Wikipedia talk:Attribution/Archive 8

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 5

Archive 6

Archive 7

→

Citing yourself

The "citing yourself" section says that you may cite yourself, but only if you're considered a reliable source. However, the link to WP:Autobiography gives stating your shoe size as an example of original research.

Information from a reliable source is, by definition, not original research, and you're considered a reliable source for information about yourself in articles about yourself.

These rules contradict, unless you want to interpret them in very strained ways. (for instance, allowing your shoe size if it's on your web page but not if you first mention it in Wikipedia). Ken Arromdee 15:14, 29 December 2006 (UTC)

A source is not reliable for Wikipedia purposes unless it is verifiable. For information from the subject of a biographical article to be included, the information must be fixed in some verifiable form, such as the subject's web site, a published autobiography, etc. The Wikipedia userid George Bush is not a reliable source of information about any of the people named George Bush because there is no mechanism to verify that the Wikipedia userid is actually under the control of a person named George Bush, or if so, which one. Even if an ad-hoc mechanism could be devised for one particular article, the ordinary reader could not be expected to understand or believe such a mechanism. --Gerry Ashton 18:19, 29 December 2006 (UTC)

Try me. I'd like to see if I could understand or believe in such a mechanism. Can you do that, please?74.38.35.171 02:14, 1 January 2007 (UTC)

It would be original research to state your shoe size if either nobody has published that anywhere but Wikipedia, or you publish it in your self-run publisher and that publisher has not gained enough reputation to be considered reliable. 74.38.35.171 02:14, 1 January 2007 (UTC)

Relativating WP:NOR and WP:V

One should not overemphasize demands for sources and take a moderate stand as to determine what is original research:

Statements for which, if correct, it is most likely to find proper sources, should generally provide a proper source.
Deductions that throw a whole new and surprising light on a topic, require a proper source.
Statements which, if these would be false, can be expected to be immediately spotted and counter-acted, do not necessarily need a source.
Deductions that merely allow generally synthesizing numerous sources into a simple sentence or paragraph without causing a statement that can be expected to be entirely new or controversial, and seem likely to have been widely assumed by people who probably did not thought it to be new and important and therefore did not explicitly publish the evident synthesis or conclusion, should not be wiped as if it were highly original research.

I am now thinking of articles like French fried potatoes, in which is stated that in Belgium, the thickness of the cut potatoes decreased over the years. One cannot assume the Belgian journalists or scientists to have published such, as the decrease was never sudden, and yet millions of people have been eating thinning fries and thus – with daily contributions to the article – one can safely bet that if this statement would be incorrect, it would quickly have been disputed. It would then be wrong to have a single user living in Wales, to kick out the relevant phrase under pretext of it being original research without proper source: most sourced material gives less guarantees on the correctness of statements. I'm well aware of Wikipedia articles not necessarily having to be correct, but rather reflecting what is considered known, which can be shown by sources. But it is an encyclopaedia and readers expect it to be right, not a press review. Thus the importance of a subject (French fries are notable and the article is busily edited, but few people would call it a highly important subject – Most noteworthy event of 2006? I have eaten French fries!), the nature of the statement (by millions effectively experienced as verifiable without published sources), the degree of controversy (a larger thickness of chips is considered essential by its British consumers, but perhaps not by Americans and moderately by Belgians; but the evolution in Belgium is not controversial), the consequences of what the statement describes (thinner fries have a larger surface exposed to oil than a same weight of thick-cut potatoes, thus causing a higher consumption of fats) ... should all be considered with some common sense that cannot be reflected by exact rules. — SomeHuman 3 Jan 2007 03:25 (UTC)

I disagree. This statement needs a source as much as any other: we can't expect wikipedia readers to take it on face value. Yes, for the reasons you suggest we can assume that it is correct, but I'd hesitate to say that means it doesn't need a source. What I'd suggest it means is that it shouldn't be a high priority for a source: the statement shouldn't be removed because of the absence of one, and probably that no {{fact}} tag should be added that may be interpreted by readers as casting doubt on the correctness of the statement. JulesH 09:47, 3 January 2007 (UTC)

It will most likely be impossible to find a published source. One just might find a 40-year old, a 20-year old, and a recent text that happen to mention the thickness of French fries at (as mentioned in the article) street fries stands; even if the figures would not have been merely visually estimated, it is most unlikely that any of these figures would be reliably valid for the whole of Belgium. So, seeing the probability of correctness, you suggest not acting upon the lack of a source. I quite agree, but would like to see such advice in the to-be-guideline 'Attribute'.

Suggestion: Perhaps we could have a {{common}} that would show a minute symbol in the text and insert a footnote (as a ref 0 position) in the <references/> section, behind that same symbol stating: No reference for this statement is brought forward, but it appears of a nature allowing factual accuracy to be assumed for not being disputed. (only once in case of several {{common}}).

Details: This could be done with current <ref name="common">No ... disputed.</ref> but that would cause a seemingly true reference number, inviting to be clicked. After a few of such, readers would get enervated. Therefore we need a recognizable small symbol. I'm sure some bot will soon track recent {{common}}-tags and its supervisor can spot unduly usage. The occasional reader who spots the symbol while knowing where to find a real source, might feel compelled or at least invited to improve the article. — SomeHuman 3 Jan 2007 18:56 (UTC)

Woo, I just quoted W:ATT for the first time!

I've argued against the awkwardness of wiki-shibboleth "verifiability, not truth" before and today realised I was getting nowhere with it at Wikipedia:Articles for deletion/Thakur Sher Singh Parmar; so I whipped the better terminology we have here out of my sporran. For this proposal to be accepted as better than the existing policy pages, it has to be more useful in practice, and I very much found it so in this case. (More assessors of that page welcome, by the way, as the "keepers" are springing up from the ground like dragons' teeth, all dressed in red type.) qp10qp 17:24, 3 January 2007 (UTC)

S'OK. Bingo! Deleted.qp10qp 23:01, 3 January 2007 (UTC)

This policy is much clearer, excellent job!

The formulation I think is brilliant: "Wikipedia is an encyclopedia, not a publisher of original thought. The threshold for inclusion in Wikipedia is whether material is attributable to a reputable published source, not whether it is true." I would even make that bold, red, blinking and all the worst html tags, if it would only make people finally get it. This would make life on wikipedia so much easier! I really hope this formulation becomes official policy as soon as possible. See the discussion on Talk:Ontological argument#Rejected by aquinas where I'm denied the inclusion of "cited untruths" and you'll see that this reformulation is urgently and desperately needed! I mean, the guy I'm arguing with seems to be a very reasonable person, but he has simply not understood wikipedia policy. --Merzul 18:06, 3 January 2007 (UTC)

Damn it, I just noticed that WP:V already states "The threshold for inclusion in Wikipedia is verifiability, not truth." and it is boldface, yet I had actually missed that myself. It seems then sadly, that it is not actually the problem of the formulations, but that we editors simply don't read the policies :( I still believe this is a better proposal because WP:V and WP:NOR are just the two sides of the same issue, and maybe if there is less policy, we might actually read it. --Merzul 18:22, 3 January 2007 (UTC)

Also, as I said above, "Verifiability, not truth" isn't always grasped by editors, because, paradoxically, they may disregard it if they believe they know the truth, or they may think "verifiability" applies to the truth rather than to the source.

I've just had a glance at your disputes over this, and of course, you are entirely right in your position. But if you can't prevail there for the time being, may I suggest leaving the article for a week or two and coming back to make your edits then. That's my approach in such cases; often the point does sink in with editors over time, but in the short term they may be blinded by reactance. I admit I've sometimes energetically argued a point and not realised till the battle had cooled that I actually had the wrong end of the stick all along. I now follow a thing I call the "3 ar" rule (in real life too): I'll argue a point three times in one conversation and then stop: this does me a favour whether I am right or wrong. qp10qp 19:20, 3 January 2007 (UTC)

Yes, you are right about the "3 ar" rule, I have calmed down about that argument; but it is funny that all this time, the statement "verifiability, not truth" had not stuck in my mind. I think you are absolutely right: verifiability is too close a relative to truth, while attribution, not truth is much stronger, so again, this proposal is urgent needed! :) --Merzul 19:45, 3 January 2007 (UTC)

But I'm still very mischievous, I created WP:NOTRUTH that I can use next time :) It goes directly the FAQ entry. --Merzul 23:46, 3 January 2007 (UTC)

Analysis of facts from statistical surveys

When I first read WP:NOR#SYNTHESIS, I was very suprised because I think I have broken that rule myself and seen it broken many times on wikipedia. Taking the synthesis claims to their logical conclusions, here is another example about analysis of sources. I have written it like the FAQ entry:

Example of unpublished analysis of published material

Facts and statistical surveys can be used (or abused) to argue a conclusion that are not attributable to the original study. Statisticians have methods and criteria for when correlations are valid and significant. Drawing conclusions based on statistical data is therefore original research.

Here is an example from wikipedia:

Although some studies show an inverse relationship between education and religiosity, there are several counterexamples of religious groups among which a positive correlation between educational level and religiosity has been shown.

And here we have a number of facts, all of which are attributable to a study:

A 2004 study by the General Social Survey showed that in general 30.4% of those with a graduate degree attend religious services weekly or more. This was higher than any lesser educated group. Further the group with the highest percentage of "never attending," was composed by those with less than a high school education.

The use of these facts to support the claim that there is a correlation between educational level and religiosity, is not from the study, but our own. In the wikipedia article, editors that disagree with the implied conclusion of this study have done some original research of their own and added other facts from the study:

However, those with graduate degrees were the least likely to believe the Bible as the literal word of God. Moreover, graduate degree-holders had the highest percentage of a religious preference of "Jewish" in comparison with the lesser educated groups.

The inclusion of these facts, both in favor and against the conclusion, are cases of original research that are not attributable to the cited source.

Discussion (of statistical analysis)

So what do you think, do these inclusions of fact constitute original research or not? --Merzul 19:58, 3 January 2007 (UTC)

Determining that a statistically significant positive or negative correlation exists requires carrying out a mathematical analysis, and is distinct from dividing people by education and observing how many people answered certain questions with certain answers. It would be original research for a Wikipedia editor to carry out the mathematical analysis; that is a job for the authors of the source. --Gerry Ashton 21:44, 3 January 2007 (UTC)

If all of the cited facts come from the study (which is the way I read it), I don't see any problem with this (except, perhaps, that it creates a POV that does not exist in the original source... but that's not what this discussion is about). Merely selecting parts of a published work to use in order to illustrate a point does not constitute OR. JulesH 22:52, 3 January 2007 (UTC)

Introducing a POV that is not in the original source is precisely what WP:SYNT is all about. And while selective quotation from a source is not OR, it is still horrible research. I generally lose any respect for authors that claim Darwin believed the evolution of the eye was "absurd in the highest degree." This is not proper attribution, one should represent the idea presented in the source, not cherry-pick whatever furthers one's own POV. --Merzul 03:10, 10 January 2007 (UTC)

Yes, but the policy for dealing with that is WP:NPOV, and therefore it doesn't need to be addressed in this proposal. JulesH 09:11, 10 January 2007 (UTC)

I don't think it's OR. Squidfryerchef 00:27, 4 January 2007 (UTC)

It seems this is a difficult issue, and whatever conclusions we reach, I will move the above example to WP:ATTFAQ with a clear answer that "this is" or "this is not" original research. Personally, I feel very strongly about this. I think this is original research of almost the worst kind. It is the kind of research that would put serious statisticians out of business. I don't mind stuff like List of songs containing covert references to real musicians because those aren't misleading: you can choose to believe each of those statements or not, and if we can find sources for them, then all the better. The above example, however, is harmful and misleading! It cites a number of facts as implying to support a conclusion, and even if I know enough about statistics to know that it is not that easy to establish correlations, I somehow still think that the argument makes sense and it feels backed up with evidence, and that's what I mean with harmful and misleading, because the evidence are not really supporting my intuitive conclusion! --Merzul 03:10, 10 January 2007 (UTC)

But it doesn't (as I understand it) state any conclusion that isn't in the cited paper. Now, if it isn't presenting the whole story then that is clearly a problem. But the problem is one that's already dealt with by WP:NPOV#Undue_weight. JulesH 09:11, 10 January 2007 (UTC)

You are correct that no policy works in isolation, but the current example in WP:SYNT also does not state any conclusion that aren't in the cited sources. The problem is that juxtaposing two cited statements create new meaning: "Jones didn't commit plagiarism", and these implied conclusions are not backed up by any sources. I think it is the same here. --Merzul 12:25, 10 January 2007 (UTC)

Some changes I made...

Well, I restructured this page, that's just a matter of taste, and can easily be reverted, but I wrote in some sense rather specific guidelines about requesting sources. My main reason is that I wish I had seen something like that when I began editing wikipedia, and in particular the {{dubious}} template, which has a link to the exact section in the talk page, is extremely helpful. There are so many pages with a huge disputed tag on top of the page, and then you click on the talk page link and you don't find what the problems were, etc. The dubious template I think is heavily underused, see for example this edit on BBC, an entire article was disputed because of a simple argument over whether it is the biggest or not! (it was finally solved as "one of the biggest" :) I'm also very pleased with the use of it here, because it let's the reader understand what was the issue, and if some editor who knows more about the issue than I do, they can probably fix it. So well, what was my point? Just having slightly more concrete guidelines for what to do when facing unattributed statements would be very helpful, and also encouraging the use of templates that link to the right sections in the talk page, rather than blanket DISPUTED tags, I think... --Merzul 02:18, 4 January 2007 (UTC)

Why haven't my edits been reverted yet? Don't people care about this proposal anymore... Please don't abandon this project, the formulations here are really cleaner than official policy and it could make a difference. Of course we can live without it, and I am quite happy with people's behavior on wikipedia, but it still bothers me that reasonable people can have completely different understanding of the policy. --Merzul 21:12, 8 January 2007 (UTC)

I reviewed your changes, and only found one point to quibble about. I changed from never accepting a self-published expert over a peer-reviewed publication to hardly ever. In the FAQ page I mentioned two exceptions: an author issuing errata on his/her personal web page, and fields that have become inactive, and the only up-to-date publications are self-published experts. --Gerry Ashton 21:38, 8 January 2007 (UTC)

Yes, indeed, and I also agree with your last change, I went a bit too soft with adding "harmful", if one is certain that an unsourced statement is wrong, then one should of course remove it. (It will almost certainly be reverted back in, and then one can discuss it :P) --Merzul 02:36, 10 January 2007 (UTC)

In-line citation

I regret seeing the mandatory in-line citation business show up here; this confuses a necessary result with three mechanisms which can accomplish it. Citations can be clear and precise without being in-line; for example, Homotopy groups of spheres is (with a handful of clearly marked exceptions) largely drawn from two sources, neither much longer than the article itself. In a shorter article, regular numbers does the same thing with one source.

On the other hand, in-line citation is not a warranty of clear and precise citation. Consider this edit: The page cited supports only the claim made in the text added; and does not mention the rest of the paragraph (which is verifiable elsewhere). I included the word "scandals:" to show this; the footnote would be incomplete without it. Septentrionalis PMAnderson 06:37, 11 January 2007 (UTC)

Yes, but what about articles with a bunch of offline references and multiple points of view represented? It's well and good to say "Everything's sourced! Look, there are references!" but having to comb through 10 articles to find one minute fact is just annoying, especially when the author of the statement could have just added an inline citation in the first place! Besides, the space where people assume everything is WP:V and WP:NOR-compliant because there are too many references to get through is also the space where the OR and outright falsehoods come in. In short, inline citations are key to WP:V, and really shouldn't be that much extra work assuming articles are being sourced right in the first place, so why not require it? GertrudeTheTramp 06:58, 11 January 2007 (UTC)

Because doing so will require it where it is not useful or appropriate. The case you envisage the usual case; that's why clear and precise citation "usually" has to be inline citation. But it's not true in every case, and if policy says it is, some fool will be unreasonable about it, even where no-one in this discussion would be. Septentrionalis PMAnderson 17:49, 11 January 2007 (UTC)

That said, maybe there's room in the policy to allow for some sort of "Where not otherwise specified, the source of factual statements is XYZ" when the majority of statements in an article come from one or two sources? Would that satisfy your objection re: Homotopy groups of spheres? GertrudeTheTramp 07:01, 11 January 2007 (UTC)

I think certainly there's room to give a general source for the majority of an article and specific sources for individual claims not covered by the general one. That seems to me to be an acceptable way of organising an article, and probably the most convenient way to work on a whole class of articles, particularly well-established academic topics where the current state of knowledge isn't changing rapidly. I know it's the way I originally wanted to work on Kernel (computer science), although the number of different sources eventually worked out to be too large there for it to work. JulesH 08:59, 11 January 2007 (UTC)

Equally sorry to see the Math Project's ongoing objections to citing articles (fallout from WP:WIAGA) show up. I'm not a fan of the "Where not otherwise specified, the source of factual statements is XYZ" approach because it fails to account for the dynamic nature of Wikipedia. Many articles are not regularly watched by original editors, and subsequent unsourced edits would be inaccurately included in that umbrella - it works for hard print, but not a dynamic document which anyone can edit. The article could appear to be well sourced when it's not, and as we've seen at WP:FAR, sorting it out two or three years down the road can be impossible - much easier if the article is accurately cited. SandyGeorgia (Talk) 14:43, 11 January 2007 (UTC)

And I, in turn, am sorry to see attribution/citation (which a reference at the bottom of the article manifestly satisfies) confused with inline citation (which should be a stylistic and pragmatic decision for an article). Inline citations are often a good idea. They should not be mandated by policy. Robert A.West (Talk) 16:46, 11 January 2007 (UTC)

Why? GertrudeTheTramp 17:08, 11 January 2007 (UTC)

(edit conflict) Because I do not believe in falsehood. They are two different concepts. A reference is a reference. If the reference is not inline, it still remains a reference. One may believe that an inline citation is needed in particular circumstances, or one could believe that Wikipedia should require that every morpheme be footnoted, but it is simply false to say that only inline citation is attribution. Robert A.West (Talk) 17:41, 11 January 2007 (UTC)

(edit conflict) Because they should be encouraged by guidelines. We should be inflexible on our ends, and flexible to accept any means that accomplish those ends. Therefore our ends should be policies, and our means guidelines; there can be no exceptions to verifiability; but a diversity of means of verification. Septentrionalis PMAnderson 18:04, 11 January 2007 (UTC)

The problem with the policy, as it's written, is it allows a "heckler's veto" for statements with no in-line citations in articles that satisfy WP:V with references on the bottom. All I have to do is come along to an article (such as, say, DOMA) and say "I challenge the accuracy of X, Y, and Z. Please provide inline citations per WP:ATT's policy re: challenged or likely to be challenged statements." Then, either someone has to reproduce the work of writing the article in the first place, search through the sources and add inline citations, or I come back in a week and say "I'm removing X, Y, and Z as unsourced per WP:ATT's policy regarding inline citations on challenged or likely to be challenged material!" (Note that that's not what I'm doing on DOMA-- I'm not really an asshat.) The solution, though, isn't to remove the inline-citation requirement for ocntentious or likely to be contentious content, fofr reasons User:SandyGeorgia mentioned above. Rather, it's to simply require inline citations for statements in any article. Yes, PMAnderson, sometimes citations are just fine without inline notation, and it is possible to game the inline citation format, but this still increases the verifiablilty and the ease of verifiablilty substantially, without that much more work. GertrudeTheTramp 17:36, 11 January 2007 (UTC)

That position strikes me as the heckler's veto personified. Have an entire section that comes from an article whose title makes it manifestly obvious that it is the source for the section? No matter, demand inline citations, its policy. And, no matter how many you have, someone can demand more. Its policy. Don't bother reading the references. Far better to demand form over substance. Its policy.

I agree that the present formulation is wrong, but for the opposite reason. I regret not having noticed sooner that the inline citation fetish got put in there, but I oppose making inline citations (as opposed to references) a policy requirement. Robert A.West (Talk) 17:50, 11 January 2007 (UTC)

The point is we should err on the side of easy WP:V-ifiability. There won't be a never ending demand for more citations, because once an article is fully cited, there is nothing left to demand citations *for*. Remember, policies are not iron-clad, there is still consensus-building and WP:IAR to deal with people who abuse policy to impede progress. GertrudeTheTramp 17:59, 11 January 2007 (UTC)

The problem with mandating inline citation is that it makes the job of the asshats easier. I have seen too many newbies come along and add a (often in fact true and verifiable) fact to an article, and watched the resident PoV-pusher throw it out on the basis of the "may be removed" language in WP:V. This would require the newbie not only to have a source, but to know our policy on citation, and how to format a footnote. Septentrionalis PMAnderson 18:02, 11 January 2007 (UTC)

On DOMA, I would not object to Category:citation needed tags; those would be appropriate, and indicate which points need to be fished out of the law review articles. Septentrionalis PMAnderson 18:02, 11 January 2007 (UTC)

Jimbo says unsourced material can be [shot on sight] anyway, and that doesn't just apply to BLPs. Adding this to the policy makes things easier for n00bs, because then there's an official policy they can be steered towards regarding why their statement was removed. GertrudeTheTramp 18:23, 11 January 2007 (UTC)

But, it is simply false to say that information supported by a reference is unsourced. We have {{request quote}} for times when a reference is cited, but without sufficient specificity for verification. On the other hand, how does one footnote a section that comes from a single source without looking silly? One footnote at the end of the section? Ibid after ibid?. What about a section that is an interleaving of two papers, each two or three pages in length, presenting the two sides of an issue? If the title of the section and the titles of the papers contain the word "Constitutionality," any reader of ordinary intelligence should be able to make the connection. IIRC, that is the situation with DOMA.

I think Wikipedia will be much better served by having policy require simply that the reference be easy to locate given the information supplied and having guidelines handle the specifics, rather than by overspecifying policy and relying on IAR. Robert A.West (Talk) 18:28, 11 January 2007 (UTC)

I'm a little concerned that people are interpreting the phrase "Material [...] should be accompanied by an inline citation" as requiring an inline citation. In my field, the word should is generally interpreted as offering a recomendation, but not binding. Is that not how other people interpret it? JulesH 18:58, 11 January 2007 (UTC)

Hmm... "Should" generally expresses an obligation or duty. In the context of policy, I am not sure there is going to be much of a practical difference. An editor who tries to appeal to such a difference is likely to be accused of Wikilawyering. Robert A.West (Talk) 19:49, 11 January 2007 (UTC)

Consensus?

As I read the archives, the text requiring inline citations was left despite several strong objections -- I do not see the hard-won consensus that User:SandyGeorgia is claiming. Robert A.West (Talk) 18:48, 11 January 2007 (UTC)

I think that the "challenged or likely to be challenged" version is likely the compromise position between those who would like to see every fact cited and those who would not. Since it is possible for material that was once challenged to stop being challenged once a citation is provided to the user who requested it, it is not true that a one-time challenge must result in a permanent inline citation. So really it's facts that are likely to be challenged, and direct quotes, that the current version of the proposal covers. CMummert · talk 20:20, 11 January 2007 (UTC)

Suggestion

Would people object to changing "should" to "should normally" in the section being discussed? This would leave a potential for an argument that an inline citation is not necessary in that specific place, but only where there's a specific reason for that. JulesH 21:50, 11 January 2007 (UTC)

I would prefer "should usually" as above; but "normally" is acceptable. The unconditional "should" is ambiguous, as the section above shows; and the interpretation as mandate is not consensus, and is not likely to be. I also do think that the distinction between the mandated end and the often prefered means is valuable. Septentrionalis PMAnderson 22:38, 11 January 2007 (UTC)

Ill effects of in-line citation

In addition to the potential for abuse above, the appearance of reliability that in-line citations offer is largely fraudulent:

To begin with, it is quite common for a sentence to be reversed in meaning without changing the citation. Often this is in good faith; the editor has a different version to explain, and doesn't notice or check the footnote.
Several editors use in-line citation as a means of POV mongering. They will cite a source which does exist, but is not readily available on line, and which does not in fact support the text for which it is cited. In many cases, this appears to be the result of googling a phrase which supports their preconceived ideas, and never seeing the context; in some cases, of reading through rose-colored glasses; in some cases, plain lying.

There are other abuses, but these should be enough to explain why I do not choose to support an unwise proposal. Septentrionalis PMAnderson 22:38, 11 January 2007 (UTC)

I prefer to assume that the majority of inline cites support what they purport to, and in cases of doubt it's rather easier to check pp. X–Y than to trawl through an entire book. As for the ones that don't, never attribute to malice ...

Half of the problem here is that different disciplines have different ways of doing things. Referencing that might serve adequately for an article on mathematics would be varying degrees of inadequate for a biographical or historical piece. Let's be reasonable: most articles have no references. Most are factually incorrect or incomplete. It's not as if we're on the home straight, we haven't even passed the starting line in terms of verifiability, or attributability, or whatever you want to call it. Arguing about what kind of references are required is rather missing the point. Angus McLellan (Talk) 00:22, 12 January 2007 (UTC)

"Most are factually incorrect or incomplete". How exactly has that been determined? The only study I saw compared WP with Britannica and they came out quite close. As to whether most have no references, that might be easier to verify, but again I have never seen any data about it. CMummert · talk 00:49, 12 January 2007 (UTC)

I believe inline citations are important and helpful, but we have to be fair and acknowledge that Nature's comparison was on science articles, and it is ironic that our science projects, who are probably doing our most accurate articles, are being pushed to add citations. I ask the people in those projects, who are frustrated by this, to try to empathize with those who are working on the more subjective disciplines. If there was a similar study by Mind comparing our philosophy articles with Britannica's, I would not want to know the result! Let's try to work something out without compromising the high demands for attributions that we need in less technical areas! --Merzul 01:03, 12 January 2007 (UTC)

Actually, one of the editors I am thinking of when I write of abuse of in-line citation works in political science; another in history. DOMA, which came in at the start of the discussion, is a legal article. The subjective disciplines won't like mandatory in-line citation as much as they think they will. Septentrionalis PMAnderson 04:33, 12 January 2007 (UTC)

Yes, the irony never ends. Like I said above, I think the "challenged or likely to be challenged" standard is already a compromise, especially if it is interpreted so that challenged can be resolved on talk pages. This subject came up at WP:V today, by the way.[1] CMummert · talk 01:34, 12 January 2007 (UTC)

Most articles are effectively stubs; only 34% were over 2K as of September 2006, and 20% were less than 0.5K. Stubs are almost certainly incomplete. Also, as Merzul said, the Nature survey flattered Wikipedia by choosing one of WP's strong fields. Had they done the survey on the humanities, the results would have been rather different. Then there are all those {{Catholic}} and {{1911}} type articles, which are only a small step up from unreferenced, and are very unlikely to correspond with modern scholarship. I could go on, but it's late.Angus McLellan (Talk) 01:20, 12 January 2007 (UTC)

I thought that by "incomplete" you meant "seemingly complete but leaving out important viewpoints". Of course there are a lot of short articles, but when I see a one-paragraph stub I don't expect it to have any sources - that's why it gets tagged as a stub. Most importantly, I don't see how improved sourcing can lead articles to agree with modern scholarship, or to be comprehensive. At some level we have to trust the wiki process to lead to improvement, rather than trying to dictate it through a sourcing policy. CMummert · talk 01:34, 12 January 2007 (UTC)

Quaero: Are the figures above taking into account dab pages and short lists (like USS Cole)? We have a lot of those. Also, we have a lot of bios of historical figures that will probably never be more than 2K, no matter how impeccably referenced. Robert A.West (Talk) 11:32, 12 January 2007 (UTC)

Conclusions?

Maybe we can take a time-out and just state our standpoints on the issue of in-line citations in this proposal. For clarity, we could keep this section a bulleted list and not answer each other here, then normal debate can resume below or above.

I support the original version, but I don't feel this is the most crucial aspect of this proposal, it is simply the best mechanism that is currently available and maybe some day we have things like m:wikicite. The main reason I'm interested in this proposal is because attribution is a better concept than verifiability, not because of the inline citations. In conclusion, I prefer the strict wording, but don't object to the softer proposals. --Merzul 04:39, 13 January 2007 (UTC)

I support WP:ATT because it says "attributable" at the top. It this was changed to "attributed" then I would no longer support it, because it would be much stronger than WP:V. Some common sense is necessary, rather than legalistic citation requirements. I feel that the "challenged or likely to be challenged" language is reasonable, so long as "likely to be challenged" is determined by consensus on an article-by-article basis. CMummert · talk 05:00, 13 January 2007 (UTC)

I agree with Cmummert's statement. I also assert that inline citations should only be absolutely required for direct quotations. If a statement is challenged, and a reference produced, there will rarely be a reason not to include that reference in the list for the article. On the other hand, there may be reasons not to use an inline citation in particular instances. Since this policy is non-negotiable, it should include only what it must include, and the remainder should be left to discussion and consensus. The more we rely on IAR, the greater the chance that IAR will be applied to the core purpose of the policy. Hence, this policy and the accompanying FAQ should make clear that inline citation is only required for direct quotations. Robert A.West (Talk) 17:40, 13 January 2007 (UTC)

I don't support requiring inline citations. Ideally, the phrasing should strongly suggest that they are used, but the policy should stop short of requiring them. JulesH 18:29, 13 January 2007 (UTC)

The first exception (ho-hum)

I know this page is sort of dying, but when did the first exception go from allowing questionable sources by subjects to allowing questionable sources about subjects. What a laugh! Read it through and think about it: ipso facto Wiki allows questionable sources on any article. Let's throw up the historical tag and get out of here. Marskell 21:55, 16 January 2007 (UTC)

Concur - it kinda went nowhere. SandyGeorgia (Talk) 21:59, 16 January 2007 (UTC)

The intention is clear from the phrase, "Articles about themselves." The current phrasing was an attempt to resolve a minor unclarity about whether, say, the American Red Cross's website is only permitted in an article entitled American Red Cross, or may be valuable to an article such as Hurricane Katrina that concerns its activities, and concerns matters (relief efforts) about which the Red Cross is reasonably considered authoritative. I think the unclarity is easily fixed, now that it has been pointed out. Robert A.West (Talk) 22:42, 16 January 2007 (UTC)

Mr. West's revision is not ideal, because the subsection is about questionable or self-published sources. The current version is almost OK for self-published sources, but what about questionable sources?

Even for self-published sources there is a problem. Suppose organization X allows John Doe to put whatever he wants on the organization's web site without any supervision, and John Doe puts up an page, and signs it as the author. It's effectively self-published. Under the current policy, we could use the posting for information about John Doe, but not about organization X, because organization X is not the author. --Gerry Ashton 23:07, 16 January 2007 (UTC)

My bad -- author or publisher should do it. In the given instance, the web page could validly serve as a source for any of the following in normal circumstances:

Membership numbers for X in an article about X,
The fact that Richard Roe won an award bestowed by X during John Doe's tenure as President, in an article on Richard Roe.
The birthplace of John Doe, in an article on John Doe.
The birthplace of John Doe, in an article about the birthplace.

As for questionable sources, John Doe writes an article for the High School Gazette about his tenure as the President of X. The Gazette is not reliable because of the obvious inexperience of its board, yet it could be validly used for any of the above four purposes, since John Doe is the author, each fact is relevant to John Doe, and he should be an authority on each point. What doesn't work? Robert A.West (Talk) 23:26, 16 January 2007 (UTC)

I think adding "or publisher" solves the concern I had. --Gerry Ashton 23:37, 16 January 2007 (UTC)

Thanks. I was afraid I has missed something else. Robert A.West (Talk) 23:43, 16 January 2007 (UTC)

"The intention is clear from the phrase, 'Articles about themselves.'" The phrase actually reads: "Questionable sources should not be used except in articles about themselves or their activities."
"Questionable or self-published sources [may be used] in articles related to their authors."
"In rare cases, the best source for a particular article may be one that this policy would not recommend; for example, a self-published non-professional source or one that might otherwise be regarded as "questionable."

Compared to two or three months ago: "their activities" is a weakening in first, "related to" is a massive weakening in second, and the third remains a notwithstanding clause that renders the rest of it void ("psst, just ignore this policy if you like.") What's so stupid about the third is that it repeats questionable or self-published, which have already been caveated, then throws the door open with "[if there is] shallow coverage." We have never in bold re certain disciplines, so perhaps people won't misuse it; not holding my breath. Marskell 06:12, 17 January 2007 (UTC)

This talk section started on the first exception only, so I'll confine my response to that point for now. "Or their activities" was intended to avoid stupid arguments, such as whether the Nobel Foundation's website can be cited in an article on the Nobel Prize. It is hard for me to imagine what abuses that particular "weakening" opens us up to, provided we retain the "unduly self-serving" caveat. Can you be more specific about how you think it is harmful? Robert A.West (Talk) 06:38, 17 January 2007 (UTC)

The "activities" of a crackpot physicist include physics, correct? "Related to" is even worse because it doesn't even suggest active involvement.

As for the Nobel Foundation or the American Red Cross, I don't follow—neither of those would normally be considered questionable and they seem a red herring here. Marskell 07:34, 17 January 2007 (UTC)

(unindent) Responding to Marskell, if someone tried to justify an inappropriate source on the basis that the article is about the activites of a self-published author, or related to a self-published author, I would point out that these are merely brief summaries of, or headings for, the paragraph under the heading "Questionable or self-published sources in articles related to their authors" and a summary or heading should not be used to override the full text. --Gerry Ashton 08:04, 17 January 2007 (UTC)

Also replying to Marskell: Self-published material must be in a field in which the author is considered authoritative" and "should not be contentious or unduly self-serving" if it is to be included. Moreover, the material must statisfy other criteria for inclusion, such as NPOV, which includes not giving undue weight to fringe theories. On the other hand, if the fringe theory is notable enough to mention, it may be worthwhile to cite the author's self-published work when describing it. For example, tax protesters are legal crackpots: not one court case, not one law review article, not a single widely-recognized authority supports any of their contentions. Yet, we validly include the theories in articles such as Sixteenth Amendment to the United States Constitution, because they are notable and relevant, and readers may come to the article having heard of the theories. Having determined that the theories are valid for inclusion, the self-published works of questionable legal reasoning are clearly authoritative on the content of the theories themselves and are properly cited.

As for the Red Cross/Nobel issue, the self-published works of prominent organizations are still self-published. I'm not sure what you don't follow. I would have to search the archives to see what the actual edit wars were that inspired the change, but they weren't much less silly than the examples. Robert A.West (Talk) 10:16, 17 January 2007 (UTC)

I don't agree with Robert A West's assertion that works of prominent organizations are necessarily self-published. If the organization has established an effective system of editorial control, they are no more self-published than unsigned stories in the Washington Post. Some examples would be the Bell Labs Technical Journal or the IBM Journal of Research and Development. --Gerry Ashton 17:33, 17 January 2007 (UTC)

I was not speaking about journals and the like: Obviously, every reliable source is published by some organization. I was speaking about the material an organization puts out about itself. For example, the page soliciting advertising on the Washington Post's website is a self-published work, not a part of its editorially-controlled content. While I would accept such a page for basic demographic information about the Post's subscribers, I would not accept it as a source for the fact that readers of the Washington Post are more likely to be in an influential position than readers of papers in general. The fact that this is a large and respected organization does not make material it publishes about itself into a reliable source. Robert A.West (Talk) 19:54, 17 January 2007 (UTC)

We define it this way: "A self-published source is material that has been published by the author, or whose publisher is a vanity press, a web-hosting service, or other organization that provides little or no editorial oversight." So no, I don't follow—I wouldn't call the Red Cross or the Nobel Committee self-published under these criteria and it seems to me you're conflating "reliable sources referencing themselves" with "self-published sources"—these aren't the same thing. Given that "little or no editorial oversight" is the basis, I don't see why the long-standing "Self-published sources in articles about themselves" needs to be muddied by "activities" and "related to"; it will just invite loose interpretations. Perhaps we should make note of the reliable self-referencing thing (Brittanica's claims to be the largest English language encyclopedia always comes to my mind...) but it differs from the questionable source issue.

To go to your earlier point, I can't agree or disagree that tax protesters cited on the pages should be "considered authoritative" and are "not contentious or unduly self-serving", as I know nothing about the issue. But let me compartmentalize it: Is the idea fringe? Does the place of publication provide little or no editorial oversight? If the answer is yes to both, the cites should not be used on the Sixteenth Amendment. They might belong on the page of a given tax protester or organization that's notable enough to belong, but nowhere else. That's what the long-standing wording from RS is there for. Marskell 20:41, 17 January 2007 (UTC)

Well, we have a point of strong disagreement. If crackpot theories about X have significant cultural influence, then they are perfectly proper to discuss (including the refutation of those theories) in an article on X. It is not the job of Wikipedia to ignore that influence in the hope it will go away. An article on the assassination of JFK would be incomplete without a discussion of the various crackpot conspiracy theories, precisely because they have enormous influence. It is likely that more people have heard the misinformation about the 16th Amendment that tax protesters peddle than know the actual text and history of the Amendment. Hundreds of people go to jail and thousands pay large fines because of such beliefs. Not dealing with such a notable cultural aspect of that topic in the main article would be irresponsible and a disservice to Wikipedia's readers. Robert A.West (Talk) 00:32, 18 January 2007 (UTC)

Absolutely we can discuss crackpot theories—but not with questionable sources. That's why I asked "yes to both questions". As our various sourcing policies and guidelines have long pointed out, if an idea is sufficiently notable you ought to be able to find reliable sources for it (if hundreds are jailed in your example, for instance, surely a few newspaper cites can be found). The only place crackpot ideas belong with questionable sources is on the page directly discussing the source of the crackpotishness: "Joe Q. is a well-known lunatic; his bizarre ideas are detailed on his website www.unreliableinfo.com". The website should be linked nowhere but the article for him (assuming he's sufficiently notorious to demand an article). Marskell 09:04, 18 January 2007 (UTC)

Absolutely, there are plenty of reliable secondary sources: Supreme Court cases, Appellate Court cases, articles on the law, articles on the psychology of the movement, newspaper articles, and those sources provide the interepretive basis for the relevant articles. Most of those sources in turn cite, or even quote from, the sources propounding the crackpot theory. This is just the garden variety primary/secondary source distinction. When reliable sources are making references to and summarizing an unreliable source, why should it be treated any differently from any other primary source? We can quote from primary sources, we just don't rely on them for interpretations. The distinction between an original copy of a partisan screed of 1800 and a modern reprint without commentary is availability, and perhaps some cleanup of orthography, not reliability of the underlying text. If in 50 years a scholarly reprint, without commentary other than a short preface, is issued of The Law that Never Was, the rules as you apply them would permit the 2057 Wikipedia to include the very same source that you are arguing we should now exclude. That strikes me as following the rules out the window. Robert A.West (Talk) 18:05, 18 January 2007 (UTC)

Sorry for two days late.

"There are plenty of reliable secondary sources:..." Then what the hell are we talking about? I was using your idea as a hypothetical. If there are "plenty of reliable secondary sources" re your example, then what we're debating re this (potential) policy isn't pertinent. If "reliable sources are making references to and summarizing an unreliable source", of course you can cite the reliable source and supplement it with the primary source (if you can find it). Again, I think you're conflating issues. But on the bright side, I doubt we actually disagree that much. Marskell 22:36, 20 January 2007 (UTC)

I try not to discuss total hypotheticals -- I may construct examples, but they always relate to real cases. The type of edit that your say, "Of course you can," is actually prohibited by the current wording in WP:V.

"Sources of dubious reliability should only be used in articles about the author(s)."
"Self-published books, personal websites, and blogs are largely not acceptable as sources."
"Material from self-published sources, and published sources of dubious reliability, may be used as sources in articles about the author(s) of the material."

I will grant that most Wikipedians would interpret things the way you did and I do, but I thought one of the points of this proposal was to minimize the needed for that sort of analysis and to say what we mean. As for conflating, I disagree that I have conflated in either instance. Material written by a person in consequence of their position within an organization and reflecting the official position of the organization is, effectively, written by the organization. If we argue otherwise, material written by me and published by the corporation that I own is not self-published. That is absurd. Robert A.West (Talk) 00:06, 21 January 2007 (UTC)

On your second last I thought with "reliable sources are making references to and summarizing an unreliable source", you were suggesting something like this: "The Times has reported that tax protester John Q. keeps a record of supposed government tax fraud on his website."[1] Here, with the Times ref in place, I would find it acceptable to link to the dubious site outside of the article about the author; I don't think policy disallows it because you're not actually sourcing any info to it. But again, I didn't think we were debating this exact issue, but rather the dubious source supporting info by itself outside of the article about the author (which I'm concerned this page will make more likely, injunctions to the contrary aside)...

On the other issue, you and your corporation would be subject to a host of guidelines before being considered at all (WP:CORP, WP:SPAM). I was thinking of something like this: a publication that a reliable source puts out regarding activities related to its work. I would not call it "self-published"; if you do, we can agree to disagree, I suppose.

Now, here's an example for you. Michael Salla, a once respectable expert on international relations, decided a few years ago that Earth governments are in contact with eight (or perhaps eighteen, can't remember) alien races, and that a massive conspiracy is hiding it from everyone. I took his (currently red-linked) page to AfD; it was deleted initially, and subsequently re-deleted once or twice on the basis of WP:CSD, G4. Unfortunately, Salla is still linked on Exopolitics, along with Alfred Webre, another apparent crackpot. I've gutted the page in the past, but the shit keeps turning up. Now, bear in mind that "exopolitics" relates to his "activities"; according to partisans who edit the pages, Salla and Webre are "pioneers" in the field. I'm not saying that you would interpret this potential policy to allow links to his websites off of the page about him, but the wording as it stands can be construed to do so.

Do you know what I mean? I don't like it one bit. My weird example probably has thousands of counterparts across the site. But apparently we should weaken the wording because people might interpret the Red Cross as an unreliable source? Marskell 22:35, 22 January 2007 (UTC)

(unindent) If we are going to have articles on crackpot theories, then we are going to end up quoting and paraphrasing crackpots. I just don't see any way around it. Whether we should have such articles at all is (IMO) a matter of NPOV and undue weight, moreso than attribution. Robert A.West (Talk) 05:29, 23 January 2007 (UTC)

So we shouldn't have articles about things that happened just because the people involved were lying or stupid? — Omegatron 05:03, 25 January 2007 (UTC)

I don't think I said that. I don't think I said anything remotely resembling that. What, particularly, is your concern? Robert A.West (Talk) 17:23, 25 January 2007 (UTC)

Is it soup yet?

What needs to be done to present this proposal to the community? Are there remaining issues or are we all just waiting for someone else to do something? Robert A.West (Talk) 00:54, 17 January 2007 (UTC)

One useful thing before presenting this would probably be a short list of "diff"s between this policy and the official ones. I mean essentially this is just a simple reformulation, but I'm sure you have made some subtle differences in the details, what are they? --Merzul 01:51, 17 January 2007 (UTC)

oh, and the top of this talk page should have a nice infobox explaining very briefly the merits of the proposal! --Merzul 01:52, 17 January 2007 (UTC)

I think the only real outstanding point of serious contention is the third exception. I personally believe that almost all wikipedians would be happy to accept this exception, but a few have argued strongly against it and if anything can be done to address their concerns without reducing its value, it clearly should be. JulesH 08:30, 17 January 2007 (UTC)

A few have argued against it, yes. BrianSmithson, Taxman, Jayjg, JzG, Yomanangi, even Slim, who was largely agreeing for the sake of compromise, but didn't like it. In fact, I'd suggest "almost all Wikipedians" requires a touch of selective memory, given the arguments three months ago. A raw majority of all Wikipedians would quite probably agree—a majority of Wikipedians (in the truly broad sense) would like nothing more than to see the policy standards weakened. Thankfully, Wikipedia is not a democracy.

The real rub is the three months. I think this needed to go live sometime in Nov., for it to have succeeded. It's still on watchlists obviously, but there's not enough people here—and if more people were here, it would just be more arguing. Marskell 21:22, 17 January 2007 (UTC)

Obviously, it would be better to dump the third exception, and whatever else controversial, and postpone such discussions to after this policy is official. Whoever has written this proposal has done an excellent job simplifying 3 policies into one and FAQ, and it would be a shame to throw it all away. --Merzul 16:11, 18 January 2007 (UTC)

"Whoever has written this proposal has done an excellent job simplifying 3 policies into one and FAQ." Yes, I agree. It was initially pushed by SlimVirgin. I'm not sure how much she's up for editing it now, but you're right that the basic idea of streamlining everything was a good one. Marskell 23:07, 20 January 2007 (UTC)

"A raw majority of all Wikipedians would quite probably agree—a majority of Wikipedians (in the truly broad sense) would like nothing more than to see the policy standards weakened."

The general argument is that this codifies existing practice, and doesn't weaken any actual standards, even though it weakens the wording. Ken Arromdee 15:04, 22 January 2007 (UTC)

That's a highly controversial general argument. Wikipedia has so much material that I would rather see a stress on quality instead of making it even easier to add material. I mean if the wording should be changed at all, the "threshold for inclusion" should be higher not lower than before. --Merzul 15:44, 22 January 2007 (UTC)

"The general argument is that this codifies existing practice." Precisely the problem. See, Wikipedia:Danny's contest, which is only a few months old: "Some numbers that were run today show that we have over 230,000 without any sources whatsoever. That's almost 20 percent of our total articles without any sources."

Do you want to "codify" this "existing practice"?—articles with no sources whatsoever? The descriptivist path has been beat before on this talk page, and it is the weakest "pro-" argument, IMO. That people do something is NOT a reason to give them licence to do it (on Wiki or anywhere else). "Doesn't weaken any actual standards..." Wrong, wrong, wrong. It codifies a weakening of "actual standards."

Now is where you tell me that various caveats on the page will control this concern ("In rare cases" is a beauty). But, no. The "average" Wikipedian will use the three exceptions to add bullshit whereever they can. Marskell 22:07, 22 January 2007 (UTC)

I mostly agree, but I looked through the archives and this goes back to the prescriptive/descriptive debate, and here SlimVirgin seemed to advocate a compromise that the policy should describe what the best editors are already doing, and prescribe this to the rest of us. Now... if there are cases where otherwise good editors have to break policy to do their job, then either we have to rely on IAR or make sure the policy covers those cases. Having said that, I was not convinced by reading the archives that the third exception is needed. One option would be to move this exception to the ATTFAQ, so that at least it wouldn't be as prominent as the other two exceptions.

Finally, in an answer to your general concern, I would like to point out that not all changes have been going in that direction, e.g., the second exception has been significantly tightened by disallowing self-published material to contradict peer-reviewed articles. I think this is a bigger blow to crackpot articles than what they stand to gain from abuse of the "related to" wording. However, if reviving this proposal would require reverting to somewhere mid-october, then we should do so. Actually, I don't even know if this proposal is dead or still alive. Is this alive? --Merzul 02:24, 23 January 2007 (UTC)