Wikipedia talk:Verifiability

A general proposal

It was partially inspired by this. I noticed that some discussion about policy change do not lead to any outcome, but the opinia on the meaning of some policy clause may be very instrumental and useful. What is we create a link to some talk page discussion and add them to the policy? A possible mechanism can be as follows: if some user believes the discussion deserves to be presevted, they can initiate an RfC, and if the community verdict is "Yes" the permanent link to this discussion will be added to the policy (similar to the links to some essays). In particular, the discussion about the meaning of the word "Mainstream newspapers", which took place few years ago, was very useful.--Paul Siebert (talk) 20:42, 9 September 2021 (UTC)

This would occasionally be convenient for me, but I think it would tend to elevate past decisions and enshrine them as the One True™ Interpretation. I think the less convenient (for me) option might be better for Wikipedia in the long run. WhatamIdoing (talk) 05:46, 11 September 2021 (UTC)

I would agree with @WhatamIdoing here. Such an approach is appropriate for consensus templates imo, such as Wikipedia:WikiProject COVID-19/Consensus. Because they are meant to be much more fluid, but still less fluid than talk pages. However, as a policy, these pages are meant to be some of the most cut in stone around. I think citing past talk pages removes some of that firmness of the policy, and elevates the talk page discussion, in an inappropriate way. The policy is based on much more than a single discussion, given that it has stood the test of time and use, which are not easy to link. — Shibbolethink ^{(♔ ♕)} 19:07, 11 September 2021 (UTC)

WhatamIdoing&Shibbolethink, I meant something else. The very fact that some clauses in the policy were a subject of long debates, the breadth of the spectrum of opinia expressed, and the very fact that no consensus was achieved is an important information. It is a kind of a red flag: "Look, this part of the policy seems unclear to many people, and each of them interprets it differently!". That may help future improvement of the policy and avoiding repetition of the same arguments.--Paul Siebert (talk) 19:55, 11 September 2021 (UTC)

I think in more cases, the past discussions would tend to shut down future discussions. The ability of editors to see what they want to see in past discussions shouldn't be underestimated. WhatamIdoing (talk) 17:49, 17 September 2021 (UTC)

WP:V vs WP:NPOV

This post is inspired by this discussion. Many users argue that some source is not a RS because it is fringe and/or primary.

IMO, the main function of WP:V, WP:NOR and WP:NPOV are different, and they are as follows:

• WP:V requires that when readers read something in Wikipedia, they must be provided with an opportunity to independently verify each statement by looking at the source. That means a reader must be capable (at least theoretically) to find that source and to make sure what the source says in reality. That requires that the information must be reliably published, which means a publisher must be respectable, reputable, and the source must be stable. "Mainstream vs fringe" has no relation to that, for even a fringe views may be a RS if they were reliably published.
• WP:NPOV requires that only mainstream or significant minority views can be represented on Wikipedia pages. That means that even if a source is 100% reliable, it can and must be rejected if it is fringe. In other words, all considerations such as author's credibility, number of citations, criticism and acceptance by peers - all of that relates to NPOV, not to V.
• WP:NOR requires that primary sources must be used with cautions. That means a primary source may be 100% reliable, but it cannot be used in some context for NOR reasons. Again, that issue is irrelevant to WP:V. added to address Shibbolethink's argument

However, many participants of that discussion, which was aimed to establish if some source meets formal WP:V criteria, claim that that source is unreliable, because it is fringe. In my opinion, that happened because the core ideas of WP:V and WP:NPOV are not clearly separated in the policy text, and too much attention is given to the issues that in reality relate to WP:NPOV rather to WP:V.--Paul Siebert (talk) 17:52, 11 September 2021 (UTC)

You have just mischaracterized the position of many users. They do not think the source in question is unreliable because it is fringe. Many in that discussion consider it unreliable because the authors have no relevant expertise, are describing their WP:PRIMARY findings, are likely not peer-reviewed by experts in the area, and are cited by other RSes mainly as refutation. The fact that the paper espouses a FRINGE viewpoint and does not have any WP:DUE content adds context, but it is not the sole reason why it is not an RS.

You also have misunderstood the point of the reliable sources noticeboard. It is not to determine whether something meets WP:V. If it was, it would be called the "Verifiability noticeboard." The board exists to help determine if sources should or should not be considered reliable. Not whether they are used in a WP:V-compliant way, which is a separate issue. — Shibbolethink ^{(♔ ♕)} 17:58, 11 September 2021 (UTC)

I always understood RSN as a noticeboard where users ask question about compliance of sources with WP:V, so "verifiability" should not be understood in a colloquial meaning of that word. The shortcut is WP:V/N--Paul Siebert (talk) 18:06, 11 September 2021 (UTC)

or WP:RSN. Of course verifiability matters there. But that is not the sole purpose of the board; It appears you may be excluding the entire WP:RS guideline as if it has no bearing on the discussions themselves. See the top line of the page description: The guideline that most directly relates to whether a given source is reliable is Wikipedia:Reliable sources. It has an explanatory supplement at Wikipedia:Reliable sources/Perennial sources, listing the outcomes of many consensus discussions at this noticeboard. If WP:RS and WP:V conflict, WP:V has priority as a policy. But, in this case, they are extremely compatible. WP:V lists author and status/expertise of that author as relevant. Removing that may create unnecessary conflict between WP:RS and WP:V, reducing the ability of the project to form consensus on these issues. Why would we want that? — Shibbolethink ^{(♔ ♕)} 18:42, 11 September 2021 (UTC)

Well, WP:PRIMARY is a realm of the third core policy, WP:NOR. Yes, a good notion. Will add it to my original post.--Paul Siebert (talk) 18:00, 11 September 2021 (UTC)

I think an important aspect of this talk post is that you have not suggested any change to the policy. What would you change here to prevent the inaccurate interpretations you perceive in other users? — Shibbolethink ^{(♔ ♕)} 18:01, 11 September 2021 (UTC)

If we will come to an agreement that the policy needs in some clarification (imo, part of its content should be moved to NOR/NPOV), then I am ready to discuss it. Thus, it should be properly explained that WP:V focuses on formal criteria, such as a publisher, affiliation, peer-reviewing etc. Everything that relates to acceptance of some publication by peers, criticism, support, number of citations, quality of citations - all of that is an NPOV realm. Similarly, WP:PSTS has no relation to reliability: there are totally unreliable secondary sources and super-reliable primary or tertiary ones. Some formally peer-reviewed journals are less reliable than good blogs - but that, again, the due weight issues, formally, the former is a RS, whereas the latter is not.--Paul Siebert (talk) 18:04, 11 September 2021 (UTC)

Firm disagree that criteria based on author, perception of the field, citations, etc should be removed from this page. If we do that, we open the door to lots and lots of pseudoscience and fringe science being perceived as legitimate and in compliant with WP:V. — Shibbolethink ^{(♔ ♕)} 18:34, 11 September 2021 (UTC)

I think where PRIMARY comes into play in this situation is WP:REDFLAG. We know from NOR that PRIMARY sources are typically less reliable than secondary. And indeed the V policy states outright that independent sources are more important. So when we have an exceptional claim, such as those made in this paper, then a PRIMARY source may be more likely to fail verification as an inadequate source. — Shibbolethink ^{(♔ ♕)} 18:55, 11 September 2021 (UTC)

I'm not seeing a major conflict here. We have a peer-reviewed paper in a major MEDRS journal but one that presents a view that is very contrary to a number of other established theories. The paper meets V, but inclusion of it would appear to fail to meet NPOV, specifically UNDUE - its a minority viewpoint compared to the bulk of prevailing theories. We don't even need to bring up the question of the authority of the paper's authors (outside that they are not leading experts in the field which would factor into UNDUE).

I would point out that is the situation at this time. Maybe in time this paper will be proven right, just as Copernicus was in time. But WP definitely should avoid including theories that yet to have that larger acceptance in the broad scientific/medical community per MEDRS, even if the source is an RS/meets V. --Masem (t) 18:27, 11 September 2021 (UTC)

Gotta be honest with you, calling BioEssays a major MEDRS journal is a very flawed characterization. I have rarely, if ever, seen a BioEssays paper that truly meets MEDRS. I cannot think of a single one. They are quite often very speculative and therefore nowhere near the stability or "confirmed"-ness required by WP:MEDRS. Some may meet WP:RS, if they are from content experts and written as topic reviews, or even as WP:RSOPINION, perhaps summing up or describing some recent shift in the field... But MEDRS is a much higher bar, as it should be imo. — Shibbolethink ^{(♔ ♕)} 18:36, 11 September 2021 (UTC)

I agree its not Nature or the like, but we can even take that out of the equation here and just simply point to UNDUE being the reason to exclude the source for now per "one paper vs prevailing medical-based theory". As soon as you start begging the question of the journal quality, the authors' expertise, or the primary-nature of the work, you complicate the question that makes the OP question about these policies' overlap confusing. You can easily simplify the question that you have one source that is very much counter to everything else out there, that unless it was from the most expert source in the world, then its just UNDUE, period. --Masem (t) 18:41, 11 September 2021 (UTC)

I guess my concern is that this approach opens the door to other papers from this author, this journal, etc. I don't think it would be appropriate to use a similar paper from this publication that is anti-lab leak either. Not if it is as heavily WP:PRIMARY, WP:RSOPINION as this one is, etc. I think it's important we define which criteria we are using to reject a source on that noticeboard, and the plausible criteria are spelled out right here on WP:V! and especially WP:REDFLAG.

In the future, I would still like to be able to cite those same criteria in WP:RSN discussions. Being UNDUE may not come up in those discussions depending on the viewpoint, but it does not make that future source more reliable. I see what you mean that we should be satisfied with the UNDUE component. But this discussion was brought to the RSN, not the NPOVN, so I think many users described the reliability based on criteria here and in WP:RS, and did not touch DUE.

To me, this talk page section seems like a grievance looking for a home. rather than pointing out an actual inconsistency in this policy. — Shibbolethink ^{(♔ ♕)} 18:46, 11 September 2021 (UTC)

If I see what you are getting at: there is nothing that would immediately disqualify said paper from being a RS per WP:V (given its peer review from a non-predator journal) but under MEDRS, it can be taken as a lower quality source to be using for any medical related claim due to the low weight of the journal and that as you say, that journal is generally meant as "essays" rather than research results and thus prone to more open hypothesis rather than scientific method and results. Perhaps the closest we have is as you say, this specific journal is basically RSOPINION (in context of MEDRS), similar to any newspaper's op-ed page. I just don't think one needs to complicate the reason not to use this article without getting so far into the weeds of the complexity of V/NPOV/NOR policies. --Masem (t) 19:10, 11 September 2021 (UTC)

MEDRS is built upon several policies. Thus, when it refers to PRIMARY, it is WP:PSTS (i.e. NOR). However, do I understand it correct that the RSN discussions should be focused NOT on whether this particular source is good for this particular article, but on whether this particular source is acceptable per WP:V? In that sense, the conclusion "Yes. It is reliable" means it meets WP:V criteria, not that it can be used for this particular purpose.

However, if I am right, then most people don't understand the policy, which means it should be made more clear.--Paul Siebert (talk) 19:40, 11 September 2021 (UTC)

You are mistaken. The introductory words on RSN demonstrate that the purpose of that board is for evaluating sources in context, in fact they ask that posters include links to the specific article and the specific text to be verified, so that WP:RS may be applied to the question. It says this is because questions of reliability are context-dependent. Many sources are reliable for statement "X," but unreliable for statement "Y". — Shibbolethink ^{(♔ ♕)} 20:01, 11 September 2021 (UTC)

In other words, that RSN discussion is so long and hot because different people answer different questions. Some of them, like me, answer the question: "Does this source meet WP:V?", whereas others, like Shibbolethink, answer the question "Can this source be used in this concrete article and in that concrete context?", which is a much broader question. To demonstrate my point, had the same question about this source been posted on WP:NPON, my responce would be "this reliable source represents either fringe or insignificant minority view, so it should not be used in this article".--Paul Siebert (talk) 19:44, 11 September 2021 (UTC)

You may be right about our two questions, but your question is not the only purpose of that noticeboard. From the very first words on that board: Welcome to the reliable sources noticeboard. This page is for posting questions regarding whether particular sources are reliable in context. (emphasis mine) — Shibbolethink ^{(♔ ♕)} 19:56, 11 September 2021 (UTC)

Interesting. It seems different users understand it differently, and, taking into account that many of them express the same view as I, we need to provide some explanation (maybe, by modifying the introductory words). My (and not only my) understanding is that RSN is linted to V in the same way as NPOVN is linked to NPOV and NORN linked to NOR.--Paul Siebert (talk) 20:13, 11 September 2021 (UTC)

Wouldn't it make more sense for RSN to be linked to RS? That is how the discussion seems to be going, anyway, as there are many editors (myself included) who analyse the source as a function of the criteria listed at WP:RS and also in light of the context of where it is to be used (WP:CONTEXTMATTERS). RandomCanadian (talk / contribs) 20:34, 11 September 2021 (UTC)

Actually, it is linked to RS, but RS is just guidelines, which is not binding.--Paul Siebert (talk) 20:40, 11 September 2021 (UTC)

Most of the same criteria are listed at WP:V anyway, so this seems like a lawyerish and unimportant distinction in this context. RandomCanadian (talk / contribs) 20:43, 11 September 2021 (UTC)

If that were the case, we would never had such a long RSN descussion. WP:V is more formal, and it does not include PSTS or FRINGE. Since PSTS/FRINGE arguments were brought during that discussion, that means different people understand reliability criteria totally differently---Paul Siebert (talk) 20:49, 11 September 2021 (UTC)

Many users have gotten mired in disagreements much smaller and more nonsensical than this, so I do not find this argument very convincing. — Shibbolethink ^{(♔ ♕)} 21:46, 11 September 2021 (UTC)

But if we can easily avoid that by clarifying the rules, why cannot we do that?--Paul Siebert (talk) 22:20, 11 September 2021 (UTC)

Are you saying we should disregard the guidelines without any actual reason to do so? I believe the default in most cases is that the guidelines should be how we conduct ourselves, and only default to the policy when the policy and guidelines disagree. If you think the guidelines are not meaningful or actionable, that is very troubling indeed, and does not appear to be a view shared by most WP:RSN users, from reading the current discussions on that page. — Shibbolethink ^{(♔ ♕)} 20:57, 11 September 2021 (UTC)

I am saying that when you ask people a question, you must do that clearly. I (as well as many other users) think that each RSN question is about compliance with WP:V (and not to NOR/NPOV). And answer the question accordingly. My answer was "yes, it complies with formal WP:V criteria" (and I meant it was obvious from my answer that subsequent analysis may show that source fails WP:NPOV/NOR; however, I didn't write that explicitly, because I believe that is offtopic on this concrete page). Note, several other users responded in the same vein, which means they share my vision of functions of that page.

You may also notice that there were no claims that that source is mainstream or not primary. The objections were just about its rejections from WP:V point of view. --Paul Siebert (talk) 21:09, 11 September 2021 (UTC)

The objections (mine included) are about much more than it's lack of suitability per NPOV. There's quite a lot of discussion of the factors which determine the reliability of a source (which are identical on both V and RS); namely the source itself, it's authors, and the publisher. Hard to see how you would miss that. RandomCanadian (talk / contribs) 21:53, 11 September 2021 (UTC)

And that is a big problem, because if we apply the same approach to Wikipedia in general, lion's share of its content must be removed.--Paul Siebert (talk) 22:03, 11 September 2021 (UTC)

Why? — Shibbolethink ^{(♔ ♕)} 22:42, 11 September 2021 (UTC)

Because majority of WP articles are based on such sources as local newspapers, web sites, magazines, which, according to your (and mine) standard are total garbage. We cannot pohibit those sources, because most article will become unsourced. Meanwhile, if we get rid of those articles, Wikipedia will not be Wikipedia anymore.--Paul Siebert (talk) 23:01, 11 September 2021 (UTC)

My standards are applied to scholarly sources, I treat news sources differently because they have different properties. Speak for yourself. — Shibbolethink ^{(♔ ♕)} 03:26, 12 September 2021 (UTC)

Your "speak for yourself" is somewhat impolite. Actually, my standards are the same. The problem is that in the topics that I usually edit other users use quite different standards, and, they are in accordance with our policy too.--Paul Siebert (talk) 03:59, 12 September 2021 (UTC)

My apologies. I didn't mean to offend. By way of explanation: I get very ornery when someone claims to know how I feel about something, especially when I don't feel that way! I think you are pointing out an overall issue with wikipedia in that it does not value expertise. But I don't see your suggestions as solving that problem, and I think they may create new ones... For instance, I see that allowing users to challenge the author of a source as unreliable creates problems for you in other areas. But removing that criteria from V creates many additional problems in the areas of pseudoscience and alternative medicine. Because many crackpots get published in high quality venues, and it is not immediately obvious that the claims are UNDUE. DUE/UNDUE takes a lot more time and effort to establish than simply pointing out the flawed credentials of the author. That's why I appreciate such a criteria included here. — Shibbolethink ^{(♔ ♕)} 13:56, 12 September 2021 (UTC)

I do not propose to remove anything from WP:V. I propose to clearly split it on two parts, let's conditionally call then WP:V-a and WP:V-b. The first part is about verifiability proper (as I describe it below), and it determines if this concrete statement is verifiable (can be checked by going by the reference, which is reliably published and stable). The second part WP:V-b is about trustworthiness, and different criteria are applied here. Currently, these two components are mixed, so some users make a stress on the first part (and pretend everything is fine with the source), whereas others emphasize the second part. If these two aspects will be clearly separated, and it will be stipulated that they both must be met, pushing questionable sources will become much more problematic.--Paul Siebert (talk) 16:52, 12 September 2021 (UTC)

@Paul Siebert, I'm really interested in this question, so thanks for bringing it here. Were you around for the Wikipedia:Attribution proposal? It might have been before your time. The main proposal was to merge WP:V and WP:NOR. NOR basically has two pieces: the parts that are more or less WP:V and the WP:PSTS section. (There was some talk then, and several times since then, about splitting PSTS to its own policy page.) The reason I bring this up is: I don't think that there's a really sharp dividing line between WP:V and WP:NOR (or at least the non-PSTS parts of NOR). I do agree with you about NPOV being a separate consideration. WhatamIdoing (talk) 20:22, 11 September 2021 (UTC)

If you want my opinion about Wikipedia:Attribution, I see it somewhat differently. I would clearly separate Verifiability from Reliability, because they are totally different things.

Verifiability means just one thing: everything that is written in Wikipedia can be verified when a reader go by a reference. If something was published, printed and stored in some reliable place (or on some reliable server), and this information is (at least theoretically) available to a reader, that source is reliable. In that case, "reliable" means we can rely on it for verification purposes. If some Nobel prize winner gave a speech, but that speech has never been properly recorded, that source is not reliable. If a Flat Earth theorist published some article in NYT, that source is reliable (for verification purposes only). In connection to that, one Verifiability aspect, which is totally overlooked, is as follows: "Does this particular source have a significant risk to produce a dead link? In reality, a lot of ostensibly "sourced" information in Wikipedia are references that lead to page 404: there was some information by the moment the article had been written, but it disappeared since then. IMO, our goal is to estimate a probability of such risks. Thus, many references that lead to questionable web archives should be considered unreliable. In other words, stability is an important aspect of verifiability, which seems to be totally ignored in the policy.

The reason why I think Verifability and Reliability must be separated are quite simple: the former can be checked by simple and formal criteria, thereby eliminating a possibility of conflicts and edit wars.

Reliability is a totally different story. Actually, we should call it "trustworthiness" or "correctness". It relates to reliability of information that is published in some concrete source (of course, we limit ourselves with reliably published sources only). In contrast to verifiability, a decision about reliability of some information is much harder to make, and, frankly, that is something we must avoid (because we all must be considered amateurs with no expertise). Instead, we should use mainsreamness criteria: if the source does not contradict to what other sources say on that account, or (which even better) that source is explicitly supported by others (for example, it has positive reviews etc), than it, most likely, reflects majority or significant minority viewpoint. In other words, "reliable" is closely linked to "mainstream/majority". In contrast to the analysis of author's credential or the context (which may require some expertise), evaluation of mainstreamness is more straitforward.

That is why I think the two aspects of WP:V should be clearly separated, and the second one (reliability/trustworthiness) should be closer linked to NPOV.--Paul Siebert (talk) 22:53, 11 September 2021 (UTC)

That would allow for wikilawyering of the "but it's listed in [garbage source]". If a source isn't reliable, then no, it cannot be used to satisfy verifiability; otherwise we'd be linking to bullshit predatory journals and fansites and the like. You claim on your user page to have published peer-reviewed papers: now, I don't know what discipline it is in, but I assume that no matter where, serious scholarly publications require serious sources; and that you have enough self-awareness to acknowledge that this is more complicated than what you are depicting above. RandomCanadian (talk / contribs) 00:25, 12 September 2021 (UTC)

To take an extreme example, if we were to publish a quote from the Daily Mail - despite being a deprecated and generally unreliable source - we can still show it is possible to verify it through a proper citation to the publication date/page or URL. That meets WP:V, but as we all know, it is very unlikely this is reliable information. To me, that falls under the context of where WP:V and WP:NOR overlap - we want to include information from sources that we know are verified (so no through-the-grapevine or personal calls with experts or email chains) but that show that no original research is needed to extract the necessary information, meaning that we are looking to the reliability of those sources to make sure they are telling us what is that information. And while there are necessary ground rules (that we prefer third-party sources, and generally favor secondary sources over primary), this is also where there are a lot of field-specific aspects to consider. As soon as you step into the world of science and peer-review, then we know there's a range of quality of journal publications that come into play, something that doesn't affect more common mainstream topics like politics, sports, or entertainment. Its just that I don't think we can we write absolute reliability rules beyond the minimum set out in WP:V, and instead leave this to the individual fields. (This would parallel with how notability has a general base rule but also field-specific considerations) I can see where there is the idea of reliability can come from, but in considering that V/NOR/NPOV and NOT form the core content policies, reliability is a sourcing aspect. --Masem (t) 00:50, 12 September 2021 (UTC)

I don't see any problem here. Instead of the current single step (mixed) check procedure, we will have a two step procedure: (i) "Is that information verifiable?" "Yes, it was published in Daily Mail, and everybody can check the newspaper archives by themselves", (ii) "Is this source reliable" "No, according to other, reliable sources, information published in Daily Mail is generally non-reliable".

The advantage of that approach is obvious. Since the current approach is an implicit combination of these two components, some users make more stress on the first or second aspect (depending on what supports their POV better). However, if the two aspects are separated, and we explicitly say BOTH of them must be met, the WP:V related disputes will be more formal and structured, and there will be less freedom of maneuver for POV pushing. --Paul Siebert (talk) 04:08, 12 September 2021 (UTC)

Are you suggesting we formalize the criteria in WP:RS as a bigger part of WP:V? If so, I would agree with that approach. But I am not sure you will find consensus, because I do not see all that many instances where POV charlatans really care about the PAGs anyway. They usually haven't read any of these pages, and don't care much about the structure of policy vs guideline... — Shibbolethink ^{(♔ ♕)} 13:58, 12 September 2021 (UTC)

If we assume that any good proposal will be filibustered by POV pushers, then editing Wikipedia is senseless. But we are still here, aren't we?--Paul Siebert (talk) 16:53, 12 September 2021 (UTC)

@Paul Siebert, I think your view of WP:V aligns with my view of NOR: it's proof that someone else said it (i.e., it wasn't just made up by a Wikipedia editor), regardless of whether the source is suitable for the statement in question. WhatamIdoing (talk) 03:26, 18 September 2021 (UTC)

V v NPOV

I think the linked discussion occurred on the wrong noticeboard and conflated "is reliable" for "I can use it (in some unspecified way)". Or at least it was perceived as some that way. Perhaps the Wikipedia:Fringe theories/Noticeboard would have been more appropriate. Paul, you may be right that you were want to answer general question whereas others were answering a specific question but really, everyone discussing Covid conspiracies knows that "give and inch and you'll take a mile". Most of the above (and earlier) discussions seems to be two editors talking past each other. Covid never really brings out the best kind of policy discussions and I think this one isn't going anywhere useful. -- Colin°^Talk 13:08, 13 September 2021 (UTC)

"give an inch and you tale a mile" is an argument that is not relevant to policy or guidelines. Although I myself find the hypothesis about artificial origin of SARS-CoV2 totally fringe, I may argue that it might be useful to mention such articles (in a context of their debunking), because general public still believes in the lab leakage hypothesis, and if Wikipedia will ignore such articles (instead of their explicit refutation), that may reinforce laymen's belief that Wikipedia is dominated by leftists/Communist/CIA/FSB , or similar consporacy bullshit. However, it is possible that I am missing something. I amm not editing that topic (SARS-CoV2 is my job in RL, so I believe I have a right to relax here, and to do something that is not related to it, like history).--Paul Siebert (talk) 20:57, 13 September 2021 (UTC)

Paul you know the old joke: "In theory, there's no difference between theory and practice. In practice, there is." Most sourcing discussions are contaminated by the agenda and worldviews of the participants, and Covid is the outlier that breaks everything. Many participants are playing the game "If I can get consensus that policy/guideline says X or that policy/guideline does not apply/does apply, then I can write Y or I can remove Y". It mostly isn't about trying to improve policy or guidelines for the benefit of all articles, but about being able to win whatever argument they are involved in. IMO anything covid related should be banned from influencing guideline and policy because the particpants really are not concerned with anything other than their immediate agenda. I think your approach of avoiding a subject related to real life is a good one. -- Colin°^Talk 22:17, 13 September 2021 (UTC)

But I will say, in general, I think it's a good thing that we treat policies and guidelines with very special care re: changes, even outside of these contentious areas. Most people come to this talk page wanting to change something that was an annoyance in another discussion, and it's difficult to see the full and total implications of that change unless it is examined in hindsight... I really do believe "if it ain't broke, don't fix it" applies here more than almost anywhere on wiki. Because the changes have lots of repercussions... Hence why we need an awfully large consensus to make changes, and why it's really okay with me that the massive BMI RfC failed. It probably should have failed to pass. Because, in hindsight, we were making unnecessary changes to fix current problems. We were trying to make an Asshole John rule. — Shibbolethink ^{(♔ ♕)} 23:17, 13 September 2021 (UTC)

I don't see what current problems this thread is supposed to address. My proposal was a result of my observations that I made during several years, and the recent discussion just pulled a trigger.

I think it is important to make things as formal as possible for a very simple reason. Although it is not recommended, closure of long discussions is performed by de facto vote count. Now imagine a situation when 70% of responces were: "Reliable, because it was published in a peer-reviewed journal and well referenced". I perfectly understand that that is just a part of truth, but, according to our policy that is a correct conclusion. And it would be not a surprise if the outcome will be interpreted as "RS". However, if we explicitly stipulate that the question is actually TWO combined questions ("Is it reliable?" "Is it trustworthy?") the answer is supposed to have a format: "Yes, RS. No, not trustworthy", which significantly facilitates formal closure.--Paul Siebert (talk) 23:56, 13 September 2021 (UTC)

Paul, you keep using this word "Reliable" as though it is something that fully and easily describes a whole thing, like a car being red. Always ask yourself "reliable for what?" A small study looks at a cancer drug and found a few patients get better for a while. That sort of thing gets published in a "peer-reviewed journal" and I'm sure the paper is full of "references". But it isn't "reliable" for a statement that "drug X treats cancer" or "drug X is a promising new cancer therapy". And even the marginally better "A study found drug X helps cancer patients" is misleading the reader by missing out important details like how small the study was, how the patients were selected, which got better and for how long, etc, etc. This is why we use secondary sources. A primary research paper in a good journal is "reliable" for describing what the research did and the very specific and limited results it got. Wikipedia articles aren't just a list of research studies. An encyclopaedia might sometimes mention some groundbreaking or notable research, but mostly we are wanting to describe the consensus and knowledge these studies led to. And judging whether that research established a new consensus or confirmed or rejected existing knowledge that we might want to write about in an encyclopaedia .... that's not a job for Wikipedians. So if your "reliable" is "Is being published in a peer reviewed journal enough to make a study reliable for us making health claims about a treatment?" Very very much no. -- Colin°^Talk 07:48, 14 September 2021 (UTC)

• Comment

  I just wanted to comment on a couple of things that was stated: "Actually, it is linked to RS, but RS is just guidelines, which is not binding.", Any guideline, or even an essay, can be relevant or "binding" (certainly if there are enough editors that support it) if it is not contested as going against or undermining a policy, Many times things are not contested and enjoy consensus by silence which can change if contested. Maintenence is always far behind article creation.

  I don't see a "versus" as this section is titled. NPOV is often subjective because it includes "fairly, proportionately, and, as far as possible (my emphasis), without editorial bias". The NPOV aspect is not a particular concern of the reliability of a source but is assuming the source is considered reliable. Is a source reliable in the context wherein it is being used? Wikipedia core content policies are to work in harmony with NPOV concerns as the more important. This is probably why WP:NPOV states: Editors are strongly encouraged to familiarize themselves with all three. This will add a stumbling block to many "POV pushers".

  Many news media outlets may not be actually independent (An "explanatory supplement") but this is an important aspect of sourcing. Gannett (for example) owns "100 daily newspapers and nearly 1,000 weekly newspapers". Advertising is used to influence people and so does mass news media or the other name for "mainstream media". This is where WP:Balance is important. If a source is generally considered reliable, and it does not misinform or mislead a reader, it is acceptable as a source but content using the source (thus the source) can very well slant the article to a particular POV. I have seen many articles "slanted" by content using dead links. If I run across an article with several 404 errors, that I can't verify, I might add a WP:OR tag to the article or bring it up on the talk page. . This type of content, if not corrected, can be removed (and the source) if it presents undue weight. This is where consensus becomes important. ​I was not in the discussions but wonder if it was a good idea to depracate the controversy section? Otr500 (talk) 21:29, 12 September 2021 (UTC)

Now my proposal is to add the following sentence: (after the current statement "If available, academic and peer-reviewed publications are usually the most reliable sources in topics such as history, medicine, and science.")

Note that a scientific paper documenting a new experiment conducted by the author is a primary source on the outcome of that experiment, and primary sources should NOT normally be used as a basis for biomedical content.

The two parts of the sentence are both direct quote from existing policy or guideline pages. I only added the linking words "note that" and "and". If you find them problematic, you need to change the policy or guideline pages first.

Why is it needed here if other policy or guideline pages already contain it?

Because this page Wikipedia:Verifiability#Reliable sources is the entry point for sourcing ("WP:SOURCE" redirects to there). It's unrealistic to expect a new editor to read all the secret rules buried in the mountain of policy and guideline pages. This entry point should be a consice summary on the policy of sourcing.

Cons:

This policy page is one sentence longer.

Pros:

Makes thing a little clearer. Betty (talk) 06:01, 13 September 2021 (UTC)

I agree with this revert by user:Colin and the comment "This is covered by WP:PSTS. There are reasons beyond verifiability why WP prefers secondary sources". In the past when changes like this are added to a policy, although they may initially be in harmony with other policy sections that cover a similar point, over time there is a tendency for them to diverge and this give Wiki-lawyers wriggle room. -- PBS (talk) 15:43, 14 September 2021 (UTC)

Ok, it seems I understand the problem. We have three core content policies V, NOR, NPOV, and three noticeboards (WP:RSN, WP:NORN, WP:NPOVN). Many users, including myself, conclude (which seems quite logical) that each of the three noticeboards is linked to the corresponding policy. That is true for NORN and NPOVN, but it seems that WP:V is not liked directly to WP:V, but is linked to WP:SOURCES, WP:MEDRS and some other guidelines. That may lead (and frequently leads) to long disputes where some people argue that the source X is acceptable per WP:V, whereas others say that it is not acceptable per WP:FRINGE or WP:SOURCES, etc. This situation is not unique, and I think we need to clarify that misunderstanding in either way. What if we start an RFC about that?--Paul Siebert (talk) 20:51, 15 September 2021 (UTC)

I think what the above discussion shows is that most users here do not believe any clarification is necessary, and could lead to further problems. — Shibbolethink ^{(♔ ♕)} 23:00, 15 September 2021 (UTC)

What? WP:SOURCES is not a guideline, it is policy. WP:SOURCES is a central, perhaps the central, part of WP:V. Unreliable sources do not comply with WP:V, that's flat out stated in the first sentence of this policy. If people are discussing WP:SOURCE, it is a discussion directly about WP:V. -- Alanscottwalker (talk) 01:42, 16 September 2021 (UTC)

Alanscottwalker, it was just a typo. The header says: Welcome to the reliable sources noticeboard. This page is for posting questions regarding whether particular sources are reliable in context.

That means I made a typo: instead of WP:RS I typed WP:SOURCES. However, that does not affect my major point: unlike other two noticeboards, WP:RSN is linked not to the policy (WP:V), but to guidelines, and WP:V doesn't have its own noticeboard.

Clearly, if the header were Welcome to the reliable sources noticeboard. This page is for posting questions regarding whether particular sources are reliable in context, I would never start this discussion.--Paul Siebert (talk) 02:07, 16 September 2021 (UTC)

WP:V is about relaible sources. What do you want to discuss, if not reliable sources since that's what WP:V is about? Perhaps analogies help; V is the bricks, NOR is the mortar, and NPOV is the superstructure, or V is the paper fibers, NOR is the paper glue, and NPOV is the binding ready sheet. I'm not sure why it matters that RS is linked since RS' purpose to to elaborate V -- no one is going to prevent you from addressing V there, since RS incorporates and actuates V. And the inseperable issue for V is publishing process and context. -- Alanscottwalker (talk) 17:46, 16 September 2021 (UTC)

I don't think we should make such a big a deal over the difference between WP:SOURCES and WP:RS. Almost the first thing that the former does is link to the latter for "Further information". Ultimately all the policies and any relevant guidelines should be considered when considering article text and its sources. I don't think we need to be rigid that there is a 1:1 mapping from policy to noticeboard or that discussions started on one board may only reference the associated policy. For example, people quite often try to push a alt POV on Wikipedia by bigging up the quality of the source: "Look it was published in a peer reviewed journal" and discussion really needs steered away from tedious debates about whether that journal is reliable and more towards the WP:WEIGHT side of "Why are you pushing something the secondary literature ignores / rejects?" -- Colin°^Talk 19:38, 17 September 2021 (UTC)

I still disagree. The "Additional notes" section of WP:RSN header clearly says that the page is linked to WP:V, and it specifies that NOR/NPOV aspects should be discussed at different noticeboards. Meanwhile, as the recent discussion demonstrated, many people claim that some source is not reliable because it is fringe, which literally means the source fails WP:V criteria because it is fails WP:NPOV. That leads to long and senseless disputes. In my opinion, a solution would be to link the RSN page directly to WP:V, and clearly explain that the page's function is to find the answer to the following questions:

1. Does the source of interest meet WP:V criteria sensu stricto?

2. Does it comply with WP:RS recommendations?

Actually, a positive answer to the first question does not mean a blanket approval of the source. For example, during a recent discussion, I answered "Yes", but I made a reservation that other two policies need to be consulted. That means, the discussion should move to, e.g. WP:NPOVN, and the question that has to be asked is: "Can this reputably published source be used in a context of a lab leakage hypothesis per NPOV?", and the most probable answer would be "No". Similarly, the question if that source can be considered primary or secondary (in that concrete context) could be asked at NORN.

The advantages of this approach are obvious: each time, a single and a relatively simple question is asked, which dramatically accelerates achievement of consensus. Thus, had we quickly agreed that BioEssays is a good journal, we could easily leave this question behind, and focused on the main issue: whether this source is fringe. However, as you can see, more and more participants were joining a discussion and bringing the same argument about the quality of the journal, which was really distracting.

--Paul Siebert (talk) 23:09, 28 September 2021 (UTC)

...A lot of people in that discussion do not believe that the editorial board has the expertise necessary to evaluate those papers, and that the authors are not experts in the field. We did not "quickly agree" that BioEssays is a good journal for this type of content. Hence why this is a straw man argument. Many did not agree that the source was "reputably published." Quality of the journal for the topic at hand is a component of WP:V. And it is not as simple as "good Impact factor = good journal." A really good petroleum engineering journal is not reputable for publishing papers about Climate Change. A really good fundamentalist christian religious studies journal is not reputable for publishing studies about evolution of saurischian fossils. etc etc. "reputable" requires context. — Shibbolethink ^{(♔ ♕)} 00:11, 29 September 2021 (UTC)

If you are a scientist, you are expected to understand the core idea of Born-Oppenheimer approximation: a complex problem can be solved much easier if it is split on simpler parts. In addition, it is much easier to resolve disputes in Wikipedia if that is done in a formal way, because expertise of participants may be very different. I cannot understand why you are persistently refusing to make your life easier if there is an opportunity to do so.

Your arguments about BioEssays are not convincing, and the whole approach is non-productive. It is not my goal to defend BioEssays or to advocate Sergeto&Deigin, but, if I were a POV pusher, I could easily beat your argument against BioEssays. Your analogy with "christian religious studies journal vs evolution of saurischian fossils" is absolutely flawed. Just take a look at this list of publications about viruses in this journal. BioEssays publishes a lot of articles about viruses, and many of them are well cited. That means, from the totally formal point of view (i.e., according to WP:V), BioEssays IS a quite good and reliable source. If you disagree, just ask a general question at RSN, something like: "Is BioEssays a reliable source for virology related topics?", and the answer will be: "Of course, it is". Now explain me please: why some specific article in this journal, which is RS per our policy, can be considered unreliable per WP:V? There is nothing in that policy that allows us to make such a conclusion, and your arguments are shaky (from a point of view of an admin or another non-expert who will be closing RSN discussions). Again, I do accept your arguments as a scientist, because my expertise allows me to make a correct conclusion. But your approach is flawed, because we must assume most of us may be non-experts, and those users who look at the problem from the point of view of our policy may quickly and easily beat your argument.

Therefore, instead of pushing weak arguments, why cannot we switch to really strong ones? Why cannot we admit "Yes, it is RS per WP:V", and switch to a much stronger argumemt: " Now, let's see if we can use this reliable source per WP:NPOV/NOR". And that immediately disarms S&D's proponents, for there is virtually no evidence of a non-fringe nature of this source.

In other words, instead of endless repetition of "This source is reliable per WP:V because it was reputably published" - "No it is not reliable per WP:V because it is fringe per NPOV" - and so on ad nauseum, I propose a simple and short stepwise procedure:

• Is it reliable per WP:V? If "No", the source is rejected
• If yes, is it acceptable per NPOV? If "No", the source is rejected
• If yes, is it being used correctly per NOR? If "No", the source is rejected
• If yes, does it comply with RS and MEDRS recommendations? If "No", the source may be rejected

This discussion is split on small subtopics, which makes the whole process more organized and leaves much less freedom of maneuver for various POV pushers. Unfortunately. I am rarely editing science related topics, so I don't know how frequently you meet civil POV pushers here. However, in my topics, the probability to meet that type users is significant. That is why I am constantly keeping that possibility in mind. Our policy is good if all users are good faith users. However, it can be easily misused by civil POV pushers, and we need to make their life as hard as possible.--Paul Siebert (talk) 01:46, 29 September 2021 (UTC)

If you are a scientist, you are expected to understand the core idea of Born-Oppenheimer approximation: a complex problem can be solved much easier if it is split on simpler parts. As an aside, I have made it through undergraduate training, a PhD, publishing lots of papers and so on, and now a few years of medical school, all without ever hearing about this principle of quantum mechanics. And, likewise, I would not expect a physicist to understand the principle of somatic hypermutation or clonal selection. But I suppose I would expect them to understand the core principle of evolution and natural selection driven by mutation, so that's fair. But to pretend that all problems are better solved by breaking them into component parts is to misunderstand how Wikipedia works. More and more rules and finer and finer distinctions in policy do not necessarily make a better encyclopedia. See: Wikipedia:Asshole John rule, Wikipedia:Wikilawyering, and Wikipedia:BUREAUCRACY. It is almost never a good idea to change policy based upon one's feelings about a particular dispute at one point in time. The spirit of the rule is not changed here. Very few, if any, commenters in this discussion have agreed that this distinction is worth making. I understand your concern, I get why you're frustrated with how WP:V works. But I do not think your solution will fix any problems. I would suggest that you read WP:1AM and consider whether this is worth your effort. And consider whether Civil POV pushers actually care how WP:V works, or if they would just find another way to argue the rules. I do not think solving POV pushers is a matter of better tailoring policy. It is a matter of better enforcing the policies we already have. — Shibbolethink ^{(♔ ♕)} 12:19, 29 September 2021 (UTC)

Perhaps one way of looking at this is that Paul is taking a rigidly reductionist approach to answering the question:

"Can I write this text in this article while citing this source (or these sources)?"

He argues this should be broken down into separate policy discussions on separate notice boards, each looking at possibly simpler questions, and then presumably combined together again to conclude. But we are just here to answer that one big question, and while determining which of any policies and guidelines influenced the outcome can be enlightening, concentrating too much on that is distracting from the problem of building an encyclopaedia. Editors may come to a particularly noticeboard thinking that the answer to their question is best found by examining WP:V or WP:NPOV or WP:MEDRS or whatever, and they may be mostly right or completely wrong. I think it is more important to answer the encyclopeadia-building question and move along, than to get totally bogged down in a mire as happened at the linked discussion. -- Colin°^Talk 14:50, 29 September 2021 (UTC)

Shibbolethink, first of all, I during that discussion I came to a conclusion that there is no urgent need in policy change. IMO, it would be more correct to say that the question is in its correct application of the policy. More concretely, we may discuss a possibility to change the RSN page header.

Secondly, I am not saying all problems should be split into simpler components. My point is that, when possible, such splitting significantly simplifies a discussion, so it is just silly not to use such an opportunity.

Re "consider whether Civil POV pushers actually care ...etc" Not only they care, that is their most standard approach". In a discussion where two or more aspects are mixed together, civil POV pushers usually make a stress at those aspects that are most helpful for pushing their POV. Thus, if a question combines V and NPOV issues (e.g. "Can it be used per WP:RS"), they may vote "Yes, the source is acceptable because it is reliable per V" (if that source was reputably published but expresses fringe POV). Alternatively, if publisher's reputation is questionable, they may argue "This source represents a significant minority view". Again, when you ask a composite question, it is more likely that a partial "Yes" may be interpreted as "Yes" during the discussion's closure.

Colin, No. I am not arguing that "this should be broken down into separate policy discussions on separate notice boards, each looking at possibly simpler questions, and then presumably combined together again to conclude". By saying that, you demonstrate you poor understanding of our content policy. A correct logical operation here is AND, not OR: Wikipedia content must comply with each of three content policies, which means we don't need to combine anything. If some source is ok per WP:V, but not ok per WP:NPOV, that means it is NOT acceptable. That is why breaking each question into smaller components is always good (when that is possible). I can give you an extreme example: if some Nobel prize winner publishes some article in Nature, that publication is ALWAYS a reliable source per WP:V. However, that publication may express some totally fringe ideas, which are ignored or directly rejected by the whole scientific community. That means that source is fringe, and cannot be used in Wikipedia per WP:NPOV, independently on the result of the RSN discussion. --Paul Siebert (talk) 15:57, 29 September 2021 (UTC)

I tend to favor Paul's stepwise approach myself, but I also find that people who want to add something take the first "yes, technically it just barely complies with WP:V" and ignore the "but all content needs to comply with all the core content policies, not just WP:V, so you need to keep thinking" part.

I may occasionally annoy editors by telling them that they're citing the wrong rules (e.g., "per MEDRS" when the problem is really "per NPOV"). For most purposes, the exact reasoning isn't important. But if you are thinking about policy writing (which is hard), the stepwise, micro-question approach may prevent you from trying to put neutrality or notability rules into a sourcing policy, and vice versa. WhatamIdoing (talk) 04:49, 30 September 2021 (UTC)

WP:V is about material in articles being verifiable to reliable sources. It always needs the "material" for anyone to make a judgement call that goes much beyond "often is" or "likely to be". To pick your example, Nobel prize winner Linus Pauling published a short article in Nature PMID 13036888 (referencing a longer one in PNAS PMID 16578429) about the structure of DNA. It says The structure involves three intertwined helical polynucleotide chains. (my bold) This would not be a reliable source for our article to say "DNA has a triple helix structure." Primary research papers are not reliable sources for scientific facts. A huge amount of research turns out to be in some way incorrect. It is a failure at WP:V. So, I disagree that Nobel prize winners in Nature get an automatic pass at WP:V. As MEDRS explains, we achieve greater reliability when people conduct systematic reviews of the research, or where a good publication and knowlegable authors produce a comprehensive literature review. For long established things like the structure of DNA, pretty much any science textbook by a reputable publisher would be a reliable source. -- Colin°^Talk 14:49, 30 September 2021 (UTC)

You describe a totally artificial situation, because the source you are talking about is outdated, and everybody, including the author himself, agrees it is wrong. However, it does not make it not a RS. Thus, it would be quite legitimate to write "In 1953, Pauling proposed that DNA is a triple helix", and that would be in a full accordance with our policy. Nobody could remove this statement from Wikipedia under a pretext that the source is unreliable. That does not mean, however, that this Pauling's article can be used instead of the Watson&Crick's article published in the same journal immediately after that: the Pauling's article should be used only in a context of early, historical models of DNA structure, but it is a quite reliable source for that purpose. However, a source that is reliable per WP:V may contradict to other reliable sources, it may be outdated, fringe etc. That may significantly limit its usage in Wikipedia, but it doesn't make it unreliable per WP:V. Indeed, had that article been retracted? No. Can a reader go by the reference and verify that Pauling really wrote that? Yes. That means the source is reliable, and your example perfectly demonstrates my point: WP:V and WP:NPOV are two different policies, and they should be clearly separated.

It is fundamentally incorrect to mix WP:V, WP:NPOV, and WP:NOR. To say that "This source is not reliable because it is fringe, or because it is primary" is a total nonsense. Moreover, you are forgetting the main goal of the content policy: to make a decision which source can be used in this particular context, and which is not. That is what we need, and by mixing the three content policies you make a discussion long, confusing, and vulnerable to manipulations by various freaks. Do you really need that?

Let me re-iterate it again:

• WP:V sets a formal and rigid standards for technical quality of a source (is it reliably published, and can the publisher guarantee the technical quality of the text). If these formal criteria are not met, the source cannot be used in Wikipedia in any context. That is the first screen.
• WP:NPOV evaluate a linkage of each concrete source with mainstream views (is the information presented in this source in agreement with the current knowledge of the subject). If the source says something that dramatically contradicts to the currently accepted views, this source cannot be used in that context. That is the second screen.
• WP:NOR checks if Wikipedians can use this concrete source as a support for some concrete statement. If the nature of the source creates a risk that it can be easily misinterpreted or misused by the editors, this source cannot be used. That is the third screen.

This is a crystal clear and perfect scheme, and to pass our quality control check, the source must comply with all three policies. If the screening procedure is clear and simple, that minimizes the amount of conflicts and misunderstanding. Can you please explain me what is a reason for making it more convoluted? What is a need in mixing WP:V, WP:NPOV, and WP:NOR?

Actually, to say "Ok, this source is RS per WP:V, but we cannot use it in this context per WP:NPOV, and cannot use it for non-descriptive statements per WP:NOR" essentially means "No, this source is unacceptable per our policy". In connection to that, I cannot understand why the same outcome must be achieved using much more convoluted argumentation that you propose ("This source says something that has been refuted by other sources, and it is primary, which means it is not reliable per WP:V". It seems obvious that, although my and your approached lead to the same outcome (rejection of the source), your argumentation is much easier beatable. --Paul Siebert (talk) 16:15, 30 September 2021 (UTC)

Paul, when you wrote "In 1953, Pauling proposed that DNA is a triple helix" you had different "material". And WP:V says "Questionable sources should be used only as sources for material on themselves" so citing 1953 Paulings paper about what Pauling proposed requires actually the weakest possible source. This is entirely my point. You keep thinking we can asses sources (plus perhaps authors) without the context of article text. We just can't. -- Colin°^Talk 21:12, 30 September 2021 (UTC)

I have to agree with Paul here, particular applied to the Pauling example. The source 100% meets V/RS as a peer-reviewed journal, but its usability falters at the NPOV test as it would be one novel scientific viewpoint against accepted scientific views. An example of where such a case would fail V would be if that same claim was made in a predatory journal or one that lacked peer-review. --Masem (t) 16:59, 30 September 2021 (UTC)

Not a fan of that mechanistic process, because it elevates "steps" over whole analysis and treats V as if something that has no context, when the way to look at V is within the three core polcies context. It sounds like there are many reasons to denigrate the source under V, and at the same time denigrate it under NPOV. Alanscottwalker (talk) 18:19, 30 September 2021 (UTC)

But WP:V is written wholly to be contextless with respect to the content of the article, outside of evaluating the reliability specific to the topic itself (eg as a recent RS/N / RS/P discussion, articles from Rolling Stone are find for anything related to entertainment, but are unreliable for anything political). A key part of WP:V is "Verifiability does not guarantee inclusion" and points to all the various parts of NOR, NPOV and other relevant policies that require us to consider that once a source is determined as reliable, if that content is appropriate to include. Trying to conflate reliability with anything coming from NOR/NPOV or other content policies is what makes these discussions a mess. Mind you , there are some places where WP:V overlaps into the other policies - RSOPINION, BLPSPS, etc. but those are more exceptional places, and it is far easier to consider a source by first simply asking if it reliable under WP:V/RS, and then breaking down appropriateness for inclusion by any other policy that is appropriate. --Masem (t) 18:33, 30 September 2021 (UTC)

No. V is about context. In what manner is it being said (work: things like, is it formal work); who is saying it (author: things like author expertise), in what means are they saying it (publisher: things like, is this a publisher that looks to appropriately publish and provides safeguards to that effect), and what are they talking about (is this bit of info the proper context, for this work/author/publisher's musings). Thus as V policy says, "The appropriateness of any source depends on the context," and that is 360 degrees of context. -- Alanscottwalker (talk) 21:51, 30 September 2021 (UTC)

WP:SOURCE lists just three criteria: the work, the author, the publisher. And it is absolutely formal.

The rest is discussed in "Other issues" section, which implies those considerations are of a secondary importance.

That seems totally logical. Using the Pauling's paper as an example, if we decide it is unreliable for the topic (DNA structure), that immediatelly causes serious problems. "Unreliable" (per WP:SOURCE) means that the work, the publisher, or the author are not trustworthy. In this concrete case, neither teh author not a journal can be considered unreliable, which means the work itself should be considered unreliable. However, some facts and observations presented there are quite correct, and they were corroborated by latter data. How do you propose to resolve that issue?

Furthermore, if some source it not reliable per WP:V, it cannot be used in any context. Does it mean we cannot use the Pauling's article for a discussion of early models of DNA structure? Again, that is nonsence. --Paul Siebert (talk) 22:12, 30 September 2021 (UTC)

Well since you begin falsely, your argument makes little headway. Your attempt to take those 3 things out of context of WP:SOURCE is poor reading, "the appropriateness of any source depends on the context" is part and parcel of WP:SOURCE. -- Alanscottwalker (talk) 22:30, 30 September 2021 (UTC)

Do you understand the difference between "reliability" and "appropriateness"? I do. I can imagine four different situations: some source may be

• 100% reliable , and 100% appropriate. Example: Watson&Crick's article in Nature (1953) about DNA structure is totally reliable and absolutely approptiate to the Wikipedia article about DNA;
• 100% reliable, but totally inappropriate. Example: Pauling&Corey's article in Nature (1953) about DNA structure is absolutely inappropriate as a source for modern view of DNA structure;
• Totally unreliable, but highly appropriate. Example: Watson's facebook post explaining some details of DNA structure;
• Totally unreliable and absolutely inappropriate. Example: virtually any facebook post written by various freaks about DNA.

Each of those examples are totally realistic, which demonstrates that formal reliability and appropriatenes in some context are quite independent things, and they should not be mixed.--Paul Siebert (talk) 04:45, 1 October 2021 (UTC)

Again wrong. Cricks paper is unreliable for 99.9999% of the information in the world, and, in addition, depending on any part of it, perhaps entirely unreliable in that part, given advances in the field. Large parts of Paulings&Linus paper has been demonstrated to be unreliable, other than for the fact that they said it. Those Facebook posts are reliable for the fact that a person wrote it and can be reliable for the author's identity, but nothing else. Your attempt to cleave reliable and appropriate is unsupportable as a matter of policy or contextual reading or good sense. This policy and page does not exist to discuss reliability in abstract, only in use - in context - for the purpose of constructing the pedia. -- Alanscottwalker (talk) 10:43, 1 October 2021 (UTC)

Paul, what's Pauling's paper 100% reliable for? Presumably it is both 0% reliable and 0% appropriate for a statement about who the US presidents are, or the year that Julius Caesar was murdered. But what statement is it actually reliable for? WhatamIdoing (talk) 16:39, 1 October 2021 (UTC)

The question is incorrect. Reliability (as described in WP:V) is a contextless criterion. Entia non sunt multiplicanda praeter necessitatem. We have one contextless parameter (verifiability a.k.a. reliability) and one parameter strictly linked to a context (appropriateness). To say "this source is non-verifiable (unreliable) in that context" is an oxymoron. Moreover, to say "it is 0% reliable for a statement about US presidents" is a bullshit, simply because it says nothing about US presidents. However, had Pauling written something about US presidents in Nature, that would be 100% reliable (verifiable) information, although it could be totally inappropriate in most context. I even can imagine some context where it could be 100% verifiable and 100% appropriate, for example, in the article about Pauling's opinion about US presidents. --Paul Siebert (talk) 18:16, 1 October 2021 (UTC)

Masem, you are just, em, wrong. WP:V right from the get-go talks about "content" and "material" and "The cited source must clearly support the material as presented in the article". It is entirely contextual. -- Colin°^Talk 21:18, 30 September 2021 (UTC)

There is multiple levels of context at issue here, which is partially why it may be confusing to talk about. There is the context of the topic at a broad level, without discussing any details of the topic, and this - as I read and have understood V/RS to have been developed - is the context that should be used to evaluate when a source is reliable (though not necessarily useable) for that topic - the Rolling Stone for entertainment, not politics example. Then there is the detailed context of specific passages and statements within the body of the article where now we talk if sources are actually useable there due to NOR and NPOV. The overall topic is one broad filter of what are possible sources based on the topic, but NOR/NPOV is a much finer filter based on the specific presentation of that topic.

And then of course, you have the most narrowest context of WP:V specifically in that citations must be present were information is being used, that is, placing citations in the right context, but this is more a mechanical factor and less about determination of reliability or usability of sources. --Masem (t) 21:27, 30 September 2021 (UTC)

(edit conflict)An obvious disadvantage of you "holistic" approach is that it leads to incessant disputes "This source is not reliable because it is fringe - no, it is reliable because it was reputably published". In addition to distracting users, it may lead to unpredictable outcome, depending on which party is more persistent and numerous. In contrast, by splitting the discussion on separate subtopics, each of which with clear and unequivocal answer, we can rule out a possibility of such outcome. Indeed, those who are making stress on reputability cannot win a dispute unless a non-fringe nature of the source is proven. Similarly, those who oppose to that source because they believe it is fringe do not have to question the publisher's reputation: this issue is not relevant if the source is fringe.

However, there may be some advantages of your approach that I haven't seen. Can you please explain them?--Paul Siebert (talk) 18:38, 30 September 2021 (UTC)

Paul, it isn't helpful to say that a source is fringe. The viewpoints and opinions expressed in a source may be fringe, but the source is just a source of information. -- Colin°^Talk 21:18, 30 September 2021 (UTC)

Yes, and, going back to your example (Pauling's paper), it IS a reliable source for many statements. Thus, Pauling and Corey reported that nucleobases are separated by 3.5 Angstroms, and they are stacked. And that is exactly what we know about the DNA structure. I don't know if you are familiar with biophysics, but a standard Watson-Crick model also assumes that the bases are stacked, and the distance between them is 3.5 A. Furthermore, triple helices are also reported for DNA. That means only Pauling&Corey's conclusions are wrong, but their data are quite trustworthy. Which proves my point: the source is reliable per WP:V, but some Pauling's thougths are fringe, and should be treated accordingly.-Paul Siebert (talk) 21:30, 30 September 2021 (UTC)

Paul and Colin, you two seem to be violently agreeing with each other. I wish the agreement could be less violent.

Colin wrote "This would not be a reliable source for our article to say "DNA has a triple helix structure." Primary research papers are not reliable sources for scientific facts" (e.g., for Wikipedia to say that something is definitely so, rather than for saying that someone said or did something).

Notice that Colin doesn't say "not reliable for any purpose at all"; to find that sort of anti-policy thinking, you'll have to go to find less-experienced editors at RSN and RSP. Colin has already agreed with you that Pauling's paper is reliable for many statements, including for statements such as "In 1953, Pauling proposed that DNA is a triple helix". WhatamIdoing (talk) 00:39, 1 October 2021 (UTC)

:)--Paul Siebert (talk) 04:45, 1 October 2021 (UTC)

My problem is that sometimes Paul writes things that seem to be agreeing with me and then spoils it by writing something that seems to disagree with what they wrote. For example, wrt "a source for modern view of DNA structure" Paul claims the paper is "100% reliable" and just "inappropriate". And I don't even begin to understand the "Totally unreliable, but highly appropriate." comment about Watson's facebook. We don't consider such research papers to be reliable sources for scientific fact (something that is definiately so). All such papers are a kind of letter to the scientific community: "Look we did these experiments, we got these results, we think it means this, what do you think?" They are tentative assertions of something that others will confirm and agree or disagree on. As such that makes them an unreliable source of the kind of confident facts Wikipedia is mostly about. Of course, NPOV and WEIGHT and other policies can help here too because it can help to reinforce when a paper got it wrong. But pluck a random scientific research paper out of PubMed and it simply isn't reliable that the thrust of that paper represents current consensus thinking on the topic. And that's quite a bit of what WP:V is about: reliability. I dare say there are some articles in the Daily Mail that are not completely made up. -- Colin°^Talk 07:50, 1 October 2021 (UTC)

The line about Watson's Facebook post is presumably an updated version of what we said years ago about a blog post by Albert Einstein about his physics theories being a reliable source for statements about his physics theories. This was the point behind SOURCE: Einstein is a reliable source about physics, even if he's just posting on his blog, just like people you've never heard of are reliable sources on physics when their articles are published in a reputable journal.

I am not entirely sure that the community agrees with this principle in practice now. The proponents of RSP have been pushing the idea that the publisher/periodical matters the most, so nothing written by Einstein in The Daily Mail could possibly be reliable, even if it were a word-for-word copy of something he published in a more reputable forum. (Which, to be fair, in such a case I'd prefer citing the other version, too, but I also wouldn't claim that Einstein wasn't a reliable source just because DM published something from an actual subject-matter expert for once.) WhatamIdoing (talk) 16:49, 1 October 2021 (UTC)

Publisher absolutely matters on reliability. Einstein if published in the Daily Mail op-ed today would be outright rejected due to DM's known falsification so we can't trust those are Einstein's words, but Einstein published in the NYTimes would at least meet RSOPINION. Same would apply if we think social media and the idea of verified accounts - Just because someone claiming to be Einstein posting on Medium or as a Forbes Contributor, for example, would not be considered reliable, but if they had the Twitter blue checkmark as verified as Einstein, we'd at least again accept that as RSOPINION. --Masem (t) 18:24, 1 October 2021 (UTC)

I recall the policy already says something about that: if Einstein believed his idea deserves publication, he would have published it in some scientific journal, not only in Daily Mail. If this information was found in Daily Mail only, that means Einstein himself didn't consider it serious.--Paul Siebert (talk)

So if he posts an opinion on social media, it's fine, but if a disreputable newspaper offers to let him run an op-ed (maybe they're trying to burnish their tarnished reputation?), you don't trust anything he says? This kind of thing does happen. If Einstein were alive and used social media, the Daily Mail's marketing department should be falling all over themselves to be able to tweet something like "Read what @AlbertEinstein says about #physics and #science". WhatamIdoing (talk) 18:11, 2 October 2021 (UTC)

Do you really think Einstein could write something totally new for Daily Mail? I means something that cannot be found in his peer-reviewed publications?--Paul Siebert (talk) 19:28, 2 October 2021 (UTC)

NPOV v V

Masem, I agree with your summary. However, it seems the way RSN is organized contradicts to it. Instead of contextless discussion of reliability of sources per WP:V, the page's header emphasises WP:RS as a primary criterion. WP:V is mentioned only is the "Additional notes" section. Some people (including myself) believe RSN is linked to WP:V in teh same way as NOTN linked to WP:NOR, and NPOVN to WP:NPOV, but, in reality, WP:V is the only policy that has no dedicated noticeboard (in contrast to NPOV and NORN). Do you think that situation is normal?--Paul Siebert (talk) 18:58, 30 September 2021 (UTC)

Keep in mind that WP:V is the controlling policy while RS is the guideline that supports it. The way I see it, WP:V does not try to go into detail on how to evaluate what is a RS, but only that WP articles must be based on RS, must include citations, and simply being verified in an RS doesn't guarentee inclusion. V only gives very broad strokes on how to determine what is a RS, and read alone (without WP:RS) would be useless for source reliability discussions. It's also basically far more of a directly, hard-to-interpret-differently policy , and hence the lack of its own noticeboard. WP:RS is then the specifics of how to interprete whether a source is reliable, and as a guideline, there is a lot of wiggle room. And that's why it has its own noticeboard for that purpose. But that's still along the lines that evaluating a source for reliability via WP:RS should be done independently of trying to judge then if that's a usable source per NPOV and NOR. --Masem (t) 20:00, 30 September 2021 (UTC)

The problem with your "along the lines" is that, to reach consensus, we have to put everything into a common knowledge domain. That means every essential concept must be explicitly stated, instead of keeping them along the lines. The RSN's headers links primarily to WP:RS (guidelines), and adds, in the "Additional notes" section, that WP:V is the page's main policy. That means the primary function of the page is the analysis of sources in a context, as opposed to intrinsically contextless nature of WP:V. That is a primary reason for incessant confusion, conflicts and manipulations, because some people, like me, make a stress on WP:V, others keep in mind WP:RS, and POV-pushers capitalize on that contradiction in attempt to push their agenda.

That can be easily fixed if we change hierarchy, by explaining that in the RSN header. Instead of

• This page is for posting questions regarding whether particular sources are reliable in context.

we should write:

• This page is for posting questions regarding whether particular sources comply with the verifiability policy and if they may be considered reliable in context.

That changes a format of answers to e.g. "Yes, it is reliable, but should not be used in that context".

And, last but not least, if the opening sentence of the header links to guidelines, that implies the results of the RSN discussion are just a recommendation, which makes a discussion toothless.--Paul Siebert (talk) 21:09, 30 September 2021 (UTC)

I would expand the latter statement to "considered reliable for supporting a specific statement".

And if I thought that it would have much effect, or not get reverted within the day, I'd suggest copying the line about sources that are always/never reliable out of the FAQ at the top of this page, too. But I currently don't feel much hope about making RSN consider the statements, rather than trying to issue blanket bans on politically biased sources. You have to be able to do Close reading to do that well, and that might be a dying art in our community. WhatamIdoing (talk) 00:44, 1 October 2021 (UTC)

Just as a thought that is not fully developed yet but want to postulate here - when we talk reliability of sources, we have generally always talked about the "bulk" of the source - the newspaper itself, the journal itself for cases of periodical type works, or in the case of books or other one-off materials, the publisher of the work. That doesn't mean within a given case of, say, a newspaper that we have cases where parts of it are unreliable if the work is otherwise considered reliable, such as op-eds from a reliable newspaper. But the "reliable" quantity has always been something applied to the "many" and not to the "single" work that be from that source. Once we start talking about a single work, we just the reliability of the work it has been pulled from, and then should begin the descent down NOR/NPOV and other factors that would determine if the work is useable and should be included. I know this is not yet a complete thought because WP:RS definitely speaks to case-by-case evaluation (particularly talking primary vs secondary sources) and that complicates this through, but if we think "reliability" as a function of the whole of the work and not of the individual source, and then usability as a function of the specific source, that might simplify matters. --Masem (t) 01:41, 1 October 2021 (UTC)

That is not so, although we can certainly decide on the unreliability of a work, and/or an author, and/or a publisher, and/or on whether the source is reliable for a statement made. Reliability is regularly not an attribute of a whole work because reliability for this policy and for the pedia is not an abstract. Alanscottwalker (talk) 11:14, 1 October 2021 (UTC)

I've just realized that I wrote essentially the same in my 04:45, 1 October 2021 (UTC) post. "Reliability" means the source meets some general and contextless criteria of information quality, and that is WP:V's domain. "Appropriateness" defines if that source can be used for this particular purpose and in this particular context, and that type decision is made based on WP:NOR/NPOV, and taking into accound WP:RS/MEDRS recommendations.--Paul Siebert (talk) 04:48, 1 October 2021 (UTC)

Reliability is context-dependent. — Shibbolethink ^{(♔ ♕)} 18:14, 2 October 2021 (UTC)

Again, your "reliability" is a composite term that combines "verifiability" and "trustworthiness/appropriateness". The former is contextless, and it is obvious from a purely logical point of view: indeed, the information is verifiable if a reader can go by a reference and see that it really says that. The latter does depend on context, and I never said it doesn't. However, WP:V says virtually nothing about a procedure of determination of "trustworthiness/appropriateness". That means we can say some source is verifiable per WP:V, because verifiability criteria are explained in the WP:V page. However, we cannot say the source is reliable (a.k.a trustworthy/appropriate) per WP:V, because the page says nothing about a procedure of its determination. If you disagree, show me which section of the WP:V page provides any details on that.--Paul Siebert (talk) 20:26, 2 October 2021 (UTC)

Masem I don't find the statement "when we talk reliability of sources, we have generally always talked about the "bulk" of the source - the newspaper itself, the journal itself for cases of periodical type works, or in the case of books or other one-off materials, the publisher of the work." to reflect the sourcing discussions I've experienced. WP:V says "Use sources that directly support the material presented in an article and are appropriate to the claims made. The appropriateness of any source depends on the context". This is in the section "What counts as a reliable source". So "appropriateness" isn't some other concept covered by some other policy. I don't understand the idea of thinking of "sources" without context (the material or text in the article) or claims that WP:V is contextless. The thing being evaluated has to be the source of article text or it isn't a source. Although a source manifests as an document in a publication by an author and publisher, if it isn't propping up some article text then it is just a newspaper column or a paper or a book or a website. A source is where something came from. Without that something, it isn't a source.

At WP:MED, nobody has ever said "That paper is in The Lancet so meets WP:V and is a reliable source". An op-ed in a newspaper or an editorial in a medical journal are simply not reliable sources for facts about the world. They are reliable for what the author believes and may present the opinion of a newspaper or journal or even a body who the author represents. The "reputation for fact-checking and accuracy" is contextual. Different kinds of article in Nature, say, will be subject to different kinds of fact checking and accuracy checks. The editorial by definition is not reviewed by peers or superiors, and merely the reputation of the journal prevents the author from getting too carried away. A research paper is only checked that the research is described accurately and the conclusions are not unreasonable. It isn't checked to ensure it reflects the consensus opinion of the field, for that is exactly what happens after publication.

A newspaper such as The Guardian, may generally meet the reputation for fact checking and accuracy. But really only for straightforward reporting and investigative pieces. The opinion pieces are less reliable to varying degrees (as WP:V says, the author is a factor). And sketch pieces such as those by John Crace can best be described as "inspired by actual events". Crace's humorous politics essays fail WP:V even though we happily accept some other article about a recent murder case, say. -- Colin°^Talk 09:06, 1 October 2021 (UTC)

The policy explains in details what "reliability" means, but it mentions "appropriateness" just in passing. From that, I conclude "appropriateness" is NOT a part of WP:V. --Paul Siebert (talk) 21:36, 1 October 2021 (UTC)

Paul, that doesn't sound like what you've been saying for the last two days. See, e.g., your comment at 04:48, 1 October 2021 and the "100% appropriate" options. Is this the first time you have thought about that line? It's possible that creating a distinction between reliability and appropriateness would be a path forward. We could define reliability as a general quality, and appropriateness as whether a reliable source is useful in a given situation.

(We might then need to come up with a different word for the current/past use of that term, which I believe is more like discouraging the use of a source to support a contentious claim if the source barely mentions the subject.) WhatamIdoing (talk) 18:19, 2 October 2021 (UTC)

This is arguably the crux of this discussion - that some editors group "appropriateness" into "reliability" while others see them as two wholly separate concepts. I would strongly agree with what I've written here myself these should be handled differently, reliability being from V and RS's oversight, while appropriateness falling to NPOV/NOR and other content policy aspects. --Masem (t) 18:38, 2 October 2021 (UTC)

WhatamIdoing, I think it is quite normal that my position changes during a discussion. Isn't it a goal of any organized and mutually respectful discussion?

Masem, grouping two different criteria is always a source of manipulations, and separation of these two criteria would be a huge step forward.--Paul Siebert (talk) 20:09, 2 October 2021 (UTC)

I tried to allude to the fact that when we discuss the reliability of the "whole" of a publication, that there can be specific subset rules, like op-eds for newspapers, and I would expect similar for peer-reviewed journals that include non-peer-review letters or the like. (To this point WP:RS/P has some principles around this) The point those is that categorically, to take the Lancet as an example, that the Lancet as a whole is a reliable source for any work that has been through its peer-review process. Any subsequent questions on a specific article from that subset are questions left to the specifics of appropriateness of NOR, NPOV and other content policies. If a Nobel scientist had a peer-reviewed paper in Nature that first postulated a completely new model of quantum theory that was 180 degrees from accepted thinking, it would be a work from a reliable source, but it likely would not be appropriate to include per NPOV due to the disagreement with the accepted model. That paper itself it not unreliable, just that from an encyclopedic view, its just not appropriate to include yet. If an editor were trying to offer up a non-peer reviewed letter in Nature or Lancet as a source, we'd immediately just say that such letters categorically fail reliability as a whole due to their lack of peer review (not just the specific letter). --Masem (t) 13:13, 1 October 2021 (UTC)

That is not the case, there are a billion things for which that peer reviewed Lanct article is not reliable, and only a small set of things for which it might be depending on the context. -- Alanscottwalker (talk) 13:32, 1 October 2021 (UTC)

I have a feeling that most of what one would use to say an individual source (and for reasons, specifically a peer-reviewed paper) from the Lancet is "unreliable" is really something that is "unusable" under NPOV/NOR and other content guidelines. In practice, the same net result comes out, but the terminology can be important to make sure that source evaluation in other cases is done properly. --Masem (t) 13:43, 1 October 2021 (UTC)

Perhaps it's best to back-up, our purpose is creating encyclopedic content and that's what WP:V is designed to support. Thus, reliability is judged in relation to what it is being used (for). Our purpose is not to vet the Lancet as some phispohical excerize, our purpose is to take a content statement we write (our, yes, minimally original writing), together with the source, and determine the relibility of the source for that proposed (our written work of minmal originality) encyclopedic statement. If the source is not reliable for that proposed content, we need to reject the content, the source replaced, or both the statement and the source rejected for use in the article, or for that use, in the article. -- Alanscottwalker (talk) 14:11, 1 October 2021 (UTC)

The thing to keep in mind is that verifyability is mainly that 1) the reader is given enough information to locate the source 2) the source has a reasonable amount of editorial control/fact checking that we are able to take their published word as as statement we can justify (but not necessarily as "truth"). A peer-reviewed paper in Lancet or Nature is such a case. A news article in the NYTimes is such a case. An op-ed article in the NYTimes is not such a case, nor is a non-peer-reviewed letter in Lancet. That's the extent of what WP:V and reliability is meant to cover. Now, as I have said, we do just reliability relative to the broad topic as per things like MEDRS, but reliability at an individual statement level is the wrong approach.

To take the hypothetical example of a Nobel laureate that has published a completely groundshaking theory in Nature as a peer-reviewed paper that is diametrically opposed to standard theory, and lacking any collaboration - that paper is still reliable for what it is reporting. Remember that verifyability is not about truth, just that are are reasonably justified that we're not looking at made up material, hoaxes, or any other deliberate misinformation. It doesn't matter what topic area we're talking about, as a peer-reviewed paper in a top level journal, its reliable. But because it presents a theory counter to long-standing scientific understanding, we absolutely are not bound to include it on topics about that theory or give it any weight anywhere, per NPOV/UNDUE. That doesn't change reliability of the source at all, that's still a factor coming from being a peer-reviewed paper in Nature.

As to come back to what I know has been an issue, now we take a Nobel laureate's letter or opinion piece (non peer reviewed) in Lancet. That doesn't immediately qualify as a reliable source under normal RS, but it does qualify as a proper RSOPINION source - particularly as we don't expect the Lancet to be falsifying letters or other non-peer reviewed materials from writers, we can readily presume that that piece is the writer's own thoughts, and thus reliable for their opinion. But whether their opinion is justified to be included is the caution that RSOPINION says to defer to NPOV/UNDUE to judge, and in this case, which involves the questioning of the COVID-19 origins, there's a lot of reasonable questions about inclusions to be asked. But it's reliability for the author's opinion does not change regardless of those questions; that reliability (for being the author's opinion) is established due to being an opinion piece written in the Lancet. There's plenty of reasons why not to use that opinion piece, but it is not a reliability issue as that's a fixed point once the publication is identified. --Masem (t) 14:47, 1 October 2021 (UTC)

(e/c) No, the function of verifiability is that the reader can check what WE WRITE, and it's purpose is to create encyclopic content. There are billions of statements we write and can write and have written for which that peer reveiwed Lancet article is entirely unreliable, and for which such an opinion piece is entirely unrelaible. Alanscottwalker (talk) 16:30, 1 October 2021 (UTC)

In some respects, The Lancet makes an unfortunate example, because they are notorious for "groundshaking" articles like the Lancet MMR autism fraud. But let's take that example. The Nobel laureate has published a completely groundshaking theory. The article is peer-reviewed and in a reputable journal.

The question isn't whether the article is "reliable". The question is what that article is reliable for.

IMO it is reliable for statements like "Alice Expert published research on the size of the Sun in 2021, claiming that the diameter of the Sun is 1.05 million miles and the volume is twice the 20th-century estimates".

IMO it is not reliable for statements like "The Sun has a diameter of just over one million miles" or "The Sun is twice as large as previously thought".

So is it reliable? The answer is both yes and no. Whether it is reliable depends on what you want to say in the article. The source is reliable for some statements and not reliable for other statements. WhatamIdoing (talk) 17:04, 1 October 2021 (UTC)

To me, this tells me something about how we write, rather than reliability, in line with WP:VOICE, though it also speaks to understanding primary verses secondary sources and other details around how we deal with classifying sources. By default, any peer-reviewed research paper presenting results from research done by the authors is primary, so regardless if their conclusions agree with established theory or deviate significantly, the results should still be presented in Wikivoice as "According to (authors)...", the source still reliable as the researchers' reporting of their results. If we had a peer-reviewed secondary source capturing a review of the literature, itself not introducing any new ideas, then that source is still reliably but we don't have to worry about the attribution in Wikivoice. There's lot of little nuances in this, but in thinking in this approach, we're not changing the reliability of the source, but separating out how WP editors should be writing to properly capture the source's information based on PSTS and other such details. --Masem (t) 17:21, 1 October 2021 (UTC)

Or another way to put this: Our writing should follow from what the source is (primary/secondary, etc.), rather than what's being argued as trying to justify sources that are used to be appropriate (whether that's reliable or usable or other measure) for what we've written. Yes, a lot of times we are ending up chasing down "citation needed", but most cases where debates on sources are being used are based on this feedback loop between what's being written by editors and then finding sources to back that up, rather than the reverse that we should be doing of finding what sources there are that are reliable and then summarizing that in our writing. That's part of why this is an issue, as this all reflects on this poor approach of developing articles. --Masem (t) 17:36, 1 October 2021 (UTC)

Surely a good writer should, but the process you decribe is internal to the USER -- it's not what we discuss on the pedia, we discuss almost entirely only after it has been written and a citation already purportedly represented in the pedia. Alanscottwalker (talk) 18:12, 1 October 2021 (UTC)

I recognize that the approach (write first, source later) is extremely common, and in non-controversial areas of common/non-specialist knowledge, this is not really harmful. But when this is applied in controversial topics and particularly in speciality areas, where some WPians claim expertise (regardless if they have it or not) it creates a pre-deposition of what sources are going to be used to back that up. And this is where I see it important to distinguish when we are excluding sources specifically due to the unreliability of the work that they are published in (which is a non-starter for inclusion in the first place), and their lack of usability due to polices like NOR/NPOV/etc. --Masem (t) 18:21, 1 October 2021 (UTC)

@Alanscottwalker:, yes, the function of verifiability is that the reader can check what we write, and that is what WP:V (a.k.a. "Verifiability") should focus on. When a reference is provided to some reliable source, a reader can go and check by themselves, and that is why "verifiability" criterion is intrinsically contextless. We actually have two different "reliabilities": "reliability aka verifiability", and "reliability aka correctness/mainstreamness". The first reliability is not dependent on a context, the second one is strictly linked to the content of some concrete article (relevance) and to the whole body of other sources on that subject (which is a NPOV's domain).

In other words, WP:V analyzes the source (as a whole) taken separately from other sources, and it is doing that in a purely formal way. In contrast, NPOV analyzes some concrete source in a context of all other sources. That is why WP:V is intrinsically contextless, and everything that is related to a context is addressed by NPOV.

@WhatamIdoing:, Lancet publication you are talking about was debunked and retracted, so this your argument is moot.

With regard the rest, the question is not correctly formulated. Not "what this source is reliable for?", but "how this information should be presented?". In other words, the question does not belong to the WP:V domain, because reliability of that source in indisputable. The question belongs to the NPOV domain: if the Nobel prize winner's statement is supported by majority of other sources, it can be used without reservations. If it contradicts to what majority sources write (but is supported by some other authors), it should be used with attribution as a significant minority POV. If the source contradicts to everything what other sources write, that reliable should probably not be used at all.--Paul Siebert (talk) 19:10, 1 October 2021 (UTC)

Almost all of what you have argued is incorrect as a matter of policy, and it is not getting better with you repeating it. -- Alanscottwalker (talk) 19:29, 1 October 2021 (UTC)

That argument is unbeatable. The only possible responce is: you are wrong. :)--Paul Siebert (talk) 20:53, 1 October 2021 (UTC)

Your argument has already been beaten. There is just no use in going over it again. Alanscottwalker (talk) 21:19, 1 October 2021 (UTC)

I would say the opposite: my arguments are essentially in agreement with what other users (e.g. Masem) say, whereas you even cannot clearly articulate what you are disagreeing with.--Paul Siebert (talk) 21:32, 1 October 2021 (UTC)

I and others have several time clearly articulated your errors and you just keep repeating your claims. -- Alanscottwalker (talk) 23:23, 1 October 2021 (UTC)

I don't remember you pointed at my error. It is more a disagreement over interpretation of the term "reliability". I can agree that I am wrong in two cases: (i) you find some wrong fact that I use for my speculations, or you point at some fact that I overlooked, or (ii) you find some error in my logic. So far, I saw no arguments of that type.

I think we can easier achieve consensus if you ask the following questions:

• Why the policy's name is "Verifiability", not "Reliability"?

• The policy explains in details formal criteria of reliability, and each of them is contextless. It also makes a reservation that relevance of each particular source depends on a context, but it tells absolutely nothing about a procedure. Literally it says: if editors achieve consensus that the source is relevant, it can be used. Does it mean the policy authorises users to decide which source is reliable? (Keep in mind that guidelines are not some strict rule, they are more a recommendation).--Paul Siebert (talk) 19:53, 2 October 2021 (UTC)

That reminds me of the discussion JBchrch and I were having at Wikipedia talk:Reliable sources this summer. It seems that we never produced a definition of "reliable source". We have information about how to figure out whether your source is reliable, but not a statement that says what it is. WhatamIdoing (talk) 02:30, 3 October 2021 (UTC)

(outdent)

Paul says:

With regard the rest, the question is not correctly formulated. Not "what this source is reliable for?", but "how this information should be presented?". In other words, the question does not belong to the WP:V domain, because reliability of that source in indisputable.

The FAQ at the top of this page says:

Are there sources that are "always reliable" or sources that are "always unreliable"?

No. The reliability of a source is entirely dependent on the context of the situation, and the statement it is being used to support. Some sources are generally better than others, but reliability is always contextual.

One of these things is not like the other, right?

If "The reliability of a source is entirely dependent on...the statement it is being used to support", then there are no indisputably reliable sources, and "what is this source reliable for?" is a critical question. WhatamIdoing (talk) 17:09, 2 October 2021 (UTC)

An issue that this discussion is raising with with respect to "the statement it is being used to support", backing the "write first, source later" approach, can lead to lots of conflict on topics of controversial nature where editors have established how the topic should be discussed based on their own expertise and then expect the sources to be found to back it. That approach causes misclassification of sources that would be reliable to be dismissed as unreliable (when there are other valid reasons not to possibly use that source) particularly in the more controversial areas. We want editors to follow the sources and that means we should be reviewing sources first and then writing to meet that, and for that purposes, reliability of a source should be decided before we write the article. --Masem (t) 17:20, 2 October 2021 (UTC)

If that's the issue being raised, it certainly isn't answered by the anti-V-policy argument that you can decide reliability out-of-context. Per policy, deciding a source is reliable is a function of putting it in context, the context of it's production, it's use, and it's own content in relation to reality, which requires more than just the incompetent taking what a source says and declaring it reliable in total. There are a multitude examples of "generally reliable sources" (which is how the issue is, perhaps too often, raised at RSN) that print bad information that no one should rely on (probably every generally reliable source has done so (being generally produced by humans), except maybe the Holy Book(s) for the true believers). Conversely, there are reliable source uses for on Wikipedia per policy, for sources declared "generally unreliable". -- Alanscottwalker (talk) 20:21, 2 October 2021 (UTC)

(edit conflict)It is quite easy to prove that your statement ("Per policy, deciding a source is reliable is a function of putting it in context") is wrong. To do that, it is sufficient to give just one example of a source that is unreliable independent on any context. And the policy gives a lot of examples: blogs, patents, personal websites are considered unreliable. They are allowed in some context, but that does not make them reliable. The policy clearly says Questionable sources should be used only as sources for material on themselves. In other words, the policy doesn't say some questionable sources may be reliable in some context, it says that unreliable sources may be used in some concrete situations (which doesn't make them reliable).--Paul Siebert (talk) 20:49, 2 October 2021 (UTC)

You have not proved anything except you don't understand Verfiability or reliability or the section you cite, when a such a source used as a cite in Wikipedia it can only be, and must be, a reliable source for that Wikipedia content. Alanscottwalker (talk) 21:26, 2 October 2021 (UTC)

I agree with Masem. Not only the situation described by him can happen, it is happening.

The FAQ you are referring to are not on the top of this page, they are on the top of guidelines page. Actually, it is a big question if this statement is in agreement with the policy.

Indeed, it literally claims that ANY source may be reliable in some context. That directly contradicts to WP:SOURCES. Indeed, if any source is reliable in some context, doesn't it mean Daily Mail or someone's personal blog may be considered reliable in some situation?--Paul Siebert (talk) 19:25, 2 October 2021 (UTC)

Yes, the Daily Mail is still reliable for some narrow things (every source that is not blacklisted is, even the ones that are generally unreliable). Since you seem say above you don't know why this policy is called Verifiability (which brings into doubt all your arguments), it is in the first sentence, and that sentence makes certain that it is a matter that only exists in context. If there were such a thing as your contextless, platonic reliable source, Wikipedia would need no other source, and anything we write would be supported by that magic source. Alanscottwalker (talk) 20:41, 2 October 2021 (UTC)

No, see my above post. The policy says questionable sources are acceptable in some specific circumstances, but it does not say they are not questionable in that context.--Paul Siebert (talk) 20:51, 2 October 2021 (UTC)

Of course, I may be wrong. Please, show me where the policy refers to questionable sources as reliable ones. To the best of my knowledge, the whole logic of WP:V is that sources are subdivided on reliable, questionable, SPS, and WP mirrors. It never mix these four categories, but it explains some concrete circumstances where usage of questionable/SPS sources may be appropriate.--Paul Siebert (talk) 21:02, 2 October 2021 (UTC)

You are wrong, if a source is used in Wikipedia to support any content it must be reliable for what Wikipedia has written, questionable or not. Alanscottwalker (talk) 21:19, 2 October 2021 (UTC)

No, you are wrong. The policy outlines strict categories of sources, and it never says the sources that fit the criteria described in "What counts as a reliable source" section may be unreliable. It says that a degree of reliability may be different, but that is a different story. And, more importantly, it never says the sources that fall into a category of "Questionable sources" may be considered reliable. No. It says they may be used (in some very specific cases), but it doesn't say they are considered reliable in that context.

I found just one exception: the policy says "Self-published expert sources may be considered reliable when produced by an established subject-matter expert, whose work in the relevant field has previously been published by reliable, independent publications". That is the only exception, and, interestingly, it is also purely contextless.--Paul Siebert (talk) 21:30, 2 October 2021 (UTC)

You are still wrong. Your argument like your other arguments seem to have no competent idea of context. Only sources reliable for the Wikipedia content may be cited, if a source is used for that content it has to be reliable for that content, that's the first sentence of V, and the rest follows from that. You have not even read that section in context, it is called, "Sources that are usually not reliable", "usually" has never meant always. --Alanscottwalker (talk) 21:43, 2 October 2021 (UTC)

I understand what you mean. It seems that issue deserves a separate discussion.--Paul Siebert (talk) 00:30, 3 October 2021 (UTC)

It is probably important to point out that the only place "context" comes up in WP:V is in the statement The appropriateness of any source depends on the context. under WP:SOURCES. In WP:RS it comes up multiple times but all related to "appropriateness" save for the part in WP:CONTEXTMATTERS but in the full statement, its clear it is about "appropriateness" that we are looking at with that context: The reliability of a source depends on context. Each source must be carefully weighed to judge whether it is reliable for the statement being made in the Wikipedia article and is an appropriate source for that content..

That said, I still will agree that there is a part about context affecting reliability but this is at a very broad level - the topic level as I've been stating, following the principles of MEDRS/SCIRS and BLP (and WP:RS#Reliability in specific contexts gets to this idea). But we should not let specific writing affect context, because that means WP editors can established the context for reliability by how they write the article, and if they are not following the appropriate "research first, write second" approach we ask, it creates the feedback loop. That's why its important to talk about "appropriateness" at the context of a specific statement. --Masem (t) 01:06, 3 October 2021 (UTC)

@Paul Siebert, the FAQ is at the top of this page. It's the same FAQ for both pages (so that we don't end up with divergent answers). I can post a screenshot if it'll help you find it, but try the equivalent of ⌘F on your computer, and search for the words "Frequently Asked Questions". WhatamIdoing (talk) 02:03, 3 October 2021 (UTC)

You are right (I just didn't see it because FAQ are collapsed here). Then this is equally relevant to this page too. The answer contradicts to the policy.--Paul Siebert (talk) 02:23, 3 October 2021 (UTC)

WP:V vs NPOV

• This discussion is only covering half of the issue. The key to both Verifiability and Reliability is that the source must directly support what we write in our article. A source may reliably verify what we write when phrased one way, and NOT reliably verify a similar statement that is phrased a different way.

The classic example of this is when we include in-text attribution, writing: “According to Ima Expert, The Moon is made of green cheese” vs “The Moon is made of green cheese”.

The attributed statement is reliably verified by citing the primary source…Ima Expert actually saying that the moon is made of green cheese. In terms of VERIFIABILITY, it does not really matter whether Ima Expert is qualified to opine on the Moon. What we are verifying is that Ima said it, and the primary source does this (reliably).

The non-attributed statement requires Ima to be a reliable secondary source… in other words, it may or may not be reliably verified depending on whether Ima is qualified to opine on the Moon.

That said… verifiability does not guarantee inclusion. To merit being included, even the attributed statement has to pass our NPOV and NOR policies. These also are tied to what we write in our article. Phrased one way, it may pass these other policies, phrased a different way it may not. Blueboar (talk) 16:01, 1 October 2021 (UTC)

Regular criteria for making a quality article ahould also apply, not just exclusion politcies. For example, it may be an irrelevant or worthless piece of informaion what Ima Expert said about the moon. (the info / sentence is about what Ima said, not about the moon) A more succinct and logically clean way to state the actual reality is "Verifiability is a requirement for inclusion, not a reason for inclusion" North8000 (talk) 16:36, 1 October 2021 (UTC)

That seems obvious, altough the current version of the ONUS section does not explain it properly (it literally means if some information is verifiable, and users achieved consensus about their inclusion, it can be added; that is wrong, because other policies must be observed too). Actually, to be included, some piece of information must comply with ALL our content policies. If that does not seem clear from the current policy text, we need to make it more explicit. Something like "Verifiability is the first necessary condition for inclusion, but it is not the only condition. The content must comply with other policies (show lnks)".--Paul Siebert (talk) 17:52, 1 October 2021 (UTC)

I agree with Blueboar. Also, it's not enough to be a secondary source; it should be a secondary source that is reasonably representative of the mainstream view. A single cherry-picked secondary source that reviews the primary research doesn't trump three secondary sources that come to the opposite conclusion.

North, your line might be succinct and logical, but I'm doubtful about its effectiveness. It's useful only if people think about it, and I doubt that they will. (I wonder whether there are enough links to Wikipedia:Balancing aspects in this policy.) WhatamIdoing (talk) 17:08, 1 October 2021 (UTC)

@WhatamIdoing:That statement just puts current structural reality in a more obvious form. It's "effectiveness" is only to put an end to persistent false memes which have no structural basis in policies and guidelines. E.G. "undo removal of sourced material" being considered to be a reason to force in material.North8000 (talk) 19:34, 1 October 2021 (UTC)

Some seem to be not discussing that aspect because they seemingly don't get it, but yes that is more than half-the-issue. Alanscottwalker (talk) 17:44, 1 October 2021 (UTC)

I am not sure there is any disagreement among us, but let me say that in somewhat different way (I apologize in advance if that is just a repetition of previous posts, but I have a feeling we need to bring that into a system).

• A source may be reliable or unreliable, and, at the same time, it may be (i) primary, (ii) secondary, (iii) tertiary, (iv) biased, (v) outdated, (vi) it may contain disputable or debunked information, (vii) marginally relevant to the subject (viii) etc... All of that has absolutely no relation to the reliability/non-reliability issues. That means to say "This source is unreliable, because it is fringe" is against our policy. Some Nobel prize winner in Physics and member of the National Academy of Sciences may publish a paper in Proceedings ... where they may express some weird ideas about human genetics. According to our rules, it will be a reliable source per WP:V, and when that question is asked at RSN, the answer should be "Yes, it is a RS". However, to that, it is necessary to add: "A decision about its appropriateness in some concrete context should be done as described in WP:NPOV."

• Regarding attribution, that is a WP:NPOV domain. If some piece of information is likely to be a majority viewpoint, no attribution is necessary. However, that has nothing in common with reliability of a source.
• The same source may be highly appropriate in one context and totally inappropriate in another. Taking into account that "Verifiability" is a property of the source as whole, it is obvious that "verifiability" (which is contextless) and "approptiateness" (which is strictly linled to context) must be clearly separated in out policy.--Paul Siebert (talk) 17:52, 1 October 2021 (UTC)

No. What you have said in main is wrong, and you have repeated these statements despite them having been refuted. Your policy assertions have been wrong, and are wrong, and no, your repeating them is not helpful. (For the sake of this page, I'll leave it at that, instead of refuting the same claims again) -- Alanscottwalker (talk) 19:46, 1 October 2021 (UTC)

@Paul Siebert, your first bullet point begins by saying that sources can be reliable or unreliable. Can you give me an example of a WP:Published source that is unreliable? WhatamIdoing (talk) 17:53, 2 October 2021 (UTC)

Most editors seem to agree that the article from which this dispute originally stemmed fits those criteria. — Shibbolethink ^{(♔ ♕)} 18:03, 2 October 2021 (UTC)

See my responce to Colin. The main difference between Wikipedia and scientific publications is that the former is dealing with (potentially) all published sources, whereas the scientists limit themselves with what WP:V called "university books and peer-reviewed publications". That means ALL sources that are being used in scientific publications are RS per our policy, and to scientists "reliability" means "trustworthiness". They can, and they are expected to judge about each RS based on their own expertise, and we are explicitly prohibited to do so per policy (which is correct, because we should be considered "self-appointed experts with uncertain credentials").

In contrast to scientific publications, Wikipedian may try to use such sources as magazines, local newspapers, youtube videos, etc. Clearly, most of them, despite being published, are not too reliable sources, and their reliability is questionable independently on context.--Paul Siebert (talk) 19:24, 2 October 2021 (UTC)

Paul, I assume that you're not trying to say that magazines, newspapers, etc. are unreliable in all subject areas. Presumably, if you want to write about the latest Hollywood film, then an entertainment-focused magazine is just the ticket.

But let's talk science. Imagine an article in a fairly reputable news magazine that says (as many did this past year and a half) that the World Health Organization has announced that people should wear face masks when they are around people they don't live with, because it reduces transmission risks. You think that the Wikipedia articles should cover this change, and you want to write something like "In June 2020, the WHO recommended that the general public should wear face coverings in public". Do you think such a news article would be a reliable source for that statement? (I do.)

What if the statement you wanted to write said "Wearing face coverings reduces the spread of COVID-19"? Would you accept that source for this statement? (I probably wouldn't.) WhatamIdoing (talk) 02:22, 3 October 2021 (UTC)

Wikipedia, by its nature, is a very unhomogeneous resource, and it could not exist if it were based on peer-reviewed publications only. That is why we need WP:SOURCES: in contrast to scientists, we are potentially allowed to use a much broader range of sources (actually, everything that was published can potentially be used in Wikipedia).

With regard to masks... Per WP:SOURCES, three components that define reliability are the work, the author, the publisher. The first statement is a piece of news about a public announcement made by a government organization. Normally, such information is issued in a form of news publication, which means the first criterion is met (the source's format is news, and it tells some piece of news); furthermore, the author of this source is probably a professional journalist working for that agency. That is exactly the same type of authors that usually write that type sources, which means the second condition is met; lastly, the publisher is a news organization with a good reputation of fact checking and accuracy. That means everything is ok with that source, and it can be used for that statement.

The second statement is not about some official statement, but about some research paper. Normally, that type publications come from more specialized publishers, and they are expected to be authored by expert. Whereas the professional reporter is quite capable of correctly transmitting the WHO statement, their expertise may be insufficient for correctly summarising the research paper.

In other words, I agree with you.--Paul Siebert (talk) 21:39, 3 October 2021 (UTC)

Paul, SOURCES doesn't "define" reliability. That section says "The word "source" when citing sources on Wikipedia has three related meanings" and "All three can affect reliability", but there is no actual definition in that section. We have no definition of reliability (anywhere). WhatamIdoing (talk) 02:15, 6 October 2021 (UTC)

The actual measure of reliability of a source is expertise and objectivity with respect to the item which cited it. Regarding sources, the core of wp:ver basically defines RS's by various attributes unrelated to context. The real discussions at the RS noticeboard actually tend to understand the reality, even though wp:ver does not recognize it. — Preceding unsigned comment added by North8000 (talk • contribs) 19:26, 1 October 2021 (UTC)

This cannot be true, because if it were true, then an WP:ABOUTSELF source could never be reliable. We do not believe that people are objective about themselves, but we do believe that what people write about themselves can be a reliable source (for what they say about themselves). WhatamIdoing (talk) 01:59, 3 October 2021 (UTC)

I'm going to duck out of this for the sake of my own sanity. Some of you must be reading a different WP:V and have a different concept of what reliable means (something you can rely on) and what a source means (something cited to support article text). It seems that nothing of what I've said has been taken on board and statements I believe to be quite clearly wrong keep getting repeated as axioms. Each post gets longer and longer, containing three then four then five then six impossible things before breakfast.

The folk at NICE review therapies to see if they are safe, effective and affordable before making recommendations that are used in practice. Their reviews collect all studies mentioning key words and they set quality criteria to filter which studies are worth even looking at. Then those selected are combined if possible or at least looked at together to weigh up the evidence. They look at the size of the studies and the type of study performed. They do some really difficult maths. They talk to experts and to patient groups. Treatments are then rejected or supported with varying degrees of confidence. What I have never seen, is the reviewers start doing that and then writing "But then we noticed this paper that was published in The Lancet by Bright Chap, et al. And we looked at each other and concluded we had been wasting our time. Here we had already a reliable source of information about our therapy. There was no point in examining the literature further for Wikipedia had declared that peer reviewed papers in this journal are reliable. If Chap 2021 says the therapy is effective and safe, who indeed are we to question? We made our recommendations and took the rest of the day off." -- Colin°^Talk 19:22, 1 October 2021 (UTC)

Well, at least, I agree this discussion is now pretty worthless, most things Paul Siebert has been saying are wrong and then he repeats them, so yes good luck, nothing in this discussion is going to change a policy, and there is no use to Paul repeating himself, over and over again. Alanscottwalker (talk) 19:46, 1 October 2021 (UTC)

Colin, now I understand what you mean.

Your problem is that you are trying to apply standards that are common for scientific community to Wikipedia. That is fundamentally wrong, and I can easily show that using your example.

The author(s) of the NICE review is a professional scientist, and they are writing for peers. And the main difference between them and us (as Wikipedians) is that whereas they ARE a source of new information, we are explicitly prohibited from posting our own thoughts. The main difference between the NICE review and a Wikipedia article is that the former is expected to contain some new information (at least, author's own analysis, synthesis or evaluation), but the latter must reproduce only what reliable sources say. Therefore, your argument is silly: thus, a journalist who works for some a news agency cannot wait for today's newspapers to obtain recent news, because he himself is a source of that information.

In addition, since the authors of NICE review are scientists who write for scientists, some things that are far from obvious to most Wikipedians are totally obvious to them. To them, all sources are just a subject of analysis and evaluation. There is no such a concept as reliable source in science: all information published in peer-reviewed journals can be (and sometimes is being) questioned, refuted or debunked by peers. In contrast, whereas I, as a RL scientist, can see errors or flaws in some research article, I (as a Wikipedian) cannot do anything with that unless I have another reliable source that demonstrates those flaws.

In addition, the author(s) of NICE review, as well as scientists in general, are dealing only with verifiable information, and predominantly with peer-reviewed publications listed by Scopus or Thompson-Reuter. That means the very concept of verifiability/reliability, as described in WP:SOURCES, is totally redundant to them. They do not divide sources on reliable/non-reliable, because they, by definition, are dealing with reliable sources only. In contrast to us, they never deal with youtube videos, newspaper reports, blogs etc.

You must understand firmly that "Verifiability", as well as a "reliable source" as a whole, is the concept intended for our internal usage in Wikipedia. It has nothing in common with how scientists treat information.--Paul Siebert (talk) 20:43, 1 October 2021 (UTC)

I got pinged so.. Paul, no, you don't understand what I mean. My little story about NICE was just a wee joke. As with all analogies, or fables, it isn't intended to be dissected. I think we've both reached the point where we think the other's POV is "silly" as you put it. My frustration is I see you and Masem saying things about WP:V that WP:V (and WP:RS) explicitly contradicts, to my reading of it, and for which I don't experience in source discussions about medical topics. So the argument seems to be at a level of "Ok, forget what WP:V actually says for a moment, that's beside the point, those are just words on a page, what it actually means is...." I'm not really interested in an argument on the internet about what WP:V means that isn't interested in what WP:V says. You guys have clearly developed strongly held opinions about what it means and I don't have anything new to say against that except quoting bits of what it actually says. Which, you can do for yourself, if you are interested.

Ok. I really am ducking out now. Please be polite and not ping me or talk about what I actually mean after I've left the conversation. Just continue discussing between yourselves, if you want. -- Colin°^Talk 09:48, 2 October 2021 (UTC)

With regard to yours "The real discussions at the RS noticeboard actually tend to understand the reality", you are just jumping one step forward. WP:V is needed to screen out an obvious garbage, and it is doing that by formal criteria. That is an important preliminary step, but that is just a preliminary one. After the sources that pass a "verifiability" criterion are selected, we must analyze them in a context of each other: even if we are RL experts in some area, we cannot say: "this source is unreliable because it contains incorrect information". Instead, we should say: "This source contradicts to what 10 other reliable sources say, so it represents a minority (or fringe view), and should be used with reservation/attribution, on not used at all". I agree that is an important (I would say, the most important) part of our job, but it is a neutrality, not verifiability issue.--Paul Siebert (talk) 21:03, 1 October 2021 (UTC)

Does "authority" need to be a separate concept from "reliability"?

An idea from some of the discussions above gave the idea to me that while "reliability" as we have it now incorporates the concept of source authority - that is, the "expertise" of the source relative to the topic, what MEDRS tries to capture related to biomedical information. The NYTimes is clearly an authority related to anyting in terms of news (particular US and international news), whereas per a recent RS/P discussion, Rolling Stone is an authority of the entertainment industry but not an authority on politics.

By separating the idea of reliability (that we know the work has an editorial process and oversight that we know that the publisher have reasonably vetted the material published and will take steps to redact if they mispublish information) and authority (that a source has expertise in certain areas and may lack authority in others), this takes the nature of the issue of "reliability in context" out of the equation, making it "authority in context", which still effectively aligns with practice but is far easier to explain why reliability is something that is strictly a function of the work and not really the context of what we using it for on WP. This also better feeds into areas like UNUDE and RSOPINION, where we want authority or expertise related to opinions and analysis as weighting factors for inclusion, leaving the question of reliability out of that.

I'm only postulating this, since this would require a bit of reworking of WP:V to express this. --Masem (t) 15:13, 3 October 2021 (UTC)

WP:SOURCES say that three aspects must be taken into account to make a decision about reliability:

- the work itself (thus, if the work is about composition of the Moon, it is directly relevant to the article about the Moon);

- the author (if the author is an expert in planetology or extraterrestrial geology, that adds credibility to the source);

- the publisher (if the work was published by OUP, or Nature, it is much more likely to be a source of reliable information).

That means all three components defines a quality of each concrete source. A source authored by a renown expert, but published in Daily Mail can hardly be considered more trustworthy than the article published in PNAS, but authored by absolutely unknown person. Similarly, a book about history of medieval astronomy is hardly a trustworthy source for the article about Moon structure.

As I explained below, the policy's language is inconsistent. It uses the words "verifiable", "reliable", "questionable", "appropriate", "acceptable" interchangeably, which confuses a reader. Thus, I still don't understand if some source can be totally unreliable per our policy. I also don't understand if the sources that meet criteria described in WP:SOURCES can be unreliable, or they can be just irrelevant in some concrete context. And so on, and so forth.

Let me reiterate: it seems the policy already says everything what you write, but its language is so confusing that everybody reads it differently.--Paul Siebert (talk) 21:21, 3 October 2021 (UTC)

Right, my point still should be this is that the body of works from a publication (non-op eds from the New York Times, peer-reviewed works from Nature, published books from Random House, UN/WHO scientific reports, etc.) should be considered "reliable" (as I am defining above, separate from "authorative"), a property that is only a function of those works and not how they would be used in WP at all. To contrast, The Daily Mail or RT are works that we have deemed completely "unreliable", though in considering "authority", they can be used when the topic involves them as actors in the topic itself and we are only using their articles that speak to their own actions (self-statements). But the general point here, going to what I think is one underlying problem in the confusion on these pages, is the idea that the body of works from a publication has "variable" reliability; that should not be the case ever - its either reliable or not, and we can reassess reliability if the publisher has a major editorial shift (eg like Newsweek), but that's still a black/white assessment applied to the body of the publication's work or a well-defined subset, and does not at all depend how it will be used on WP. --Masem (t) 13:38, 4 October 2021 (UTC)

Masem, it seems here I was able to formulate all of that and put it in the broader context. Please, comment.--Paul Siebert (talk) 15:22, 4 October 2021 (UTC)

I would say, the first step would be to stop using the word "reliable" as a some separate, specific criterion. It seems most users implicitly assume "reliable" means "passing all policy/guidelines criteria in this concrete context". That means, that terms explain nothing except we can use a "reliable source" for some concrete purpose in some specific context.

Therefore, a possible solution is to stop introducing criteria in addition to reliability, and think which criteria, taken together, define reliability.

As I already wrote, there criteria are as follows:

• Relevance of the work (and its professional structure), author's expertise, publisher's reputation (these criteria come from WP:V);

• A degree of mainstreamness of the source: if it is a majority or significant minority view, it is acceptable, otherwise it is not (this criterion comes from WP:NPOV and WP:REDFLAG)

• A type of the source (primary/secondary/tertiary): if the source is secondary, it is acceptable in all contexts, primary/tertiary sources are acceptable with reservations (this criterion comes from NOR).

If all these considerations have been taken into account, and the source passes all these criteria in some concrete context, the source is deemed reliable.

The problem is the policy does not explain that properly: it just implies that. I propose to make the policy's language more clear.--Paul Siebert (talk) 15:44, 4 October 2021 (UTC)

Even if worded this way with all three factors contributing to what is being called "reliability" I think that still confuses the matter. I still strongly believe that by how we've treated reliability in general (ignoring what these policy pages say for the time being) that that this is an intrinsic property of the publication and context in terms of where it will be used has no relevance that will change that reliability. Authority/expertise and other factors are functions of that context, of course. This is going back to the debates that I know launched this thread, as well as concurrent RSN messages (such as the one right now about whether Yahoo News is reliable given one report related to Julian Assange). Applying "reliability" piecemeal to individual articles/works from a publisher is what weighs down all of these discussions. Reframing this approach that a publisher is or isn't reliable before we even talk WP content would help a lot because then we get to more meatier questions on authority and appropriateness which are better frameworks to discuss in regards to WP context. At the end of the day, the practice still would be evaluation of sources based on multiple factors, which will include this idea of reliability I talk about, and similar to how you present these factors, just that we shouldn't be grouping them all under reliability. But I agree this is still an issue with how disjointed the writing between WP:V and WP:RS and other content policies may be. --Masem (t) 17:49, 4 October 2021 (UTC)

I don't think introducing the concept of "authority" into this is helpful. The idea of an authority being an expert is only one secondary meaning of the word, which is more often used to mean someone in power, with official weight to decide on things. Your definition of reliability seems rather strong in requiring a publisher to redact incorrect material. I would image that vanishingly few do and please consider dead trees as a source, which don't magically auto-correct.

Nor is the concept of "expertise" helpful. MEDRS does not try to capture the 'the "expertise" of the source relative to the topic' whatever that is supposed to mean. People are experts, not sources. And MEDRS regards primary research papers as unreliable for medical claims even if those are written by experts, even if written by world authorities on the subject. And "topic" is insufficiently vague. To use WAID's example above, most people would consider that a WHO recommendation announcement about wearing facemasks, and a health claim that facemasks prevent covid spreading are the same "Facemasks and Covid" topic.

Ultimately these ideas of someone being an authority or an expert on some topic are just ways of judging whether what they are saying is reliable. If we use this as a source for our text, can we rely on it to be accurate and fair?

Masem, before we start inventing new concepts around which a rewrite of WP:V would be required, I don't think you've convinced us that WP:V is contextless or reliability is contextless. I've repeadedly seen this claim by you and Paul made as though it is an axiom and a truth. I'm pleased to see now an admission ("ignoring what these policy pages say for the time being") that this is just a personal opinion that actually contradicts our current policy and guidline text. I get that you think that in practice things are different, but a few more "I personally think that..." or "In my experience I've see..." should prefix much of these claims.

Our models and language are only worthwhile if they prove useful. I don't think an approach of judging publication reliablity in a contextless manner takes us anywhere particularly useful. All we get is an approximate bias for or against a publication, and at worst a harmful prejudice.

I think a better axiom would be that both WP:V and WP:RS are concerned with sources, which are published materials being used to support statements we make (or want to make) on Wikipedia. Why do you think those pages use the word "sources" hundreds of times, rather than "published material" or "publications"?

The first line of WP:V says "verifiability means other people using the encyclopedia can check that the information comes from a reliable source". Our readers are concerned with the text they have just read, not whether a Guardian subscription is worthwhile. -- Colin°^Talk 18:28, 4 October 2021 (UTC)

Colin, I already explained that to you, but you preferred to completely ignore that. MEDLINE is not a good example, because all sources that you can find in MEDLINE are written by experts and for experts. There is no need to emphasize authority, because all sources has already passed a preliminary filter (peer-review), and each MEDLINE reader (a profesional) can judge about authority of each source by themselves. Wikipedia works according to totally different principles: it can potentially use everything that was published, and it is being written by non-experts, who, in contrast to scientists, have no built-in concept of authority, so such a criterion as authority IS important here. And, keep in mind that scientists, being the experts, are the source of authority by themselves. They don't need any external authority to decide which source deserves their attention. Thus, in the Segreto&Deygin story, many authors decided these two scientists are not notable enough to cite their work (and that was the reason to decide this source is not appropriate). In contrast, in Wikipedia we cannot do that (NPOV prohibits that). Therefore, what Masem says is right, and you are wrong.

With regard to the rest, you contradict to yourself. You say that the idea that WP:V is contextless has not been convincingly proven, but you cite (in the last paragraph) the most contextless statement in WP:V. Indeed, your quote literally means that the main requirement of WP:V is to provide a reference to some source that a reader can take and check that that source and Wikipedia say the same. That's it. This concrete quote require us neither evaluate a quality of a source, or check if it is appropriate. It requires just to provide a reference - and a reader will check the rest by themselves. I am not saying WP:V says only that, but this quote (taken separately) really confirms Masem's idea on verifiability as a contextless parameter. --Paul Siebert (talk) 19:04, 4 October 2021 (UTC)

It wasn't that I "completely ignore[d] that". I read it and it just was incorrect in just so many ways, as is the above about medline. But explaining why would be a distraction. A problem with this page currently is that there are so many misguided claims, that explaining each and every point is wrong would be exponential (Brandolini's law is what I'm getting at, though I wouldn't be as rude as to use the word "bullshit" but the point is it is easier to say something wrong than it is to explain why it is wrong). I thought I was going mad before because two editors kept saying things about WP:V and WP:RS that were obvious "failed verification" if one thinks that WP:V and WP:RS are reliable sources about WP:V and WP:RS. I'm not going to get into an argument about "need to emphasize authority" because this "authority" thing is a Masem proposal and as I explained I don't think the terminology is helpful.

The line from WP:V is not "the most contextless statement in WP:V". The "information" (that the reader wants to check) is the context. Your next sentence is really important because it is a fundamental misunderstanding that keeps reappearing in this discussion. No, the main requirement of WP:V is not just to verify that the source and the article correspond. I've seen you suggest this before but if that were true, we would not care one hoot what kind of source it was. The line says "reliable source" yet you say it doesn't require us to "evaluate a quality of a source". Even the section that focuses on citations says "reliable source" six times. The word "reliable" appears 68 times in WP:V. It is kinda like someone wanted to keep repeating that word so that we get the hint that it is important. Like how "source" is repeated over 200 times, and "publication" just seven. Paul, you can't ignore the fact that "the information comes from a reliable source" contains the word "reliable" and "source" and "source => [article] information" is context. -- Colin°^Talk 20:08, 4 October 2021 (UTC)

Don't be afraid to call something "bullshit" in a discussion with me. That may save a lot of time and space. It would be much better if you explain what is wrong in my words (even if that explanation is rude), than ignore them. We are here not to please each other, but to solve a problem.

"Authority" is not Masem's invention. That is a second component of the "reliability" concept (see WP:SOURCE). "The author" (which means their personality, in other words, if the author is an expert in the field) means essentially the same what Masem said.

Regarding "The "information" (that the reader wants to check) is the context", I am not sure if we a talking about the same things. This policy's statement means that if Wikipedia says "The Moon is made of blue cheese(ref)", a reader may go to a library, open a book (if the reference is to a book) and read essentially the same information. That's it. This book may be a kid's textbook, it may be a fiction, it may be any book (or journal, magazine, newspaper etc). This sentence does not care. The only thing this sentence says: if you go by the reference, open the source and read, you are expected to read that the Moon is made of blue cheese (not cast iron, not rocks and minerals etc). Of course, that is only a first step. The policy explains that not every source is acceptable, but the idea of this concrete sentence is there must be a source for any Wikipedia statement. That is an example of what I (and Masem) call "contextless".

If we agree that there are some contextless requirements in WP:V, we can move further.--Paul Siebert (talk) 20:46, 4 October 2021 (UTC)

Paul, what you said about "This policy's statement means" is entirely invented and has no basis in what the policy says. You might think it and believe it strongly, but repeating it here doesn't make it true. WP:V does not require "there must be a source for any Wikipedia statement" "and that's it", because it sticks the word "reliable" in front it. And it sticks that word "reliable" in front of it so often I am puzzled why you keep missing it out. What I understand Masem means by contextless is his desire to assess publications without considering that they are sources for some article text or proposed statement. He wants WP:V to rule that BigName magazine is reliable full stop, or DodgyFacts blog is unreliable full stop. He doesn't want the article text context. I'm now not sure what you mean by contextless. Paul, it would really really help if you were to base your statements on policy as written, not policy as you think it might be better understood. If you want to argue for a different way of assessing sources, then you need to be very explicit that you are no longer talking about WP:V. -- Colin°^Talk 12:56, 6 October 2021 (UTC)

Colin, please, go to WP:V. There is a box at the top of that page that says: This page in a nutshell:. Immediately after the colon, you may see the words: "Readers must be able to check that any of the information within Wikipedia articles is not just made up." That is the core idea of the page, and that is exactly what I say, for the only way for a reader to make sure that the information was not made up is to go by a ref, open a source, and compare what the source and Wikipedia says. I cannot see another interpretation.

By saying that I agree that if the reference redirects to some kids magazine of entertainment TV show, that probably discredits Wikipedia, so we must use 'reliable sources. But you must agree that these two aspects are separate.

In other words, there are several "commandments" in Wikipedia, and they are:

• Thou shalt provide a source.
• Thou shalt not use unreliable sources
• Thou shall not take the source out of context or use in an inappropriate way
• Remember to stay neutral
• Thou shalt not commit original research

and so on. And you agree they are separate commandments...--Paul Siebert (talk) 19:07, 6 October 2021 (UTC)

• Paul, I get the feeling that you are mostly concerned with articles relating to medical topics. Our sourcing standards for those are very much stricter than they are for other topic areas (to the point where we have a separate reliability guideline for med related topics). WP:MEDRS is necessary for medical information, but it does not work in other topic areas. WP:V and our other content policies and guidelines are intentionally written more flexibly, because they have to be applicable to many different topic areas. Blueboar (talk) 19:23, 4 October 2021 (UTC)

No, I am NOT concerned with articles relating to medical topics. I am keeping in mind the topics with much less strict source requirements. The only reason why biomedical topic appear in this discussion is because it was triggered by a COVID-19 lab leakage discussion, but that was not my primary concern. I am discussing universal rules, which are vague and prone to manipulations.--Paul Siebert (talk) 20:15, 4 October 2021 (UTC)

Colin and I keep inflicting the medical context on this discussion.

I think there might be a place for mentioning "authority" in our sourcing guidelines. Specifically, what we would normally consider an unreliable (or at least questionable) source can be an absolutely authoritative one for certain purposes. A book of poems is a lousy source for facts, but it's an authoritative (and primary, and reliable) source for what the words in the poem were. A tweet from a celebrity is generally considered an unreliable source for most content (because it's self-published by a non-expert), but it's an authoritative (and primary, and reliable) source for what the celebrity said.

What these authoritative sources don't provide is some reason to believe that the article should include this information. I think you would only want to use them if you had a second source that demonstrated why anyone should care. WhatamIdoing (talk) 02:47, 6 October 2021 (UTC)

While "authoratative" is an adjective one could carefully use and we get what you mean WAID, I think for PAG purposes the alternative meanings give us problems. Specifically the idea of having the sanction or weight of authority. It could lead us into cases where people are arguing over whether something is official or not (i.e. like whether a biography is authorised or unauthorised). The definition "Someone or something that is authoritative gives an impression of power and importance and is likely to be obeyed." leads us into thinking that an authoratative sources really is the "last word" on something. And of course, in your example, if the questionable source is the subject of the article text then it really is "the last word" on what that subject literally contained. And it also leads us into thinking an authoratitive sources does have a say on whether we should mention something. If they are an "authority" then who are we, mere Wikipedians, to question what they say and the importance of what they say. -- Colin°^Talk 12:56, 6 October 2021 (UTC)

References

Researching before we write

Not sure if this is the right venue, but the discussion above re WP:V and NPOV (and NOR) has me thinking… if I had to choose one criticism of WP that has never been properly addressed in my many years of editing, it would be that too many editors do not properly RESEARCH before they write. All too often, we decide that an article could be improved by adding some bit of material, and THEN go out and find a (reliable) source to support it.

I have always thought that this is backwards. We SHOULD read the corpus of sources first, and THEN summarize them (citing the best sources to support our summarization). Is there any way we can discourage our “write first, source second” culture, and encourage a “research first, summarize second” mentality? Do we have any essays on this? Blueboar (talk) 16:54, 1 October 2021 (UTC)

I believe it was called "source-based research", and I believe that it used to be part of NOR. Jc3s5h, do you happen to remember what I'm thinking of? WhatamIdoing (talk) 17:19, 1 October 2021 (UTC)

In the area where I am active (history of Eastern Europe), many users have some strong opinion, and some of they seem to selectively pick sources that support it. Interestingly, in responce to my attempts to bring more neutrality, for example, by presenting results of neutral google.scholar search to show the actual balance of points of views, they frequently argue: "We are not supposed to do original research, we just have to describe what sources say". I have a feeling those users resist to analysis of sources, because that may undermine the viewpoint they are advocating.

In connection to that, I myself am contemplating to write an essay about possible ways to evaluate the current knowledge of some concrete field. My idea is as follows: "Imagine you are a naive Wikipedian with no preliminary knowledge of some topic. What should be the best way to obtain an unbiased impression about the current (majority) views of that subject by scholarly community?" I have some ideas on how to do that (actually, that is the approach I myself am using when I edit Wikipedia). If you want, we can try to write that essay together.--Paul Siebert (talk) 18:04, 1 October 2021 (UTC)

I don't remember where it was mentioned in guidelines or discussions, but "source-based research" was used in contrast to "original research". It was mentioned in discussions where some commenters were implying that all we could do was paraphrase passages from sources, and writing cohesive passages based on the consensus of the available sources was somehow not allowed. What isn't allowed is synthesis that lead to a position that is not contained in the sources; synthesis that produces smooth-reading material that reflects the positions of the underlying sources is a good thing. Jc3s5h (talk) 18:23, 1 October 2021 (UTC)

I know for topics around current events, WP:RECENTISM/WP:10YT should be guidance to not try to document ongoing events to excutiating detail - particularly with respect to analysis and opinions - until well after the event has settled and then retrospectives written far after the event can be used that are going to be better secondary sources to start from on how events were perceived and opined about. A very large chunk of issues at the noticeboards and ANI and ARBCOM are related to editors that think they know how a situation can be written about as the situation is happening particularly with respect to opinions of that situation, and that created a self-referencing bias that we shouldn't be introducing that early into most of these articles.

This doesn't apply in the same way to most other topics, but there is often cases of WP editors claiming expertise on a topic (which they may be), writing to what their knowledge is, then coming back to add sources. On non-controversial topics this is usually not a problem, but any topic that is a bit controversial, that's a potential bias waiting to be blown out of whack by a sourcing feedback loop. --Masem (t) 00:19, 2 October 2021 (UTC)

While I think our long-term contributors like the folks at WP:MILHIST read the literature and then write, as you suggest. Most of who we see editing Wikipedia are cranks, dilettantes, fans, and POV pushers. None of those folks are here to write an encyclopedia; they simply want to use the platform for self-aggrandizement. We, of course, will do nothing about this because if we chased all of them off we wouldn't have hardly any editors left. Do like I did and just quit editing articles; it's a pointless exercise under current management. Chris Troutman (talk) 00:49, 2 October 2021 (UTC)

Questionable sources

These two sentences look illogical:

"Questionable sources should be used only as sources for material on themselves, such as in articles about themselves; see below. They are not suitable sources for contentious claims about others. "

They sound like

"You shalt not kill. Killing of postmen is not recommended either."

If questionable sources are allowed only as sources for material on themselves, they are definitely are not allowed as a source on something else, including "contentious claims about others". I think the second sentence makes the first one weaker. It should be removed.--Paul Siebert (talk) 20:56, 2 October 2021 (UTC)

• People sometime need a point driven home in multiple ways. Yes, it is repetition, but I think it is repetition that helps. Blueboar (talk) 22:13, 2 October 2021 (UTC)

that repetition is misleading. If you say "You shalt not kill. Killing of postmen is especially prohibited" implies that prohibition of murder of other people is less absolute. Thus, one may conclude that questionable sources are acceptable for non-contentious claims, which is not the case.--Paul Siebert (talk) 00:28, 3 October 2021 (UTC)

Disagree with your interpretation. I think it aids in understanding the scope of their use. We could reword it to perhaps: "They are not suitable sources for claims about others, especially contentious ones" But I do not find this particularly necessary. If it ain't broke, don't fix it. I see no examples of this being an issue. — Shibbolethink ^{(♔ ♕)} 00:55, 3 October 2021 (UTC)

The only logical rewording that does not undermine the previous sentence could be "They are not acceptable sources for other claims".

The most important part in that text is that questionable sources are acceptable only for information about themselves. Any attempt to emphasise some specific cases where they are unacceptable undermines the absoluteness of the main claim. --Paul Siebert (talk) 01:23, 3 October 2021 (UTC)

• Let me explain why both sentences are needed: let’s say I am writing an article on Questionable.com (a notable ultra right wing website)… If I were to write: “Questionable has repeatedly called President Biden a poopyhead” (and cite it to Questionable’s website) that is technically a statement about Questionable (allowed by the first sentence)… HOWEVER it is also a contentious claim about Biden (not allowed under the second sentence). In other words, the second sentence places a limit on what is allowable under the first sentence. Blueboar (talk) 13:54, 3 October 2021 (UTC)

Perhaps your example would be a bit better if Questionable made a clear factual claim, such as Biden being born in Kenya. If Questionable said something "subjective" then arguably it becomes a question of whether Questionable's insults are notable, which comes under a different policy? (I am seeing "poopyhead" as a vague insult, not a specific claim that Biden for example suffers from poopyhead syndrome.)--Andrew Lancaster (talk) 14:03, 3 October 2021 (UTC)

Blueboar, I am not satisfied with your explanation. I could agree with it if the second sentence said something like "However, even when they are used in that way, they are not acceptable as a support for information containing contentious claims about others." THAT wording would be in a full agreement with your example.

I would say the second sentence makes the first one less categorical. Using your example, the second sentence implies that, whereas we cannot use Questionable.com for a claim that Biden is a poopyhead, we are not strictly prohibited for using that source for the claim that Biden met Putin (which is not controversial). --Paul Siebert (talk) 15:56, 3 October 2021 (UTC)

Except the statement “Biden met Putin” would be barred by the first sentence (as it is not about Questionable.com). NOW, if we were to write: “Questionable claimed that Biden met Putin”, THAT might be allowed (assuming the claim itself is not controversial) as it reliably verifies the fact that Questionable made that claim. Blueboar (talk) 17:22, 3 October 2021 (UTC)

Yes, but the first sentence is barring any claim about others, not only contentious ones. What additional information the second sentence conveys?

--Paul Siebert (talk) 18:23, 3 October 2021 (UTC)

We're trying to leave some room for editors to use their judgment. I suspect, but haven't checked, that a trip through the history would find that the earlier versions contained some temporizing word (such as "normally") in the first sentence.

Sometimes it's hard to write a reasonably complete encyclopedia article without using such sources. Imagine writing Social media use by Donald Trump without ever citing a self-published or otherwise questionable source that referred to a third party. Also, we do make a distinction between "He said that" and "That is true". A source can be reliable, and even authoritative for the first, while being completely unreliable and inappropriate for the second. WhatamIdoing (talk) 03:00, 6 October 2021 (UTC)

The policy's language is ambiguous

Thanks to @Alanscottwalker:, I've re-read the policy, and I found that the policy language is somewhat ambiguous.

The policy separates sources on "Reliable sources" and the "Sources that are usually not reliable". Note, the policy doesn't say there are sources that are usually reliable, and there are sources that usually don't. According to the policy, some source can be either reliable (period) or "usually not reliable". And, in both cases, separation is done based on formal criteria (contextless).

Next, the policy says that some "usually not reliable sources" may be used for some concrete purposes (e.g. for information about themselves). Note, the policy doesn't say those sources are "reliable for information about themselves", it says those "non reliable" sources can be used for this concrete purpose. The only exception is made for self-published expert sources by an established subject-matter expert. According to the policy, they may be considered reliable. With regard to all other "usually non reliable sources", the policy says they are "acceptable" in some concrete cases, but it doesn't say they are "reliable" in some concrete context.

This inconsistency leads to ambiguity. Thus, what does this statement mean: "Self-published and questionable sources may be used as sources of information about themselves". How do we interpret it? That SPS are reliable as a sources of information about themselves, or that they are non reliable, but can be used as sources of information about themselves? The current language allows both interpretations.

This situation can be resolved in two ways.

• First. We can clearly write that sources are either reliable or non-reliable (currently, we have a different classification: "reliable" vs "usually not reliable"). After that, we can explain that not reliable sources may be used in some specific cases (e.g. for information about themselves).

• Second. We can write that some sources that are "usually reliable" may be unreliable is some context (and explain the rule that define that context). Similarly, we can write that some "usually not reliable sources" may be reliable for some concrete purposes (for example, for information about themselves), and explain how to determine when they are reliable.

Currently, the policy uses a mixed language, which, in my opinion, leads to a significant confusion.

Finally, I think, since the section "Sources that are usually not reliable" applies the term "reliable" only to the SPS authored by experts, it would be logical to move this part to the "Reliable sources" section.--Paul Siebert (talk) 02:49, 3 October 2021 (UTC)

I think the community, or at least a significant part of the part who has thought about this, thinks that pretty much all sources are reliable for something, and also unreliable for something. I think one potential source of confusion for newcomers to these discussions though, is that "reliable source" has become a sort of jargon shorthand for the kind of source which is strong enough in a specific field that there is no real common sense doubt about its general status for most statements in that field. We used to always say that we want to keep it a bit fuzzy and remind editors to look at the context of each case. --Andrew Lancaster (talk) 14:09, 3 October 2021 (UTC)

• Reliability ITSELF is a somewhat ambiguous concept. There is no such thing as a 100% “reliable source”, nor such a thing as a 100% “unreliable source”. As Andrew has said, context matters.

For example, while the news reporting in the New York Times is generally considered quite reliable, there have been specific NYT articles that contain factual inaccuracies. Those specific NYT articles can be (and are) challenged on a case-by-case basis, and can be deemed unreliable for that specific inaccurate fact… yet these rare instances of unreliability do not change the overall reliability of the outlet.

Meanwhile, the Daily Mail may report accurately on a specific fact (say a sports score), but because it has a history of containing SO MANY inaccuracies and fabrications, it’s overall unreliability has reached the point where we have depreciated it, and only accept it as a reliable source on very rare case-by-case determinations. Blueboar (talk) 15:19, 3 October 2021 (UTC)

It seems you are inventing a wheel. When we speak about a "context", what context do we mean? Obviously, that can be only the context of the whole body of sources written on that subject. If the source X says something that is in agreement with the rest of human knowledge (i.e. majority of other mainstream sources), and it passes verifiability criteria, this source is 100% reliable. However, the relationship of this particular source with other sources is not a WP:V domain, it is a WP:NPOV domain.--Paul Siebert (talk) 16:13, 3 October 2021 (UTC)

• Ok… When I talk about “context”, I am referring to the specific WP article and the specific sentence in that WP article that we are attempting to verifying by citing the source. The same source may be reliable for verifying one sentence, but not reliable at all for another sentence ( (the attributed: “According to Joe Pundit, X is Y”, for example, is reliably verified by citing Joe Pundit saying this in an OpEd… while the unattributed “X is Y” is probably NOT reliably verified by citing Joe’s OpEd). Blueboar (talk) 19:24, 5 October 2021 (UTC)

I think there is no need to continue the discussion in this section, because the next section addresses this issue. If we define the term "source" as I defined it in the next section (something like "a piece of text that describes some topic, is authored by some author(s), and published by some publisher"), that naturally leads to a triad: "relevance and quality of the source - author's expertise and credentials - publisher's reputation". Their combination is a measure of reliability, and some of those parameters are contextless (example: work's quality or publisher's reputation) others must be analyzed in context (examples: relevance of the work, author's expertise). The problem is that we need to make policy's language more clear to explicitly explain that.--Paul Siebert (talk) 20:23, 5 October 2021 (UTC)

• The definition of the term “source” is simple: “Something you can read, view or listen to that supports what you write.” Blueboar (talk) 13:50, 11 October 2021 (UTC)

  That may be a colloquial definition, but that definition cannot be compatible with WP:V. The core idea of the policy is: "Do not use the information that can not be reliably obtained by others to verify your claims." If you have read something that others wouldn't read, you have seen something that others wouldn't see, etc that is not a source per policy. For example, you've seen Ball lightning. Is that a source? No, and I event don't need to explain why.

  A "source" of information (in a colloquial meaning of that word) and a "source" per WP:V are totally different things. Otherwise I see no need in this policy at all. Paul Siebert (talk) 15:13, 11 October 2021 (UTC)

  I would organise in in two levels. First, the information in Wikipedia must be verifiable, which means everybody can make sure it has some external source, and it was not just made up. That means a piece of information can be called a source only if it can be used for verification of some Wikipedia claim.

  Second, a source is reliable (in some concrete context) if it meets some formal criteria described in the second part of the policy.

  In other words, the set of objects called "a piece of information" can be subdivided on "sources" and "others", and the set of sources is subdivided on "reliable sources (in some concrete context)" and "not reliable sources". Paul Siebert (talk) 15:38, 11 October 2021 (UTC)

  I understand that the way to organize knowledge, and the exact words to describe them, has been a primary subject of philosophers for centuries, and we are unlikely to resolve that question to everyone's satisfaction on this page. That said, "sources" are 'where all the pieces of information come from', which includes:

  • personal experience (e.g., sunlight is bright),
  • personal communication (e.g., Mom said not to stare at the Sun),
  • documents (e.g., the book said it's dangerous to stare at the Sun),
  • tools (e.g., the light meter on my phone says it's 45,000 lux on the sidewalk today),
  • etc.

  I'm not sure what would fall into your category of "others". All information comes from somewhere. All information came from "a source". WhatamIdoing (talk) 19:36, 11 October 2021 (UTC)

  Are we going to discuss the "source of information" concept in general or in a context of WP:V? In my opinion, the former is more appropriate to some Wikipedia article, and it is NOT relevant to the policy. What IS relevant is the following: what types of information sources are acceptable for making Wikipedia context verifiable?

  Yes, all information come from some source, but

  • if that information is obtained from your personal experience, it is NOT verifiable: other people may have a totally different experience, so they cannot verify your claim;
  • if that information was obtained from on someone's personal communication, it is verifiable ONLY if that communication was made public (i.e., it was published), which means that piece of information must have at least three characteristics ("who-what-where") as I already described below;
  • if that information was obtained from some document, it can be verifiable if that document was published (similar to what I wrote above);
  • if the information was obtained by you using some tools, that is your original research, and you are explicitly prohibited to publish it per another policy;

In summary, all examples provided by you are either covered by my definition or are explicitly prohibited by this or other policies. Since this page is not a forum for a general discussion, we should not talk about some vague and general definition of the work "source" if that does not lead to improvement of the policy. Clearly, the policy text is supposed to define not a "source" in general, but what is considered a source from the point of view of WP:V. Paul Siebert (talk) 22:07, 12 October 2021 (UTC)

Meaning?

English is not my mother tongue, so my question may be stupid. However, I would like to make sure if this paragraph is logically and grammaticaly correct:

The word "source" when citing sources on Wikipedia has three related meanings:

• The work itself (the article, book)
• The creator of the work (the writer, journalist)
• The publisher of the work (for example, Random House or Cambridge University Press)

All three can affect reliability.

In this text, the subject is "meaning", and it says "meaning affects reliability". Does it means that reliability is affected by just a meaning of some term? To me, it is nonsense. I suggest to think about better wording. I have some ideas how to improve it, but I would like to have a feedback from you first. Do you agree that the text looks awkward?--Paul Siebert (talk) 15:16, 5 October 2021 (UTC)

It reads fine to me as is, but I suppose we could change 'meanings' to 'definitions'. - MrOllie (talk) 16:09, 5 October 2021 (UTC)

I'm having difficulty reading the text the way you appear to be reading it. Another way to word it is The word source when citing sources on Wikipedia can refer to the work itself, its creator, or its publisher, all of which can affect reliability. Schazjmd (talk) 16:10, 5 October 2021 (UTC)

If we omit auxiliary words, we get "a work ... affects reliability (of itself)".--Paul Siebert (talk) 16:30, 5 October 2021 (UTC)

That's fine. Reputable publishers and authors sometimes produce works that are less reliable by their nature - works of satire, for example. - MrOllie (talk) 16:40, 5 October 2021 (UTC)

That is NOT fine. The text that we are discussing is not some essay. That is a policy, which is supposed to explain how to determine reliability of sources. If it says: "Some sources may be reliable, others may be not reliable, and it is up to you to decide", that is not a policy.--Paul Siebert (talk) 16:44, 5 October 2021 (UTC)

It is only a problem if viewed in isolation. The policy goes on to explain that peer-reviewed articles are superior, which is an example of this. The same author and publisher might produce both a letter to the editor and a peer-reviewed article. That one is more reliable than the other is a feature, not a bug. Also, policies are not algorithms to be applied mechanically. There is always going to be room for consensus decision making, so it will always be up to the involved editors to decide. - MrOllie (talk) 17:37, 5 October 2021 (UTC)

Yes, the policy is intuitively clear. And that is a problem: rules may be intuitively clear when they are written for people with similar background. That is not the case for Wikipedia. Some users are experts in real life, whereas others may be unfamiliar even with such a concept as peer-review. For example, I know some users who claim that peer-reviewed publications are "yellow journalism".

In that situation, we need clear and formal rules. Thus, when we provide a definition, it should be a real definition, preferably an intensional definition. For example: "In a context of the verifiability policy, the term "source" is defined as some work that has a concrete author(s) and that was reputably published by some publisher. A quality of the work and its relevance to the Wikipedia content, author's background and expertise, and publisher's reputation are the three factors that affect reliability of the source". That wording is far from perfect, but, in my opinion, it is much better than the current one.--Paul Siebert (talk) 18:10, 5 October 2021 (UTC)

• As the editor who originally added this text to the policy (although it has been slightly tweaked in the years since I originally added it)… the intent was NOT to define the term “source”. We were merely trying to alert editors to the fact that the term already has multiple definitions (definitions that pre-existed Wikipedia). The goal was merely to say that all three pre-existing definitions should be considered when assessing a source for reliability. Blueboar (talk) 18:51, 5 October 2021 (UTC)

  I cannot agree with that. From my experience, discussions between Wikipedians about sources (at RSN and elsewhere) take into account the work AND the author AND the publisher as three components of source, not three different meanings.

  In addition, the structure of V is supposed to be as follows:

  • You shalt provide a source
  • "Source" means (....). There are good sources and there are bad sources
  • Good sources are (...)
  • Bad sources are (...)

  As you can see, the current policy's structure follows this scheme pretty well (and only the description of the meaning of the word "source" is poor), so the deliberately vague claim that "different people may mean different things under "source"" is not helpful but harmful. Moreover, the actual understanding of the term "source" by experienced users is stricter (see above).--Paul Siebert (talk) 19:08, 5 October 2021 (UTC)

In addition, that text is against the policy's spirit. Thus, if I provide just a title of some work, or the author's name only, or a publisher only, that would not be considered as providing a source. Adequate sourcing always means to cite a work, author's name and a publisher, so these three components are inseparable. That is a standard practice in Wikipedia, and our users expect to see that in the policy (and I have a feeling all experienced users interpret this statement in that way).--Paul Siebert (talk) 20:50, 5 October 2021 (UTC)

Ah… but before we added the alert about all three needing to be part of the discussion, editors DIDN’T always consider all three.

As for your “flow”… it is way too simplistic. It does not account for the fact that a source may be “good” in one context and yet “bad” in another context… Reliability is not always a black and white, on/off thing. It is a muddy mishmash of factors that all need to be sifted through to determine whether a specific source reliably verifies a specific statement, and whether it is the best source for doing so. What we require are sources which reliably verify the statements that appear in our articles. Whether a specific source does so often depends on what specifically we write. Blueboar (talk) 22:23, 5 October 2021 (UTC)

A policy is not a philosophic essay on what the term "source" means in various circumstances. It is a set of rules that users are expected to observe. And, it seems there is an unspoken consensus among experienced editors that any source (as it is seen by the Wikipedia community) has three components (as defined above). Therefore, instead of philosophical handvawing, we have just to reflect that consensus in the policy.

My "flow" is not simplistic. It is simple (which is not the same). And it allows further detailisation, which can take into account many nuances, including those mentioned by you. Thus, it takes into account context, but it separates contextless aspects from those where context matters. For example, I doubt anybody can disagree that Oxford University Press is a highly reputable publisher, and that is contextless. Furthermore, everybody agree that Kip Thorn is a highly reputable expert, and his book Black Holes and Time Warps is a high quality book published by a reputable publisher, so all contextless criteria are met. However, in his book, Thorn mentions, among other things, some facts about the Great Purge and kollectivization in Stalin's USSR. Those facts in some aspects contradict to recent historical scholarship, probably because Thorn, being not an expert in Soviet history, took some outdated figures and old publications for his book. That means that that book meets all contextless criteria (a high quality work, authored by a top expert and published by a reputable publisher), but it is not a reliable source in this context. Since not all reliability criteria are not met, it cannot be considered a reliable source for that particular purpose.

Similarly, I can imagine a symmetrical (complementary) example: some blog post that contains more recent facts and figures about the same events, which may make it more reliable (in the context of Soviet history). However, if that source is just a blog post by not notable author, some important contextless criteria are not met, so that source must be considered unreliable, despite the fact that it is quite relevant to the topic and even may (potentially) contain a high quality information.

In summary, I don't see why my approach is incapable of taking into account all nuances mentioned by you.--Paul Siebert (talk) 23:41, 5 October 2021 (UTC)

A policy might not typically be a philosophic essay on what the term "source" means in various circumstances (although we do have some rather philosophical policies), but in this case, telling people that there are multiple meanings for that word is necessary before we can really get into what the rules are that the users are expected to observe about sources.

You have several times made a leap from "The word source has three meanings..." to a belief that these "three aspects must be taken into account to make a decision about reliability". This is not true. These three things can be taken into account to make a decision about reliability, but (a) that is not required, and (b) they are not the only three things that matter for that decision. For example, it is not uncommon to make a decision about reliability without even knowing who the creator of the work is. Other factors, such as whether the source (in "the work itself" meaning) is directly about the subject and whether academic peer review was involved, also matter. WhatamIdoing (talk) 04:12, 6 October 2021 (UTC)

Ok, I called Thorn "top expert", although it would be more correct to say "top expert in Physics". Actually, expertise may have two components, contextless and context related. The opinion of a Nobel prize winner on any subject is more notable than that of a layman (a contextless component), but an opinion of a Nobel Prize winner in Physics is such topics as history may be less notable than the opinion of a assistant professor in history. --Paul Siebert (talk) 00:06, 6 October 2021 (UTC)

Paul, in terms of grammar, your first problem is that you have quoted two sentences. It looks like we forgot to add the terminal punctuation for the second.

I might expand it for you like this:

The word "source" when citing sources on Wikipedia has three related meanings:

• the work itself (the article, book),
• the creator of the work (the writer, journalist), and
• the publisher of the work (for example, Random House or Cambridge University Press).

All three can affect reliability. For example:

• If the work itself has been praised, that suggests that it is more likely to be a reliable source for factual claims (or whatever other typical uses a good Wikipedia editor would cite such a work for). On the other hand, if the work is a retracted article, then it is an unreliable source for almost all possible uses, no matter who wrote it or who published it.
• If the creator of the work is a respected journalist or a subject-matter expert, then the source is more likely to be reliable for factual claims the typical uses a good Wikipedia editor would cite it for. On the other hand, if the creator of the work is especially disreputable, then it is likely to be an unreliable source for factual claims and other typical uses, no matter how convincing the source sounds to you or how good the publisher is.
• If the publisher of the source is reputable, then the source is more likely to be reliable for factual claims the typical uses a good Wikipedia editor would cite it for. On the other hand, if the publisher of the source is disreputable, then it is likely (but not guaranteed) to be an unreliable source for factual claims and other typical uses, even if the article sounds convincing to you and the author has a decent repuation.

In other words, SOURCES tells you that the word "source" is ambiguous in English. It does not tell you what the word "reliable" means. WhatamIdoing (talk) 03:14, 6 October 2021 (UTC)

BTW, if you want to know how to determine whether a given source (including: a particular publication, anything published by a particular author, and anything published by a specific publisher) is "reliable", then WP:NOTGOODSOURCE is my favorite summary.

NB that it's not giving you an actual definition of what a reliable source is. It's only giving you a list of the basic rules of thumb for deciding whether a source is WP:LIKELY to be accepted as a reliable source for factual claims and other statements that a good Wikipedia editor would support with such a source. WhatamIdoing (talk) 03:22, 6 October 2021 (UTC)

Thanks, but that does not solve the problem. The policy is not a correct place to discuss what "source" mean. The policy is supposed to define source (similar to "verifiability", which has some specific meaning in Wikipedia).

That "source" has three different meanings is blatantly wrong: you cannot provide just, e.g. Oxford University Press as a source, you must provide something like John Smith, "On Plenipotentiaryness of Omnipresence" (1982) Journal of Sensationalism, v 12, p 1984, Cambridge University Press. --Paul Siebert (talk) 04:34, 6 October 2021 (UTC)

Ah… you are talking about the citation… in talk page conversation, if you ask me what my source is, I might well say “I read it in the Journal of Sensationalism.” Blueboar (talk) 17:25, 11 October 2021 (UTC)

Sure, but is the policy expected to tell us what words should we use during talk page conversations? AFAIK, only WP:BLP and WP:NFCC are applicable to talk pages.

Maybe, we should focus on the real purpose of WP:V? Paul Siebert (talk) 18:36, 11 October 2021 (UTC)

One of the purposes of this page is to educate new users, and part of that is explaining how words we have chosen to use in our local jargon match with the usages the rest of the world knows. MrOllie (talk) 19:03, 11 October 2021 (UTC)

It does not imply "general education". We need to inform new users about the requirement of this concrete policy, not about various colloquial meanings of some word.

However, if we write something like: "Whereas different people apply the term "source" to (...), this policy defines "a source" as (...).", that may be much more informative and useful for new users. Paul Siebert (talk) 22:19, 11 October 2021 (UTC)

If we were writing a code of laws, maybe, but that's not what we should be doing here. This policy isn't ever going to be some legalistic thing with formal requirements that are applied by a mechanistic process. Too many users (myself included) would oppose that. There's always going to be room for consensus decision making, so the policies will never be as prescriptive as it seems you would like them to be. MrOllie (talk) 01:13, 12 October 2021 (UTC)

Actually, policies document already existing community consensus regarding what should be accepted and what should be rejected. And that is exactly what we are doing here, during that discussion.

In general, the spirit of WP:BURO is: "do not follow policy's letter, follow policy's spirit", and that is not what you say.

And, whereas to follow policy's spirit may be a good idea, that doe not mean we should stop improving policy's letter. Paul Siebert (talk) 16:40, 12 October 2021 (UTC)

FWIW, while I have a personally different idea how this should be done, I agree on the idea that while we still want WP:V and other policies to reflect practice, their language (eg being clear on the difference between a publication and a source) should be formalized within policy to make the discussions on sourcing/reliability/etc. easier for all editors to follow. A lot of the mess at RS/N as a board is that it conflates issues with a overall publication (eg the Daily Mail) and individual sources that may belie the reliability of the overall publication, and those latter discussions shouldn't be had at RS/N - that's a talk page issue for the specific topic. Part of that is this implicit confusion we've created on what a "source" means, which could be resolved without changing practice or policy by just being a bit more formal in definitions. --Masem (t) 16:51, 12 October 2021 (UTC)

I am not sure I understand you. How can we convince people to be more formal in definitions without changing practice or policy?

And, by the way, if we become more formal in definitions, that automatically means that our practice has changed.

Furthermore, leaving these considerations beyond the scope, another problem is still unresolved: the wording of the policy contradicts to normal logic: it says "meaning affect reliability". That discredits Wikipedia: how can readers trust Wikipedia if its editors cannot fix an obvious nonsense in their core content policy? Paul Siebert (talk) 18:42, 12 October 2021 (UTC)

Paul, it says “Author, Work and Publisher affect reliability”. This is accurate. Blueboar (talk) 16:25, 14 October 2021 (UTC)

No. It says "The word "source" when citing sources on Wikipedia has three related meanings: a), b), and c). All three affect reliability" "Three different meaning" - what does it mean? Let's take the word "element" as an example.

According to Merriam-Webster, this word has three essential meanings and additional meanings. It may mean "a chemical element", or just a "a constituent part of something", or even "weather conditions" (in a plural form). However, what is important, one cannot use the word "element" in different meanings simultaneously. When I say "they were killed by exposing to elements", I mean weather conditions, and ONLY that. I do not mean chemical elements. And when I say "compounds are the substances that are composed of several elements", I mean chemical elements, but not weather conditions.

Similarly, if the word "source" has different meanings, that means it may mean either a work, or an author, or a publisher, BUT NOT all three simultaneously. And the policy claims that if under "a source" I mean "a work", the reliability (of what, by the way? That is not specified either.) may be different than if I mean "an author". All of that is a blatant nonsense, how cannot you understand that? Paul Siebert (talk) 17:23, 14 October 2021 (UTC)

There's the way policy is "practiced" (how its implemented, etc.) and then there's the language of how we describe that practice. Changing how the practice should be done is the type of thing that requires a great degree of consensus, but changing how its described without changing the actual practice should be less strenuous (the wordsmithing and the like). Formalizing definitions should be a act that falls into the latter. --Masem (t) 16:10, 14 October 2021 (UTC)

The way this policy is practiced is as follows. The statement "according to Einstein..." is deemed unsourced until a reference is provided, and the reference cannot be just "Einstein". It should be something like "Einstein, Podolsky, Rosen, Can quantum-mechanical description of physical reality be considered complete? Physical reviews, 1935, 47, 777"

Similarly, the statement "According to the "Can quantum-mechanical description of physical reality be considered complete?" article..." also needs a reference to be considered a sourced statement, and the proper reference should have the format as shown above.

Finally, you cannot say "According to APS...", because that statement is also unsourced until a correct reference (see above) is provided.

Therefore, the triad "Work - Author - Publisher" reflects a standard Wikipedia practice, and it is a shame that the policy (i) says something totally different, and (ii) says an obvious semantic nonsense. Paul Siebert (talk) 17:34, 14 October 2021 (UTC)

All I'm saying, which is in alignment with what you are proposing in part, is that we should be explicit on terms, and avoid the way "source" is used nebulous in multiple placed for different context related to "work", "author" and "publisher", and instead use those terms (or equivalent) where appropriate in context on the policy pages, making sure they are well-defined, such that we remove the ambiguity of what "source" means .--Masem (t) 17:39, 14 October 2021 (UTC)

It seems the new version proposed by me in the "Proposed changes (tentative)" sub-section meets these requirement. Do you have any specific comments on it? Is there anything that needs to be fixed/amended/removed? If not, I think it may be a good time to implement the proposed changes. Paul Siebert (talk) 18:14, 14 October 2021 (UTC)

Context matters

This is a theme in several of the comments above, and I'd like to address is separately.

There is a question about whether reliability is "contextless". That is, if a verifiability is context-free, then a reliable source is reliable no matter what. If verifiability is context-dependent, then a source can only be reliable insofar as that source is reliable for the specific information it's being used for in the article.

It is my understanding that the concept of reliability, as used in the Wikipedia:Verifiability policy, is inherently context-dependent. One of the reasons I believe that is the opening sentence of the policy:

In the English Wikipedia, verifiability means other people using the encyclopedia can check that the information comes from a reliable source.

The key point is that other people aren't checking that something called a reliable source exists; they're checking that the information in the encyclopedia article actually came from a reliable source. The information in the encyclopedia article is the necessary context. If the information didn't come from a reliable source, then the very next lines apply:

Wikipedia does not publish original research. Its content is determined by previously published information rather than the beliefs or experiences of editors.

Again, it's all about the information in the article. Either that information was published elsewhere, or the information is not verifiable. You cannot determine whether information is verifiable unless the information matches the sources. If it doesn't match, then no matter how awesome the source looks, that source cannot be used to verify that information. That source is therefore unreliable for that information. WhatamIdoing (talk) 02:26, 6 October 2021 (UTC)

Are you talking about "verifiability" or "reliability"? Keep in mind that, whereas the policy name is "Verifiability", the discussion is not about that, but about reliability.

If we interpret "Verifiability" literally (This page in a nutshell: Readers must be able to check that any of the information within Wikipedia articles is not just made up. This means all material must be attributable to reliable, published sources. Additionally, quotations and any material challenged or likely to be challenged must be supported by inline citations.), "Verifiability" is totally contextless: it is ONLY about a direct correspondence between some Wikipedia text and some stable source, which is potentially accessible to any reader. However, after that, the policy's text switches to "reliability", which is defined very vaguely.--Paul Siebert (talk) 04:39, 6 October 2021 (UTC)

I'm still not understanding what Paul means by contextless, and think it is different to what Masem means and what WAID is arguing against.

I don't think it is helpful to suggest that a claimed source that doesn't actually support the article text is "unreliable". I think a more straightforward point is that a "publication" only becomes a "source" if it does support a statement in an article. Abstract contextless arguments about the reliability of publications do not directly interest our readers (who are not judging whether to take out a subscription to The Guardian or The Lancet). I think that the concern about whether nor not the claimed source really does support the article text is a separate matter than whether it is a reliable source for that kind of statement, but both are essential to WP:V. If the publication cited does not support the article text, it simply is not a source for it at all. The citation is a fraud. At an extreme example, one could imagine someone placing citations to paywalled journals or rare books in the hope that few readers/editors have access to check. Whether those journals or books are reliable for the kind of statement being made becomes entirely beside the point. The degree to which the published text does or does not support the article text is certainly a matter for WP:V, but I would argue that if we conclude it does not support the text then it simply is not a source for it. The citation claims it to be the source for this fact, but it turns out it is not. Often these things can occur through editing where a citation gets misplaced or appears to cover more text than it is really a source for. Or someone rewords the article text, changing the meaning in away that isn't supported by the source. To my mind, this just emphasises how important it is that something is only a source if it really is the source of information for a fact or claim we make. And WP:V requires reliable sources, not any old source.

Paul is focused on verfication being merely checking that the claimed source and article text correspond, and thinks reliability is a separate concept. Hmm. I know Wikipedia has a saying about verification not truth but it is worth a reminder that outside of our peculiar wiki world, the word verfiy means to establish if something is true, accurate, correct, fair. How does our reader establish if something they have read on wiki is true, accurate, correct and fair? Well they can look for what we claim is a source for that text, and read it. But if that source isn't reliable for such a claim, then we haven't really helped them in their mission and may even be leading them astray. So we all agree that the source does need to have a reputation for fact checking and accuracy, that we can rely that when we use it as a source to prop up some article text, that it is likely to be correct, accurate, true and fair about that. If the source is unreliable then we are wasting our readers time making them look it up and read it. If the source doesn't agree with the article text then it doesn't matter how reliable it would have been if it had done. It is a bit like having an expensive watch but expecting it to tell you the weather, not the time. But if the source is unreliable then that's like having a watch you keep forgetting to wind up. There's no point insisting you wear a watch, if you don't also insist that the watch is ticking. -- Colin°^Talk 13:46, 6 October 2021 (UTC)

No, you totally misunderstand what I am focused on. I am focused on modifying the policy to minimise a possibility of long and fruitless discussions such as this imaginary RSN discussion:

Editor A. "Yes. The source X is reliable, because it is a journal publication"

Editor B. "No. This publication is fringe, so the source is not reliable."

Editor C. "Yes. The author is a renown expert, and the journal is a mainstream journal, so the source is reliable."

Editor D. "No. The publication is just tangentially related to the subject, so the source is nit reliable."

(and so on and so forth)

Note, each user puts forward quite reasonable arguments, but each of them tells just a part of truth. How do you imagine closure of such a discussion?

Obviously, a discussion of that type would be much simpler when it is well organized. That is impossible when each participant sees "reliability" differently, or emphasised different aspects of that concepts. One possible way to avoid it would be: explicitly define all components of the "reliability" and specify that the decision about reliability of a source can be made only after each components of reliability have been analysed. That is what I am focused on. And, to achieve that goal, the first step is to stop vague speculations about "different vision of "reliability" by different people", but in explaining how reliability is defined by our policy. If a policy says that different people see some term differently, that means the policy says nothing about it.

The policy must define some terms (such as "reliability"), provide criteria of reliability, and provide rules that help us to check if those criteria are met in each concrete case. The problem is that some terms defined by the policy may have somewhat different meaning than the meaning of the same word outside of our peculiar wiki world. Many conflicts are the result of usage of mixed terminology.

With regard to "verifiability", I mention that term because the policy is called "Verifiability", not "reliability". And the meanings of those terms are somewhat different. "Verifiable" (and WP:V explains that quite clear) means a reader is supposed to be capable of verifying (by themselves) everything what Wikipedia says. That's it (and that differs from your explanation, which you took not from our policy). Obviously, that has no relation to reliability of that information: for example Wikipedia may say "The Moon is made of blue cheese (ref)", and if the reference says the same, a reader can go to a library, take that book and find the very same information. Does it means this information is reliable? No. That source is probably just some fairy tail. Does it mean this information is neutral? No. Majority of sources say otherwise. However, this information is verifiable. That example demonstrates that verifiability is separate from reliability. To that, I would add one thing that is totally overlooked in the policy. Verifiability implies stability: the source must be stable, it is not supposed to disappear in few days. If there is a risk that that source may disappear, it is not reliable (dependable), and should not be used. It does not matter how reliable is the information in this source, if it is volatile, it is not a good source for Wikipedia.

And, as soon as we agreed about verifiability, we can switch to the second aspect, reliability, which is, of course, by and large context dependent (although some its components are not, as I demonstrated above).

With regard to your "a "publication" only becomes a "source" if it does support a statement in an article", that directly contradicts to a standard practice. Usually, the questions users ask at RSN is "Does this source support the statement A?" They rarely ask "Does this publication is a source for A?" Since majority of users prefer the first approach, I see no reason to force them to change it. It also contradicts to guidelines, which define the term source without a link to any context.--Paul Siebert (talk) 18:43, 6 October 2021 (UTC)

There is a lot of text on this talk page, and having reviewed it, including this latest statement, I am still not clear what changes you're seeking to make. Perhaps it is time to put up a concrete proposal so people can see the precise changes that you desire. MrOllie (talk) 18:53, 6 October 2021 (UTC)

Well, you are probably right. Maybe, that will make my position more clear.--Paul Siebert (talk) 22:17, 6 October 2021 (UTC)

Paul, you say I don't understand but then repeat the thing I'm complaining about. Above you again say that 'verifiable' is only about source->text correspondence and is not interested in whether the source is reliable. And again I say that this definition is not only not supported by actual policy text (which literally says "verifiability means other people using the encyclopedia can check that the information comes from a reliable source") nor is it supported with any dictionary definition of "verify" which is concerned with establishing truth, accuracy, fair comment, etc. It is simply a non-starter to try to redefine verifiability to mean "is supported by any published document even a fairy tale written by a fool published by a vanity press". So, no, I don't in fact see anyone who agrees with your definition of verifiability.

Wrt your argument about sources and publications, I'm not actually asking anyone to change their definitions. You claim "Does this source support the statement A?" is a question asked at RSN. Really? I'm sure it happens sometimes, but I don't see that being the generally issue at RSN, which, as the name suggests, is more concerned with reliability. Generally the issue of whether the claimed source, the published document, actually agrees with the article text can be resolved on an article talk page and among wiki project members who understand the topic. But even assuming this question gets asked, the questioner is clearly referring to a published article that someone has already claimed to be a source for the statement. So while technically they could have said "Does this published article, cited as a source for this text, actually support that statement?", we all know what they mean. Paul, it isn't a radical redefinition of terminology to say that a document that does not actually support a statement is not a source for that statement. The word "source" means the source of something, and no amount of misreading of policy can change a basic English word. -- Colin°^Talk 09:42, 7 October 2021 (UTC)

Well, do you agree that the question "Is this information supported by a reliable source?" is a composite question, which can be split onto at least two questions:

1. Is this information supported by some source?

2. Is that source reliable (at least, in some concrete context)? Paul Siebert (talk) 16:56, 7 October 2021 (UTC)

This really is what it is. While technically sometimes a question should have been written "Does this published article, cited as a source for this statement, actually support the article text?", do you actually find that is a common question at RSN?

Proposed changes (tentative)

The first change relates to the definition of the source. Actually, this part is mirrored by a similar text in guidelines, where it is called "definition of source". The text is supposed to define (clarify) something, but it makes a situation more confusing. The "Work - Author - Publisher" triad is by no means represents different meanings of the term "source". I don't know how can that be not obvious, because no user in clear mind can agree that Albert Einstein or Cambridge University Press can be considered as a source. Indeed, a source is a concrete document, for example, the article "Can quantum-mechanical description of physical reality be considered complete?", authored by Einstein, Podolsky and Rosen and published in Physical Reviews by American Physical Society IS a source, but APS alone or Einstein alone are by no means a source.

Second, the discussion of the meaning of the word "source" should be placed into a separate section before a discussion of reliable sources (for the reason that I already explained in previous sections). Therefore, I propose:

• First, change the text from the top of "What count as a reliable source" as follows:

The word "source" when citing sources on Wikipedia is defined as some work (the article, book, internet publication, etc) that was created by some author (the scholar, writer, journalist, organisation etc) and issued by some reputable publisher (for example, Random House or Cambridge University Press). A quality of the work, author's expertise and notability, and publisher's reputation are the three factors that define reliability of the source in a context of some concrete Wikipedia article.

• Second, move that text up, right after the "Reliable sources" header (section 2), and before the start of the section 2.1. The reason is obvious: a definition of "source" is equally relevant to both reliable and not reliable sources. --Paul Siebert (talk) 22:17, 6 October 2021 (UTC)

Paul, consider an editorial in a print newspaper… is “the source” the author (ie the editor who wrote the editorial), the publisher (ie the newspaper as an organization), or the work itself (ie the specific editorial)? Blueboar (talk) 22:33, 6 October 2021 (UTC)

Is it a piece of some text that was signed by someone? Then the work is the editorial itself, the author is the person who signed it, the publisher is a newspaper.

If this text is not signed, then the text expresses an opinion of an editorial board or an editor-in-chief. The rest is the same. What is a problem?

If I were a user who used such a source, I would write the reference as follows: Smith, A. "Is the moon made of blue cheese?", Daily Gerald, Jan. 21, 1999. All components (work-author-publisher) are here. Paul Siebert (talk) 22:44, 6 October 2021 (UTC)

That does not answer my question. If I say “this source is unreliable”, am I talking about the author, the publisher or the specific editorial? Blueboar (talk) 13:21, 7 October 2021 (UTC)

If you say that this source is reliable, that means:

• the work itself passes some criteria of fact checking and accuracy, AND
• the author's expertise and credentials meet some minimal requirements (in this concrete context), AND
• the publisher of the work has a good reputation.

Obviously, if at least one of those criteria are not met, the source is not reliable. There are two exception to that rule in the policy text.

First, if the author is a renown expert, and the work is of a good quality and is relevant to the topic, then a publisher is not too important: that work may be even an SPS.

Second, when poor quality sources are used in a context of a critical analysis of their context, they are also acceptable.

However, in my opinion, it would be more correct to say that in the latter case the source is still unreliable, but the usage of that unreliable source is acceptable for the purpose of its critical analysis and/or discussion. Paul Siebert (talk) 15:09, 7 October 2021 (UTC)

This is where we get into a mess by categorising the reliability of a source without considering the context of what article test is being drawn from it. As WAID notes above, these questionable or self-published sources may even be considered reliable (authoritative) on itself. Unless a source is such quicksand that the publisher changes the text from one day to the next, we can be confident and rely on it being a source for what words it contains. These questionable or self-published sources are generally unreliable for pretty much anything, but there are limited contexts when they are reliable. Anyway, you didn't answer Blueboar's question, and I think his question was a bit rhetorical. He's making the point that you don't actually know which aspect of the source he is referring to, and it could be any one of those. -- Colin°^Talk 15:22, 7 October 2021 (UTC)

I see no contradiction between what you and I say. You say there are limited contexts when questionable sources are reliable, and I agree with that, and I even specify that that context is their "critical analysis and/or discussion". Obviously, there may be no other context where they are acceptable: you cannot present a poor source as a fact, only as an opinion, which requires its placement into some context and supplementing of it with needed commentaries. That is exactly what I am saying.

With regard to Blueboar, let them decide by themselves if I answered the question. Paul Siebert (talk) 15:32, 7 October 2021 (UTC)

Your "unless a source is such quicksand that the publisher changes the text from one day to the next..." is absolutely correct, that is what I myself was saying repeatedly. However, that refers to verifiability, not reliability.--Paul Siebert (talk) 15:48, 7 October 2021 (UTC)

I disagree with these changes, as I find the explanation of the changes, and the result, much much more confusing than the way the policy currently reads. I think it makes sense. Sources are contextual based upon all 3 aspects of their provenance. Makes sense to me. — Shibbolethink ^{(♔ ♕)} 22:56, 6 October 2021 (UTC)

Please provide an example of a source that does not fit that definition. Sources that are generally seen as reliable by our policy are preferable.

And, correct me if I am wrong: is it correct that the current policy's text says that Oxford University Press is a source. Can you give me an example of a situation when you are ready to accept OUP (not some concrete article/book. but OUP itself) as a reliable source? Paul Siebert (talk) 23:01, 6 October 2021 (UTC)

Agree that the proposed replacement for the bullet points at the beginning of "What counts as a reliable source" is a big improvement (which is not to say that it couldn't be improved further). I have always thought that that part was poorly written and confusing. Instead of saying what a source is, it gives three different meanings for "source" and then continues as if there is only one meaning. Awful. To Shibboleth: I think you describe it incorrectly when you write "all 3 aspects of their provenance" — the policy does not describe 3 aspects but 3 "meanings". The difference between "aspect" and "meaning" is critical. Zero^talk 04:07, 7 October 2021 (UTC)

• Can we not vote please (@Shibbolethink and Zero0000:). This is a tentative proposal that is part of a discussion that is far far away from reaching consensus. There isn't even any agreement on what the problem is, never mind a proposed solution. So, if there is to be a poll, that is not now.

While I can see how Paul has got confused about this part of policy text, his proposal is to cement his incorrect interpretation of that view. I think the "meaning" part of the text has been wrongly interpreted to suggest this is a definition of source, and it really really is not. They way I read it, I take it more along these lines.. When someone talks about a "source" they may in fact be referring to the work, the creator or the publisher. For example, if you claimed that MMR causes Autism and someone asked you "And what's your source for that?". You might reply.

• Andrew Wakefield, fellow of the Royal College of Surgeons and specialist in gastrointestinal disorders.
• The paper "Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children" (catchy title eh?), which suggested that a group of children developed certain developmental disorders after their MMR vaccination.
• The Lancet, which is a prestigious medical journal.

The policy text is not attempting to define source as being defined by those and only those components, nor to limit them as the only attributes that concern reliability. So the claim these "are the three factors that define reliability of the source" is simply wrong. These are attributes of a source that are significant to have become proxies for thinking about "the source" in conversational English. But they are not an exhaustive list of the features that make up a source, and the only ones to consider when thinking about reliability. For example, we have repeatedly noted that within a publication (a newspaper or a science journal, say) there can be a variety of types of articles (editorials, opinion pieces, letters to the editor, straightforward reporting, investigative reporting, research papers, literature reviews, product reviews, political commentary, humour, horoscopes, puzzles, questionnaires, listicles, cartoons). The difference between these kinds of articles is in fact, I would say, far far more important a determinant of reliability than the reputation of the journal or the author.

On Wikipedia, unlike conversational English, we do require a source to be quite specifically cited. So "Something I read in the Lancet" is not good enough. For big articles, we even request page numbers. So if we were to define "source", which I'm not convinced we need to, then it would be the cited published text that is used to support a statement on Wikipedia.

If, on the other hand, we wanted to list the attributes of a source that affect its reliability, then I would say this list of three is merely a good start but critically incomplete. -- Colin°^Talk 10:19, 7 October 2021 (UTC)

Actually, you can see above I've confused "publisher" with "publication", when I gave The Lancet as an example. In this case, Elsevier is the publisher. I suspect I'm not the only person confusing this. Do we think "The Guardian" is the publisher? It is really Guardian Media Group. It can get messy as you go up the chain, with media groups having stakes in other groups. So, for the list of important reliability attributes, publication and publisher are both important considerations, and the former can even be more important and more likely to be considered as a proxy for a source. Nobody ever said "The Guardian Media Group" when asked what their source was. A publisher may have reliable and unreliable publications in their house. And I think the type of article is also significant enough to be thought of as a proxy when thinking about a source. Someone might reply "Some research paper" or "A systematic review" when asked about their source. -- Colin°^Talk 10:51, 7 October 2021 (UTC)

Of course you are correct that "source" in informal conversation can have several different intentions. But that is the reason I don't like the wording. This is a formal policy, not a guide to English usage. What the policy should say is "on Wikipedia by source we mean X", and then it can continue by saying that the reliability of a source can depend on several aspects, including its author and its publisher. People can use the word "source" informally as much as they like, but when it comes to giving citations and deciding reliability they have to understand what the formal meaning is. Zero^talk 11:30, 7 October 2021 (UTC)

Well Blueboar has already explained, as the author of the original text, that their intention was not to define "source" but to highlight the multiple meanings already attributed to the word. Effectively, "When someone says 'source' on Wikipedia, they may be using that word to mean ... the author, the work, the publisher, etc". So I can say that Wakefield is a bad source for facts about MMR because he holds contrarian views and has been accused of fraud in his work. I can say that that MMR paper is a bad source because it has been retracted by the publisher. And I can say that the Lancet is often a good source because it has a high reputation. In these matters I'm using the word "source" but in a casual way. This should be obvious because an actual citation reference isn't just "Andrew Wakefield". Perhaps a more correct, formal language might be "Articles written by Wakefield are bad sources ..." and "Papers that have been retracted are bad sources" or "Articles in the journal The Lancet are often good sources" but we all know what people mean by the shorthand of just mentioning the author or the publication.

If people (it seems perhaps non-English speakers?) are being confused about the language here then we could consider changing the text. It seems Paul has inverted the "this is what people might use the word to mean" into "this is what the word means and only means". Maybe this commentary on how people use the word "source" isn't necessary or helpful, and we should stick to just listing a few non-exhaustive attributes that people should consider when thinking about sources. There is no consensus or practice that restricts our examination of reliability to just those three aspects.. -- Colin°^Talk 13:18, 7 October 2021 (UTC)

I perfectly understand Blueboar's rationale, and I find it unsatisfactory. Whereas it may be a potentially good idea to describe all meanings of the word source in, for example, some Wikipedia article, the goal of this policy is different. I believe, we all agree that the policy can be summarized as follows: 1. Everything is Wikipedia must be properly sourced. 2. "Properly sourced" means a reliable source must be provided.

With regard to your other arguments, I see no significant contradiction between what you and I say. Yes, you correctly say that "On Wikipedia, unlike conversational English, we do require a source to be quite specifically cited", but I say the same.

You correctly note that "within a publication (a newspaper or a science journal, say) there can be a variety of types of articles (....). The difference between these kinds of articles is in fact, I would say, far far more important a determinant of reliability than the reputation of the journal or the author.", and I say that the work's quality is the first critical determinant of reliability of the source.

I don't want to look rude, but, maybe, you should just re-read what I wrote: maybe, the difference between your and my vision are not as big as you think?

In addition, I may be wrong when I think that that text is a definition of the term source. However, that wrong understanding is shared by majority of the community: you may easily find the very same text in the guidelines, which reflect generally accepted standards, but that text presented there as a definition of a source. Paul Siebert (talk) 14:46, 7 October 2021 (UTC)

Wrt re-reading what you wrote. Well, the bit about a work having an author and a publisher is not contentious but really isn't telling anyone anything important, other than a source needs to have been published (we say that elsewhere). But saying this is the definition of a source is just wrong. Like confusing "a home" with "a house". The thing that makes a house a home is that someone lives there. And the thing that makes a published document a source is that someone has cited it to support a statement on Wikipedia. And your conclusion that these three attributes of a source are "the three factors that define reliability" is also wrong, and you should re-read what WAID said about that further up. -- Colin°^Talk 15:58, 7 October 2021 (UTC)

In other words, according to you "a source is some piece of information that was cited by some Wikipedian in some Wikipedia article", and, according to a policy, "any information published in Wikipedia articles is supposed to be taken from some source". Doesn't it look as a circular reasoning?

And, using your analogy, during a discussion about only those buildings where their owners are living, a difference between "a house" and "a home" becomes insignificant. Here, we are discussion only those pieces of text that are being used (or are supposed to be used) in Wikipedia, so your argument, although formally correct, is totally irrelevant.--Paul Siebert (talk) 17:05, 7 October 2021 (UTC)

With regards to WAID's opinion, I am not sure what exactly do you mean, but if you mean the 02:26 (6 October 2021) post, I see no serious contradiction between what WAID and I say. As I already demonstrated, some components of the definition of "source" must be analyzed in a concrete context, which makes the whole concept context-dependent. That does not mean, however, that there are no contextless components, which I persuasively demonstrated. Paul Siebert (talk) 16:47, 7 October 2021 (UTC)

Regarding your "you can see above I've confused "publisher" with "publication"". We are reasonable people, and we all agree that that type "confusion" is quite acceptable during talk page discussion. But the policy means not talk page discussions, right?--Paul Siebert (talk) 16:39, 7 October 2021 (UTC)

Well, where else is the reliability or source->text correspondence to be discussed other than on a talk page? Paul, when you write about things you agree with me, you misunderstand and misquote, and when you write about things you disagree with me, you misunderstand and misquote. It is a bit frustrating. Your supposed quote of mine (above, 16:47) isn't what I said and your supposed quote of policy isn't what it says, and the bits you miss out or change are important. I don't find it productive to have an argument where most of it involves "No, that's not what I said" and "No, that's not what policy says". Paul, I'm going to take another break from this discussion. Can I suggest please that you take the weekend off of this to get a fresh pair of eyes and a clear head. -- Colin°^Talk 18:04, 7 October 2021 (UTC)

Obviously, if I misunderstand you, that is totally my fault, and the idea that your posts are unclear or ambiguous is totally unrealistic :)

With regards to your "where else is the reliability or source->text correspondence to be discussed other than on a talk page?", do you really believe this policy's goal is to set talk page discussion rules? I thought it is quite clear from the policy text that it tells about the article space, and only about it. Accordingly, it must say not what people may mean under "source", but what should be considered as a source per policy. A policy, by definition, must introduce some, imperative concepts, and "source" is one of the core concepts of WP:V. If a clear and unequivocal definition of this term is absent, that is not a policy, but a piece of ... (some unspecified, totally useless substance).--Paul Siebert (talk) 18:52, 7 October 2021 (UTC)

If I were going to try to define the word sources, I think I'd probably re-write it along these lines:

On Wikipedia, a source is any person, place, or thing that a Wikipedia editor gets article content from.  Some sources, such as unpublished documents and private e-mail messages and your own personal experiences, are banned.  Note:  Sometimes, when editors talk about "a source", they are referring to a document or similar work; other times they are referring to the work's creator or to the publisher.  If you are uncertain which meaning the other editors are using, you may need to ask whether they're referring to the specific document, anything written by that author, everything published by that publisher, etc.

Paul, the "reputable publisher" dooms your proposal above, even if there weren't any other problems. That proposal defines Twitter as not being "a source". Donald Trump's tweets are cited in dozens of articles. Neither Trump nor Twitter is considered a reputable publisher. WhatamIdoing (talk) 21:28, 10 October 2021 (UTC)

Correct, if that definition is supposed to cover all categories of sources, the word "reputable" should be removed. With regard to the rest, I believe it is obvious that Trump's Twitter posts are SPS, so everything depends on Trump's own expertise in each concrete context. A situation when some top ranked state official is using Twitter as a primary communication channel is very unusual, so this SPS may be probably considered as reliable, and Trump's expertise (as an author) and his reputation (as a publisher) may range from very low to very high, depending on context. However, I see absolutely no problem with that definition if we remove the word "reputable". Paul Siebert (talk) 00:38, 11 October 2021 (UTC)

And, with regard to your text, I disagree with it.

- "a person" is not a source, because two other components are lacking: "what exactly that person said?", and "where this information can be found?". Instead of "a person", we should say "a statement X made by a person Y and published in Z" (where XYZ is the triad: "work-author-publisher"). If at least one of those components is missing, this source does not meet verifiability criteria (which literally means anyone can potentially find this information and check it).

- "a place" is not a source for the same reason: how can I (as a reader) verify anything if I know "where" but I don't know "what" and "who"?

- "a thing" is too colloquial, but if that "thing" has an author, a title, and a publisher, then it is ok, but that is the same what I wrote (although, again, the style is too colloquial to be a part of a policy).

And, the most important problem is that we do not need to define (or describe) what different people see under "source", we need to define how the word "source" is defined from the point of view of the policy.

Imagine if we defined used the same approach in other policies for example Sometimes, when editors talk about "a non-free content", they mean the content you must pay for, other people think that is any content that was not issued under CC, other people think etc.. That is not what the policy is needed for. The policy must say:

• Everything in Wikipedia must be verifiable.
• "Verifiable" means some source must be provided for any statement.
• "Source" means .... ( and this part is currently missing, for the policy says people usually mean under "source", but it doesn't say what the source is per policy itself).

- Paul Siebert (talk) 01:44, 11 October 2021 (UTC)

"A source" does not imply that the source is good, reliable, usable, etc.

For the rest:

• If "a person" is not a source, then please remove the line in SOURCES about "the creator of the work (the writer, journalist)". Also, if "a person" is not a source, then we should never see any editor say "But Paul Politician said this last night" or "He is not a reliable source about medicine" or "When I took chemistry class, the prof said...", and we certainly wouldn't have bothered to write something like WP:You are not a reliable source, because if a person can't be a source, that title would be nonsensical. But we do see these things, and therefore a person can be a source. (Remember: "a source" isn't "a source that you can use in a Wikipedia article". We're just defining "a source" here.)
• If "a place" is not a source, then we need to delete half of Category:United States highway citation templates and probably Template:Cite sign, too.
• "A thing" is the most common source (and what I recommend for >99% of situations), but we do not require authors (see the FAQ at the top of this page, or titles ("Untitled" is the most common title for artwork – yes, you can cite artwork directly), and we don't need to know who the publisher is. These are all useful things if you're trying to determine whether (and if yes, how) the source should be used in an article, but they are not fundamental to the definition of "a source". An anonymous letter left on your doorstep is "a source" if you get information out of it. It does not stop being "a source" of information just because I tell you that you aren't allowed to cite it in support of a statement in a Wikipedia article.

Also, "verifiable" doesn't mean that some source must be provided for any statement. Verifiable means that it must be possible to provide a source. It doesn't mean that a source must already have been cited.

Are you instead hoping to define "a reliable source"? WhatamIdoing (talk) 19:22, 11 October 2021 (UTC)

Re: ""A source" does not imply that the source is good,.." - absolutely correct.

"A source" is something that can be used for independent verification of Wikipedia content.

"Independent" means "by a random person (not by a user who added that information)".

"Verification" means "comparison of the some concrete fact found in Wikipedia with the content of the source this information was taken from". "Verification" does not mean checking if the information in that source is correct.

And, finally, "reliability" is a property of the source: it is measure of relevance, fact checking and accuracy of some concrete source.

From that, the difference between verifiability and reliability follows.

Some information is verifiable if a source (a reference) is provided.

Some verifiable information can be taken from a source that is not reliable.

"A person" in not a source, it is a component of most sources. If I say "According to Einstein", I always mean not Albert Einstein as a human being, but some concrete piece of information produced by him. The statement "according to Einstein, two quantum objects may be entangled" is NOT verifiable (although acceptable in talk page discussions). However, the statement "according to Einstein two quantum objects may be entangled" IS verifiable. The difference between these two statements is obvious: in the first case, the statement cannot be independently verified, because it is unclear where original information can be found (unless you imply one has to read everything what Einstein wrote); in the second case, the information is easily verifiable. That means a person is not a source, but it is an essential component of most sources. I can imagine just few exceptions to that rule: one exception is the information produced by some organizations (WHO, UNO, etc), another exception is old historical documents (chronicles, etc). There are probably some other exceptions, but that is not common.

"A place", similarly, is a component (trait) of any source. Literally, it means where this information can be found. Usually, "a place" and "a publisher" means essentially the same.

I can say it in another way: every description of a piece of information that makes verification technically feasible can be considered a source. And vise versa if it is not clear from your description how can I independently verify some information, I can claim that no source has been provided.

With regard to "we do not require authors", usually any information is produced by some person. Sometimes, by some organisation (see above). There may be some anonymous information, but that is more an exception rather than a rule.

"Verifiable" means a source can be provided for any statement. It also means that anyone can put a cn tag to any statement (except totally obvious ones), and if a source is not provided, that information can be removed. I am not sure we have much disagreement about the last point. Paul Siebert (talk) 22:01, 11 October 2021 (UTC)

Consider the classic "personal communication" citation: "I saw Karp in the elevator, and he said it was probably NP-complete." Is Karp not a person? Is Karp not being cited as the source of my information? WhatamIdoing (talk) 01:17, 12 October 2021 (UTC)

That is just an abstract question that has no relationship to Wikipedia. You cannot write in Wikipedia: "According to Karp, it was probably NP-complete", and the reason is: it is not verifiable. If, for example, I want to verify that statement, I cannot do that: your conversation with Karp has not been published, and even if it was accidentally recorded, no information is provided on where that record could be found. However, if you write: "According to Karp, it was probably NP-complete", that becomes a source. Because everybody can see who said what, and where that information can be found. Paul Siebert (talk) 04:54, 12 October 2021 (UTC)

References

Historical Sources

I've seen editors source historical 'facts' from tourism, sociology and linguistic papers. I'm against this because the authors aren't historians and the papers haven't been published in history journals and therefore haven't been peer reviewed by historians. The text on this page: "If available, academic and peer-reviewed publications are usually the most reliable sources in topics such as history, medicine, and science." could offer them wiggle room since they could say that it's OK to include these papers because they are "academic and peer-reviewed". Perhaps we need a specific page describing where to find the best sources for history? Is there already one? Cheezypeaz (talk) 19:04, 13 October 2021 (UTC)

• Do you think these sources are unreliable, or just not the most reliable? Blueboar (talk) 19:12, 13 October 2021 (UTC)

Blueboar An actual example I have seen is: A tourism paper published in a tourism journal exploring the posibilites of ethnic tourism in a country which gives a potted history of the country. The potted history was used as a source for a historic claim in the wikipedia article. The potted history part was wrong. I would class them as unreliable because even though articles like these were written by intelligent people they were masquerading as "academic and peer-reviewed" which they are not from a history standpoint. So worse than a tweet from an unknown user on Twitter. Cheezypeaz (talk) 21:07, 13 October 2021 (UTC)

Cheezypeaz, are you claiming that historiography provides "facts" while linguistics and other disciplines don't? Science, like physics and chemistry, provides facts that are not dependent on political ideology. History and how it is written very often changes according to the political ideology of the historian and the political sensitivities of the society it is written in. It is very important that we don't confuse the concept of "facts" with scholarly POVs, opinions and interpretations. I am not saying that history books and articles don't contain some "facts", only that views provided in history books aren't automatically to be considered "facts".--Berig (talk) 05:31, 14 October 2021 (UTC)

I also find it a bit odd to throw sociology and linguistics in the same category as works written with a tourism-related purpose. Very often the different disciplines which study mankind are covering closely inter-twined topics, and so it is normal that debates arise about whether we should cite, let's say, a population genetics article in a WP article about an ethnic group. (Often these properly come down to discussions about what aspects of a topic that a source is best for. For example, geneticists are not good with historical detail as far as I have seen!) I do have concerns about how these often play out, but OTOH I'm not sure that such cases can easily be handled with a simple general rule. Concerning tourism-based sources, I think indeed that type of background is not likely to provide a strong source, but OTOH does that mean we need to treat all of them as unreliable? I think in a case like the one you describe the WP guidelines are already pretty clear?--Andrew Lancaster (talk) 10:18, 14 October 2021 (UTC)

Actually, the main point Cheezypeaz is making is as follows. We should always discriminate between reliable sources authored by experts in this concrete field and equally reliable sources authored by non-experts. I can continue the analogy with chemistry or physics to demonstrate that. The article authored by some expert in inorganic chemistry is less likely to be a high quality reliable source for some organic chemistry topic, especially if it just mentions some organic chemistry fact in passing.

Another example. Recently, we had a very long discussion at RSN about some peer-reviewed article that discussed a possibility of artificial origin of COVID-19 pandemic. That article was published in a very respectable Bioessays, and it was authored by experts in biology. According to our formal standards (and even taking into account additional standards described in MEDRS), that source is absolutely reliable. However, majority of users rejected that source, and among the reasons were (i) Bioessays does not focus on virology, and it has no experts in that field in its editorial board, and (ii) the authors are not experts in this concrete field.

Similarly, linguists, similar to the authors writing about tourism, are not experts in history, and, whereas they may reproduce some facts or opinia in their works, their expertise is hardly sufficient for providing independent and authoritative analysis, and, therefore, it may contain errors, omissions or distortions.

Currently, we are discussing a better wording of the policy, and I maintain that the author (their expertise and reputation) is an important criterion that affects reliability of a source in some concrete context. IMO, a situation with usage of non-history works as sources for history article is a good example of possible benefits from amendment of the policy's language. Paul Siebert (talk) 15:42, 14 October 2021 (UTC)

Paul Siebert Yes to all you have said. Thank you. Cheezypeaz (talk) 16:35, 14 October 2021 (UTC)

Things aren't always that clear-cut. When you write about topics at the borders of history, as I do, there are few clear lines between history, linguistics, literature, archaeology and religion studies. It can become interesting when a historian states something categorical and ideological about a linguistic (toponymic issue) and an archaeologist has a completely divergent opinion. If the policy is rewritten I hope it takes into account the fact that some topics are cross-disciplinary.--Berig (talk) 16:41, 14 October 2021 (UTC)

Well, that is correct, and some facts from tourism, sociology and linguistic papers may shed an additional light on the history related subjects. However, that is acceptable if, and only if they add some information that does not contradict to what mainstream historians say (and mostly is specialised peer-reviewed publications). However, it sometimes happens that whole sections in some articles, or whole history articles are written from a perspective of some non-historical sources. What is worse, some articles are based on non-specialist writings, and some additional facts are cherry-picked from good quality history publications to create a false impression of a broad support of those views by scholarly community. If you want an example, I can provide a link to some concrete articles. Paul Siebert (talk) 17:40, 14 October 2021 (UTC)

I believe you.--Berig (talk) 18:00, 14 October 2021 (UTC)

Generally I agree with Berig that there are not always clear lines between history and other branches of the humanities. And thank you for the link to the Battle of Fýrisvellir article. TSventon (talk) 23:04, 17 October 2021 (UTC)

If I'm not mistaken, what User:Cheezypeaz is concerned about is the sources used in a particular article (Welsh Not) which s/he believes he has identified as problematic. S/he does not feel that an academic work about language teaching is a suitable source for a statement in an article about a historic tool for language teaching, and has lumped this in with other sources, such as tourist guides, which s/he rightly thinks are transient and superficial. I don't personally believe this is a relevant comparison. Deb (talk) 15:55, 15 October 2021 (UTC)

Deb The linguistic example is a good one. It strongly implies that X occurred more than 150 years ago and therefore Y also occurred 150 years ago. Historians specialising in that area explicitly state that neither X nor Y occurred. Cheezypeaz (talk) 20:11, 15 October 2021 (UTC)

Cheezypeaz Anyone who looks at Talk:Welsh Not will see that this is a complete misrepresentation of the situation. Deb (talk) 08:16, 16 October 2021 (UTC)

This page should mention intereviews

Preferably with a link to Wikipedia:Interviews. I think it should clarify if interviews are "questionable sources" or not, among others. _{Piotr Konieczny aka Prokonsul Piotrus| reply here} 08:45, 15 October 2021 (UTC)