Wikipedia talk:Verifiability

Ease of access

Suppose someone cites several fictional classified papers. Access to sources states that "The principle of verifiability implies nothing about ease of access to sources". I interpret the rules quite literally. Do I have to break into the Pentagon to prove that it is a hoax? Marcus Qwertyus 21:38, 16 December 2010 (UTC)

That sentenced is prefaced by the comment: "anyone should be able to check that material". "Classified" means that not anyone can access it, so a classified document would not be a reliable/verifiable source until someone makes it publicly available. Location (talk) 21:47, 16 December 2010 (UTC)

Clarified. Marcus Qwertyus 21:53, 16 December 2010 (UTC)

No, I think the edit you made makes the statement more ambiguous, so restored the early information. It is the source of the information we need access to, yes? --Nuujinn (talk)

All I did was reassert that the source must be published. It's contradictory: "anyone should be able to check that material "≠ "implies nothing about ease of access to sources". Marcus Qwertyus 23:04, 16 December 2010 (UTC)

With respect, no, what you wrote was "The principle of verifiability implies nothing about ease of access to information published by a reliable source". My point is that it is the source we must have access to, not the information published by the source. The information published by the source may be available in many other places; wikipedia, blogs, etc., which are not reliable. To verify the information, we must have access to the source. --Nuujinn (talk) 00:08, 17 December 2010 (UTC)

Okay then. How about "The principle of verifiability implies nothing about ease of access if it has (or "as long as it has") been published by a reliable source". Marcus Qwertyus 00:30, 17 December 2010 (UTC)

Perhaps you could elaborate a bit more on what's wrong with the current wording, I'm not sure I understand what you're seeing as problematic. Sorry to be dense. --Nuujinn (talk) 00:35, 17 December 2010 (UTC)

My point wasn't to add the bit about the reliability of the source, just that it has to be published and therefore accessible. Marcus Qwertyus 01:10, 17 December 2010 (UTC)

Yes, I get that, but what is wrong with "The principle of verifiability implies nothing about ease of access to sources"? --Nuujinn (talk) 01:17, 18 December 2010 (UTC)

I agree with Nuujinn. It should be obvious that a source thats a classified document can't be used to support something in an article. Thats covered by their not being listed in WP:RS, where it says "reliable published" sources. AaronY (talk) 01:57, 18 December 2010 (UTC)

I think AaronY has a point since "publish", "publication", and "public" all have a common Latin root in the word publicus (“pertaining to the people”).  RB  66.217.118.112 (talk) 22:17, 19 December 2010 (UTC)

Copyright

I've reverted an addition which I feel is already covered under our copyvio guidelines; its pointless imho to cover guidelines in more than one place and given how daunting the scope of our entire MoS is (probably 200+ pages if printed out), I don't feel there is a need to over-explain things or lengthen rather than tighten guidelines in many areas. Lets just try to stay on topic here. AaronY (talk) 20:29, 18 December 2010 (UTC)

• It is already covered under WP:COPYVIO and that's where all detailed discussion of the issue should stay. All this page needs is a two-sentence mention of the fact that you can follow the sources too closely, plus a direct pointer to the policy. The background and context to this edit, of course, isn't completely unrelated to the circumstances in which User:Rlevse became a redlink: WP:V has genuinely been used as a reason for substantial copyright violations in the encyclopaedia. The problem the edit was supposed to address is not just theoretical.—S Marshall T/C 20:49, 18 December 2010 (UTC)

I'm aware of the Rlevse fiasco. I still don't think we need to summarize one guideline in another. There's always a tendency to over-explain in these guidelines which has the opposite effect of the one desired. Once they become too long or digress too much people's attentions begin to wander. I'd be interested to see what others think. AaronY (talk) 22:32, 18 December 2010 (UTC)

This could obviously go either way. Brevity v Spell-it-out. Suffice to say, the tension between V and COPYVIO is a real one, and a brief pointer for resolving that tension is not a bad idea. Ocaasi (talk) 23:39, 18 December 2010 (UTC)

Is 2 to 1 consensus now? Oh well I guess if someone else comments we can change it back. AaronY (talk) 15:41, 19 December 2010 (UTC)

I think all we need is a pointer, with no explanation. Explaining how to avoid copyvio here just increases the chance of confusion or conflicts between here and elsewhere. --Nuujinn (talk) 16:07, 19 December 2010 (UTC)

Pointers can do the trick in such cases, but they should be clear.--Andrew Lancaster (talk) 16:56, 19 December 2010 (UTC)

Close paraphrasing

Hi S Marshall, I removed "'Too closely' includes any sentence-by-sentence paraphrase of a source." It's not clear what that means, and I don't think we should get into it here. It's probably more appropriate just to link to Wikipedia:Close paraphrasing. SlimVirgin ^{talk|contribs} 15:00, 20 December 2010 (UTC)

• I'm having trouble seeing any clearer way I could put it. Basically, we don't want editors paraphrasing the sources at sentence- or phrase-level. We want editors reading the sources, understanding them, and then summarising what the sources say, but using entirely original language in the process. That's more than just varying the wording the sources use, and it's more than just putting the same information in a different order; it's making the point the sources make, but using completely fresh language to do it. I really do think WP:V needs to say more emphatically that it isn't an excuse for copyright violations.—S Marshall T/C 15:29, 20 December 2010 (UTC)
  • At the same time editors should be cautions when naming items or concepts, especially where sources that are not renouned for their accuracy are involved. For example "Atomic weight" and "atomic mass" are not the same, neither are the "Republic of Ireland" and the "Irish Republic". Martinvl (talk) 15:34, 20 December 2010 (UTC)

• We can't always use original language, though. It depends on the context, and when close becomes too close is a matter of judgment and sometimes negotiation. SlimVirgin ^{talk|contribs} 15:37, 20 December 2010 (UTC)

• In cases where we can't use original language there are various recourses: direct quotations, intext attribution, etc. within the overall limits of fair use. I'm right with you that editorial judgment is the king here. The point of a policy, though, is to explain the things that should rightly influence editorial judgment, and one of them is copyright.

  The reason I'm so adamant on being clear about copyright within the text of this policy is because WP:V is a big stick for dealing with problem articles and problem editors. That's a good thing, and it's important, but it also means that in contentious discussions WP:V can be deeply interrogated. It's also deeply interrogated at FAC and to a lesser extent GAN. But you can go too far in verifiability, because the ultimately verifiable article references a source for each statement it makes, and gives correct proportionate weight to each point that the source does—but the easiest way to achieve that is a paraphrase, and paraphrase can sail close to, or actually infringe, copyright boundaries. And that's why we have copyvios in both contentious articles and (as the Rlevse fiasco demonstrates) featured articles: because the editors involved are parsing WP:V without paying any attention to copyvios at all. Real and recent issues show that it's time we were clearer about copyright in WP:V.—S Marshall T/C 16:52, 20 December 2010 (UTC)

There was a discussion started about this above under the copyright heading. AaronY (talk) 17:02, 20 December 2010 (UTC)

Am I to take silence as consent for this edit, then?—S Marshall T/C 23:20, 21 December 2010 (UTC)

Well I disagree with it as do others I think. AaronY (talk) 16:52, 22 December 2010 (UTC)

Purely on the basis of WP:CREEP, as I understand it?—S Marshall T/C 17:07, 22 December 2010 (UTC)

I don't read essays. I'm voting this way because most of the analysis I've seen of Wikipedia by independent sources says one of the reasons our growth has slowed is because new editors are bitten frequently and referred to a huge unwieldy manual to learn some mistake they have inadvertently made. I favor anything that cuts down the over-explaining of the MoS while keeping its core messages intact. Ideally the whole thing should be a fraction of the size it is now. A lot of its length is due to needless repetition (the case here imho), and over-explanation. AaronY (talk) 19:06, 22 December 2010 (UTC)

It's certainly true that our growth has slowed, new editors are bitten frequently, and the rules are huge and unwieldy and not very clear. Right with you there. Unfortunately, this is what happens when policies are written by a committee. Particularly when the committee is self-selecting, and therefore consists of the people who would rather spend their time telling other people how to write an encyclopaedia than writing it themselves. (Harsh words but it includes me too.)

My problem with the current version of WP:V is that for some reason, editors including very senior editors have taken or are still taking it as a reason to breach copyright. Which is exactly ass backwards, because while it's important to have an accurate encyclopaedia, obeying the law is even more important. In fact, if you'd like to trim WP:V then I'd encourage you. I mean, the first, second, third, fifth, sixth, seventh (and then I stopped counting) paragraphs all contain the same basic thought, with only slight variations in the order in which they say "reliable", "published" and "source". Surely some of this repetition could be trimmed.

My point is that the policy that talks in so much detail about how to analyse and cite sources needs to be clearer about how far the sources can be used, and there's plenty of other spare wording that can be trimmed to make room for that.—S Marshall T/C 20:23, 22 December 2010 (UTC)

Regarding the point about "not reading essays", Wikipedia:Close paraphrasing is an essay, and referral to this essay was what started this section.  FYI, RB  66.217.118.46 (talk) 20:46, 22 December 2010 (UTC)

S Marshall, sorry for not responding; I didn't notice until now that the discussion had continued. The copvios we've had have involved extensive close paraphrasing, not just the occasional too-close sentence; and in several cases straightforward plagiarism. The issue of the occasional close paraphrase isn't something we can deal with here, because it really does depend entirely on context and it's complicated. For example, sometimes close paraphrasing is necessary; sometimes it will involve common words and phrases and therefore won't be a copyvio; mostly intext attribution will be a good thing, but sometimes it's misleading. I agree that discussion about this is good, but I don't think this policy is the place for it. SlimVirgin ^{talk|contribs} 06:35, 23 December 2010 (UTC)

• • Sure, copyright is complicated and we don't need to deal with it here. Detailed discussion belongs at WP:CV. But WP:V as presently phrased is misleading (or at least, has been extensively misunderstood, including by formerly-senior editors) because it would be possible to comply with it by close paraphrasing. WP:V describes Wikipedia's "three content policies" as WP:V, WP:NOR and WP:NPOV and quite falsely suggests that this is an exhaustive list, in a definite sin of omission against WP:CV. Until I started this argument by amending the policy, its only mention of WP:CV was somewhere under "non-English sources". This policy needs to adapt to the problems it's created.—S Marshall T/C 08:08, 23 December 2010 (UTC)

"This policy needs to adapt to the problems it's created" Can you provide some examples of problems created by the wording of the policy? It seems to me that the copyvio policy is clear and not in conflict with anything here. Policies do not need to be part of core content policies to be applicable and effective. --Nuujinn (talk) 12:35, 23 December 2010 (UTC)

"Threshold" again

Threshold is a word of many meanings, but the nearest relevant one, following wiktionary, is

• (engineering) The quantitative point at which an action is triggered, especially a lower limit

To say that verifiability is the threshold for inclusion therefore implies that it triggers inclusion, and that everything verifiable is included. This would be a ridiculous proposition, never followed in practice, so it's a bit silly to state it thus. SamuelTheGhost (talk) 13:07, 19 December 2010 (UTC)

Well, from an engineering standpoint, your deduction is not 100% correct. A trigger can be what normally causes the event to happen, but still subject to other criteria. Trigger/gun is a great analogy...pulling the trigger normally fires it, but only if the safety is flipped off. But I agree that the term is ambiguous in an area where it is important that it be un-ambiguous. Also, the inclusion of an example ("truth") in that sentence of something that does not trump the requirement is also confusing and problematic. North8000 13:54, 19 December 2010 (UTC)

[inserted later] The definition for "threshold", that SamuelTheGhost gave in the first post, is definition 4 from wiktionary.org.  From the viewpoint of definition 4, crossing a "threshold" triggers an action.  Conversely, if an action was not triggered, then the threshold was not crossed.  With this reading, if pulling a trigger does not fire the gun, then pulling the trigger is not a "threshold".  So I think that SamuelTheGhost was precise, but more to the point, I think that we should agree that definition 4 from Wiktionary is not the intent of the first sentence of WP:V.  RB  66.217.118.46 (talk) 19:58, 22 December 2010 (UTC)

[insert ends here]

Can you clarify what is ambiguous in the current version? --Nuujinn (talk) 14:02, 19 December 2010 (UTC)

Verifiability is the first of several thresholds for inclusion. But it is not the only threshold for inclusion. The fact that something is verifiable means that it passes the first test and may potentially be included (it also has to pass the other thresholds). However, because Verifiability is the first threshold, it does mean that anything that is not verifiable may be excluded. Blueboar (talk) 14:59, 19 December 2010 (UTC)

Its never implied (to me at least) that its the only threshold. AaronY (talk) 15:41, 19 December 2010 (UTC)

I'll use terms from normal logic. Verifiability, the way policy is worded, is necessary but not sufficient. Material also has to be notable, and there are also some special rules for special cases.--Andrew Lancaster (talk) 16:55, 19 December 2010 (UTC)

I think the best other requirement to consider as proof that not all verifiable material should be included is that a consensus must be reached among the article editors that material is interesting enough to include in an encyclopedia. Notability applies to an article as a whole, but the interesting criteria applies to each claim within an article. Jc3s5h (talk) 17:09, 19 December 2010 (UTC)

Bearing in mind that this is the change I wanted to introduce, it seems to me that everybody agrees that it says the right thing, some people agree that it's an improvement, and some people think that it's unnecessary because the existing text already says that, but nobody thinks that what I've proposed is actually wrong. SamuelTheGhost (talk) 17:20, 19 December 2010 (UTC)

Thanks for the patronization Alex. I guess what I'm saying is that its obvious to me if Bill Clinton attends a softball game in Peoria and it makes the local paper it doesn't merit inclusion in his article, even though technically this guideline says that it does. I don't care if the wording is changed, but I've never for a second thought it meant anything with a source gets in. Its common sense that it wouldn't. AaronY (talk) 17:47, 19 December 2010 (UTC)

This guideline sets the minimum for inclusion while WP:NPOV and WP:NOTABILTY are for weight and relevance. Just because we don't summarize those policies here (which we could) doesn't suggest anyone take the policy in isolation (in fact, that's explicitly cautioned against). Ocaasi (talk) 19:10, 19 December 2010 (UTC)

I don't see a need to change the language... The intent of the statement is to outline what may and may not be added. If no reliable sources verify Clinton's attendance at the softball game, we may not mention it (even if his attendance is "true"). If reliable sources verify his attendance then we may mention it... but that permission does not mean we must mention it. I think this is clear in the current language. Blueboar (talk) 18:37, 19 December 2010 (UTC)

The previous time these issues came up we ran into a language problem with both the word "notability" and the relation of WP:N to content policy.  As noted by Wtmitchell [here], WP:N is not part of content policy.  WP:NNC states, "The question of content coverage within a given page is governed by the principle of due weight and other content policies."  At the same time, the word "notability" is frequently used on this page as it relates to content.  It seems that the word used in the WP:WEIGHT policy is "prominence".  Since I didn't like the word "prominence" in this context, I started referring to "notability/prominence".  FYI, RB  66.217.118.112 (talk) 23:27, 19 December 2010 (UTC) [inserted later] See also: Wikipedia:Notability vs. prominence.  RB  66.217.117.192 (talk) 15:16, 23 December 2010 (UTC) [insert ends here]

Here is another definition of "threshold", the point being that a threshold isolates two states:

www.merriam-webster.com threshold 3b : a level, point, or value above which something is true or will take place and below which it is not or will not.

A threshold for inclusion is therefore also a threshold for the absence of inclusion.  This is technically an ambiguity, where "ambiguity" is defined in www.merriam-webster.com as ambiguity 1b : "A word or expression that can be understood in two or more possible ways".

The problem for policy arises in considering verifiable material for the "absence of inclusion".  WP:V says that editors cannot consider material that "could be not true" for the "absence of inclusion".  RB  66.217.118.112 (talk) 02:15, 20 December 2010 (UTC)

There's also WP:Editorial discretion. Editors are expected, even required, to use their best judgment. WhatamIdoing (talk) 05:45, 20 December 2010 (UTC)

The "threshold for inclusion" implies a necessary, not a sufficient, condition. If the threshold for your inclusion at a party is that you bring the hostess some wine, it doesn't mean you'll be let in if you arrive with the wine, but also drunk and covered in mud. But there are times when the existence of a reliable source would be a sufficient condition too: if the article is underdeveloped, for example, or if the point is one required for NPOV.

We should be careful not to add anything to the policy that editors could use to reject reliable sources, because everything depends on context. What the policy currently implies is that if you arrive at an article with a good source, there has to be a strong editorial reason to keep your material out. SlimVirgin ^{talk|contribs} 06:26, 20 December 2010 (UTC)

If a reliable source has printed a mistake, WP:V policy should not require editors to ignore the error.  Under WP:V, editors cannot consider a retraction by the newspaper for the story about the Clinton softball game, because WP:V is "whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true."  WP:V is currently a force to include mistakes in the encyclopedia, and this force has no documented purpose.

From WP:V:

• See also, argument from authority.
• Notes, 4. "Wales, Jimmy (16 May 2006). "Zero information is preferred to misleading or false information...I can NOT emphasize this enough."
  

RB  66.217.117.29 (talk) 22:29, 20 December 2010 (UTC)

*WhatamIdoing, Wikipedia:Editorial discretion has the status of an "essay", it says an essay may be a "minority viewpoint", and "Consider these views with discretion."  Discretion in this case can include citing WP:V the way it is currently written, which is that it is incorrect to challenge verifiable fallacies; since Wikipedia is about verifiability, not the truth of the verifiable statement.  RB  66.217.118.27 (talk) 19:29, 21 December 2010 (UTC)

With all of it's wording shortcomings, I think that it's clear that meeting wp:ver is a requirement for inclusion, not something that mandates inclusion.

It easy and common for a wp:"RS" to be unreliable, (wp:rs has no criteria for knowledge of the subject or objectivity) and to have information that is patently wrong, or just one of many conflicting opinions. More commonly, people misuse what's in a RS. For example, The New Your Times covers John Smith's claim that the earth is flat. Then they use the NYT as a source for the statement that the earth is flat. Either way, nothing in wp:ver mandates inclusion of anything. North8000 21:00, 21 December 2010 (UTC)

If the consensus of editors is to omit something... just omit it. The "Threshold" clause allows for this... and if that is not enough, you can always invoke WP:IAR. If, on the other hand, the editors are debating whether to include something, then we would need to know what the something is before we can address the issue further. The simple fact is, we do not require that everything verifiable be added to an article... but we do require that anything added be verifiable. Blueboar (talk) 22:54, 21 December 2010 (UTC)

I agree that WP:V is "not something that mandates inclusion", that "we do not require that everything verifiable be added", and "we do require that anything added be verifiable.

We've seen that the word "threshold" is itself ambiguous, as two definitions have been given and are in use.  We've seen that there is a connotation of "threshold" that is the "threshold for the absence of inclusion".  No one has offered a purpose for having a "threshold for the absence of inclusion".  I agree with what Blueboar said, "The intent of the statement is to outline what may and may not be added."  I submit that this proposal restores that intent, and that in the absence of further analysis, consensus exists to restore that intent.  RB  66.217.118.91 (talk) 01:43, 22 December 2010 (UTC)

I don't think it is needed. The intent is already clear in the old language. Blueboar (talk) 02:15, 22 December 2010 (UTC)

BTW, before I said I thought there was a consensus, I concluded you were using SamuelTheGhost's definition of "threshold", and that my response would clarify that the word "threshold" is confounded.  The question becomes, even if you don't see the need, and think the intent is clear in the old language, is that an objection to a change which to you is not a change?  RB  66.217.118.91 (talk) 02:42, 22 December 2010 (UTC)

Blueboar, I'll take your silence as a "rhetorical answer".  I'll also withdraw my thought that you were using SamuelTheGhost's definition, i.e., Wiktionary definition 4, of "threshold".  At this point, I don't know what definition of "threshold" you meant when you said in your recent post, "The "Threshold" clause allows for this..."  RB  66.217.118.46 (talk) 20:07, 22 December 2010 (UTC)

In summary, I think that "threshold" is ambiguous, and, per it's common use, would lead towards the wrong impression. I lost track of what the proposal is, but "requirement" or "condition" would be much better (and less ambiguous) words. And, including an example of one thing ("truth") that does not override the condition, and a poor choice of a word is problematic on two levels. One, it confuses the main/core statement of wp:ver, and secondly it leads to wide mis-quoting that wp:ver states that accuracy (in cases where objective accuracy exists) is not an objective of Wikipedia. So I thin the "not truth" should also go. North8000 14:59, 23 December 2010 (UTC)

I agree with most editors here that the purpose of the "threshold" language is to say that it's not enough for something to be true, it must also be verifiable -- and even that may not be enough.

The main problem is that the principle, especially in the abbreviated form in which it is usually quoted ("verifiability not truth"), can be misunderstood as follows: We don't care whether something is true or not. We just check whether it's verifiable, and that's it.

This misunderstood version is actually convenient when dealing with fringers who claim that established and verifiable facts are false. (E.g.: "It is not true that Obama was born in Hawaii, so we can't say it." Response: We don't care whether it's true or not, only whether it's verifiable.) As a result, we have a significant number of experienced editors, including at least one high-profile admin, who believe this is the correct interpretation.

In the past we have had a number of situations in which there was a strong consensus among Wikipedia editors that all reliable sources that reported something were in fact wrong. These situations tend to attract editors who then claim that we are obliged to parrot the incorrect reporting as if it was true, because truth simply does not matter. This is particularly egregious in BLP cases and in cases such as the Sam Blacketer controversy article, which are about events that happened at Wikipedia. In these cases we are in the best situation to report the correct facts, and readers including journalists turn to us to learn about them, not for our parrotting of news sources without any warning that we know it's all misinformation.

For this reason I think the "threshold" wording needs tuning. However, the proposed change is certainly not optimal for this purpose. Hans Adler 15:33, 23 December 2010 (UTC)