Wikipedia talk:Verifiability

Then is a first change just to point to relevant policies?

What has emerged is that there is no agreement about the intended purpose of the two-word phrase "not truth" even by those who want to retain it in the policy.  This means that, and continuing to accept for the moment that we aren't changing or removing it, that we are currently unable to document what it means.  Several of us agree that "truth" and Truth^TM need not be discussed in this policy, and as per a discussion above, there is consensus that verifiable material may or may not be accurate.  There has been little or no opposition to providing references to related policies in the lede.  Thus I propose:

“ The threshold for inclusion in Wikipedia is verifiability, not truth—whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true. Verifiable material may or may not be accurate—policies and guidelines that address accuracy in the encyclopedia include Wikipedia:Neutral point of view, WP:RS#Reliability in specific contexts, and WP:Editing policy. ”

  Unscintillating (talk) 19:18, 4 August 2011 (UTC)

The way I see it, what has emerged is that there is no consensus to change the current wording. Yet a small group of dissenters are trying to get their way by engaging in siege warfare to wear down and exhaust the patience of everyone else for whom this is not the single most important current issue on wikipedia. Leave it alone for now, and let everyone take a break. Quale (talk) 22:02, 4 August 2011 (UTC)

At this point, I pretty much agree with Quale. I kind of regret having gotten into this discussion at all. It seems to me that all of the endless talk is going nowhere very slowly. The existing policy works well. --Tryptofish (talk) 22:13, 4 August 2011 (UTC)

I am of the same mind, a break of 3-4 months is what I recommend. --Nuujinn (talk) 22:24, 4 August 2011 (UTC)

(edit conflict)x2 The way I see it is that there's fundamentally no consensus to keep the current wording, and the pro-change party are not "a small group of dissenters", but either a significant minority or actually a small majority. On Wikipedia, where there's no consensus about something, the status quo remains, which enables the tactical filibustering that we've seen on this page for the last several months. Other, similar discussions regularly appear, although not necessarily on this talk page. (See Wikipedia:Village pump (policy)#Wikipedia is not an encyclopedia, and it will never become one unless, which comes at the same point, although from a slightly different angle; the nominator finds himself saying in despair: "nobody in this Village pump section appears to really share my real concern: NO INFORMATION IS BETTER THAN FALSE INFORMATION." He is, by and large, talking to the same people we are, and I sympathise with his frustration.)

I'm pretty sure that we'll never convince people like Blueboar or WhatamIdoing, because they've invested too much effort in defending the current policy: there's no prospect that they will climb down. But we might just convince their audience.—S Marshall T/C 22:26, 4 August 2011 (UTC)

With apologies for what is about to sound snarky, I appreciate that you and others have a legitimate concern about accuracy, but what you and others convinced me is that this talk page is just a club for endless and pointless talk. --Tryptofish (talk) 22:31, 4 August 2011 (UTC)

Nobody's forcing you to participate, of course. You and Nuujinn are welcome to take a 3-4 month break if you wish.—S Marshall T/C 22:37, 4 August 2011 (UTC)

Yes, and that's exactly what you want—For all the editors who disagree with you to get tired of your endless complaints and go away in the hopes that you'll get your way. That's why it's long past the point where you should just give it a rest. You don't demonstrate consensus by making it a test of endurance. Quale (talk) 01:52, 6 August 2011 (UTC)

S Marshall is investing time trying to represent the view of about 1/2 of the folks at the RFC. A few folks on the "zero change" side have been just as persistent, so your one-sided implication is not correct. North8000 (talk) 01:59, 6 August 2011 (UTC)

Those of us in the US just got finished with an explanation of consensus building by the US Congress, where it is said that a typical measure of a compromise is that nobody gets what they wanted.  As for the current proposal, it is based on a consensus that verifiable material may or may not be accurate, and three references previously suggested as relevant.  There is already wide disagreement here about what these other policies mean, but it is part of the point to move that part of the discussion away from WT:V.  Unscintillating (talk) 23:43, 4 August 2011 (UTC)

While each phrase is technically true, I don't think that this proposals is one of the better ones that have been put forward. WhatamIdoing (talk) 03:30, 5 August 2011 (UTC)

A genuine compromise to resolve this, or at least calm it for a while?

A genuine compromise to resolve this, or at least calm it for a while? Part A

[new subsection inserted above] Unscintillating (talk) 13:52, 6 August 2011 (UTC)

I figure that the sign of an actual compromise is something that that Blueboar opposes as too radical of a change and which S Marshall says is an insufficient change. How's this for one of those?:

Add a second sentence which says: "Not truth" means that no other consideration, such as "truth" may be substituted for meeting the verifiability requirement.

Sincerely, North8000 (talk) 02:18, 6 August 2011 (UTC)

Without commenting on the merit of the proposal, I already tried starting a sentence in a recent proposal with "Not truth" and people found multiple reasonable ways to misread it, so I suggest considering instead:

“ The phrase "not truth" means that no other consideration, such as "truth", may be substituted for meeting the verifiability requirement. ”

Unscintillating (talk) 02:37, 6 August 2011 (UTC)

I think that that is better. So, my revised proposal is to add the following as a second sentence:

“ The phrase "not truth" means that no other consideration, such as "truth", may be substituted for meeting the verifiability requirement. ”

Sincerely, North8000 (talk) 10:18, 6 August 2011 (UTC)

I think we can shorten the proposal to, "The phrase "not truth" means that "truth" is not a substitute for verifiability."  Unscintillating (talk) 13:52, 6 August 2011 (UTC)

Yes, that's even better. So now my proposal is to add a second sentence which says:

“ The phrase "not truth" means that "truth" is not a substitute for verifiability ”

North8000 (talk) 14:01, 6 August 2011 (UTC)

I would re-phrase this slightly as "The phrase "not truth" means that "truth" your belief that something is true is not a substitute for verifiability." WhatamIdoing (talk) 17:56, 10 August 2011 (UTC)

Sounds good. We could go even further and say, "your belief or knowledge that something is true is not a substitute for verifiability." --JN466 21:31, 11 August 2011 (UTC)

A genuine compromise to resolve this, or at least calm it for a while? Part B

Adding just the phrase " 'truth' " is looking at the issue of unsourced material from the viewpoint of those wishing to block the addition of material they believe to be inaccurate.  In cases where editors agrees that the material is accurate, but some still think verifiability is a first priority, WP:V could be more helpful.  What would have a broader viewpoint is:

from: The threshold for inclusion in Wikipedia is verifiability, not truth—whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true. to:      The threshold for inclusion in Wikipedia is verifiability, not truth—whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true. The phrase "not truth" means that "truth" is not a substitute for verifiability. Accuracy is not a substitute for verifiability.

Unscintillating (talk) 13:52, 6 August 2011 (UTC)

Proposal withdrawn.  Unscintillating (talk) 22:41, 6 August 2011 (UTC)

Alternate proposal X

Let's see if this dog can hunt:

The threshold for inclusion in Wikipedia is verifiability, not truth. The phrase "verifiability, not truth" is intended to convey the principle that accuracy is never a substitute for verifiability—readers must be able to check that any material in Wikipedia has been published by a reliable source.

Incorporating SV's comments and the wording I prefer, hurls rocks and stones as desired. --Nuujinn (talk) 17:38, 7 August 2011 (UTC)

Comments on the proposals

With B dropped, these are proposal A.

• Support - I have no objections to this... either with or without the "Accuracy is not a substitute for verifiability" sentence. Blueboar (talk) 15:07, 6 August 2011 (UTC)
• Support Prefer the one without the accuracy sentence, but either is good. North8000 (talk) 16:38, 6 August 2011 (UTC)

Ok, let's consider the "accuracy" sentence as an unrelated proposal, and one that is currently tabled.  Unscintillating (talk) 22:32, 6 August 2011 (UTC)

• Something along these lines would be OK with me too (and I'm happy to see people getting back to discussing concrete changes to wording). I also think that the "Accuracy..." sentence is not needed, because it seems repetitive to me. I think it would be more to the point to drop the words about "The phrase "not truth" means". Simply: ""Truth" is not a substitute for verifiability." But I also don't feel strongly about that point, so if people want to retain the part about "the phrase means", I don't particularly object. On the other hand, I also don't feel strongly that we need to do any of this. --Tryptofish (talk) 19:02, 6 August 2011 (UTC)

To me the "means" part is important and central.North8000 (talk) 21:10, 6 August 2011 (UTC)

• Well, it's a step in the right direction.—S Marshall T/C 21:16, 6 August 2011 (UTC)

• Weak Support, I can live with it, but would much prefer 'The phrase "verifiability, not truth" is intended to convey the principle that accuracy is never a substitute for verifiability.' Strictly speaking "not truth" cannot mean 'that "truth" is not a substitute for verifiability', as it is too short. But if the proposed version can bring an end to this endless discussion, sign me up. --Nuujinn (talk) 21:22, 6 August 2011 (UTC)

I think it implies explaining the intended meaning of saying "not truth" rather than a literal definition. But I think you allowed for this in your "strictly speaking" qualifier. North8000 (talk) 23:12, 6 August 2011 (UTC)

• Support  Part A.  Unscintillating (talk) 22:32, 6 August 2011 (UTC)

• Oppose. The proposals make a syntactical error. No one has suggested that the first part of the first sentence reads:

• {the threshold for inclusion} = {not truth}

Rather, it is:

• {the threshold for inclusion} ≠ {truth}

It's therefore a mistake to add a sentence explaining what the phrase "not truth" means, because it doesn't mean anything here. The key phrase is "verifiability, not truth," and the sentence explains what that means, namely "whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true."

There is no need to keep adding sentences that explain the meaning of the previous sentence, especially not when the explanations introduce errors or lack of clarity. For example, what is the purported difference between a true sentence and an accurate one?

North8000 and S Marshall have said they aim to make tiny changes that they hope no one will see as significant, e.g. S Marshall in June: "As I've explained before, the only way to make significant changes to the first sentence is via a slow accretion of stealth edits ...". I'm concerned that adding "explanatory" sentences will only lead to the next step in their effort to remove "verifiability, not truth" entirely. So if any of the supporters are doing so in the hope that it will end the discussion, I think that could be a mistake. SlimVirgin ^{TALK|CONTRIBS} 15:36, 7 August 2011 (UTC)

Slim, please AGF, especially when your assuming bad faith is creative and erroneous in my case. Addressing the structural part, the sentence is implicitly about explaining the intent of saying "not truth" in the first sentence, not trying to define those two words. And the goal of the compromise is to bolster it's intended meaning (reinforcing wp:ver) and reduce all of the mis-meanings that have been derived from it. And, while I think that removing those two words would be the ideal solution, I certainly don't view this as a step towards that. In fact, as a practical matter, the compromise actually would (unfortunately, sorry S. Marshall!) serve to entrench those two words by significantly reducing the active opposition to them. My own goals were (ONLY) two: 1. Reduce the unintended, non-wp:ver effects of those two words. 2. Try to facilitate an end to the current painful situation which has arisen because there was no consensus to either keep or remove those two words. Sincerely, North8000 (talk) 15:58, 7 August 2011 (UTC)

SlimVirgin, leaving aside the issue of intent, do you see the same problem in scope for 'The phrase "verifiability, not truth" is intended to convey the principle that accuracy is never a substitute for verifiability.'? I'd like to make sure whether I understand your comment. --Nuujinn (talk) 15:46, 7 August 2011 (UTC)

I don't see the same problem there, no. But it's still problematic. We could add "the phrase 'verifiability, not truth' conveys the principle that accuracy is never a substitute for verifiability." But what does that mean over and above "whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true"?

If people want to add something, we could say in a footnote: "Wikipedia articles aim only to offer an accurate and balanced overview of the relevant literature." But that takes us into the territory of NPOV/UNDUE, as others have pointed out. As I see it, the meaning of the first sentence is clear. I accept that a few editors don't like it, but the problem is not that it is unclear. SlimVirgin ^{TALK|CONTRIBS} 16:00, 7 August 2011 (UTC)

Briefly, the most common problem is that it causes pervasive mis-paraphrasing of wp:ver to say that that wp:ver weighs in for INCLUSION of inaccurate material, and that wp:ver excludes accuracy from conversations about possibly excluding false material. And I'm talking about situations that do not involve wp:npov. North8000 (talk) 16:08, 7 August 2011 (UTC)

But you've continued to say this for months without evidence. No one has argued that we should include material where everyone agrees that a source has made a simple error. All the policies must be applied with common sense.

The point of V is that we give an overview of the relevant literature, whether or not we agree with it. And the point of NPOV/UNDUE is to resolve how to present a balanced view of that literature, whether or not we agree with it. Editors argue about how to apply these principles, but there is no confusion about the principles themselves, no matter how often you say there is. SlimVirgin ^{TALK|CONTRIBS} 17:03, 7 August 2011 (UTC)

I submit that the last sentence of your first paragraph and that first 2 sentences of your second paragraph are correct but off-topic, and I disagree with the rest of your post and submit that wp:ver is pervasively mis-quoted / mis-paraphrased. Would be happy to discuss, give examples etc., but here is probably not the place for such a substantial discussion. North8000 (talk) 17:12, 7 August 2011 (UTC)

Please do give examples of it being "pervasively" misquoted. I've been editing for nearly seven years, with 115,000 edits to 13,000 unique pages, and I see it used and understood well. SlimVirgin ^{TALK|CONTRIBS} 18:24, 7 August 2011 (UTC)

OK here's the first one. Between not knowing how to provide a diff / link to a block of text and not wanting to personalize this, I'm putting it in here: This issue is a group of editors wanting to leave out (or offset / qualify) one error that a medium quality source made, which nobody was arguing was accurate. And it was not in a wp:npov balancing situation:

• Other experienced editor#1: Unfortunately, it seems that we still have Wikipedia editors who consider themselves to be more reliable sources than Walter Russell Mead writing in Foreign Affairs.
• Other experienced editor#2 Academic books journals are the best sources because they are peer-reviewed and writers must clearly distinguish between facts and opinions and when they express opinions must explain their degree of acceptance. If Mead had written his article for an academic journal then we would be clear whether his description was generally accepted or merely his own opinion. We could then look at later papers to see the degree of acceptance his views had, whether they represented academic consensus, a majority view, a minority view, etc. Instead, Wikipedia editors must decide among themselves what weight to assign the views.
• Me to experienced editor #1: I'm not sure whose comments you are responding to. If mine, you have characterized it. Wanting to leave out one obvious error the source made does not equate to what you describe. (North8000)
• Experienced editor #1: I was referring to your comments, which display a fundamental misunderstanding of Wikipedia's policy and mission. The role of a Wikipedia editor is to accurately summarize what reliable sources have written on a topic, not to weigh in with what he assumes is his expert opinion on matters of public policy. Mead is an expert, writing in Foreign Affairasc magazine. You are some random guy on the internet. You have no authority to "correct" a leading expert writing in a reliable source, or to inform us what is or is not obviously the case or what is right or wrong about anything. I am sorry that you find this difficult to understand

In other words, the fact that it was sourced is sufficient to force it to be in the article, any anyone who advocates leaving out an erroneous item has all of the above shortcomings. North8000 (talk) 20:52, 8 August 2011 (UTC)

As I mentioned before, each of the zillions of times that that someone says something to the effect of "do not remove sourced material" followed (if they are challenged) by the "not truth" quote, and it is not in the context of a wp:npov balancing situation, they are arguing that wp:ver weighs in on the side of INCLUSION of any /all sourced material. Also, each time when an editors are arguing for exclusion of false sourced material (and again not in the context of a wp:npov balancing situation) and somebody says that the accuracy argument for EXCLUSION of material is not allowed because of "not truth". And, as before, if someone wishes me to, I'll go find lots of examples of these. North8000 (talk) 11:59, 8 August 2011 (UTC)

Yes, please provide examples of people failing to understand the first sentence of the policy, or interpreting it to mean that clearly mistaken material must be added to articles. As this has been requested many times it would make sense just to offer some. SlimVirgin ^{TALK|CONTRIBS} 15:27, 8 August 2011 (UTC)

I certainly won't deny that I have every intention of seeking consensus to remove "verifiability, not truth" from the policy, and continuing to seek that until (a) I achieve my goal or (b) I see evidence of a genuine consensus that Wikipedians want to retain that phrase.—S Marshall T/C 17:26, 7 August 2011 (UTC)

I've put up an alternate to North8000's suggestion, sticks and stones welcome. --Nuujinn (talk) 17:40, 7 August 2011 (UTC)

I like yours better than mine, but fear that it will make things here really complicated. North8000 (talk) 17:49, 7 August 2011 (UTC)

If you mean things here, I doubt that's possible, but thanks for the compliment ;) --Nuujinn (talk) 17:59, 7 August 2011 (UTC)

How about putting that second sentence (or something like it), instead, in a footnote, and leaving the main text alone? --Tryptofish (talk) 00:16, 8 August 2011 (UTC)

That would be fine by me. What do other think, one way or another? --Nuujinn (talk) 00:38, 8 August 2011 (UTC)

I'd like the clarification to be prominent, per with "A" or "X". Sincerely, North8000 (talk) 02:48, 8 August 2011 (UTC)

SV states, "No one has suggested that the first part of the first sentence reads..."  This has to do with the ambiguity in the word "threshold", which creates both a threshold for the inclusion of truth as well as a threshold for the inclusion of not truth, but this ambiguity is not an issue currently being discussed.  In the context of the current conversation this appears to be a [red herring] argument. 

SV's direct objection does not pass inspection, if you look at the proposal it says exactly

“ "truth" is not ”

which compares algebraically with:

• {the threshold for inclusion} ≠ {truth}

Unscintillating (talk) 07:03, 8 August 2011 (UTC)

• I don't think that's quite right. Preventing that interpretation is, surely, the purpose of the comma before the word "not". To me, it's clear that the policy wants you to believe that the minimum criterion for inclusion is verifiability and the truth has nothing to do with whether something should be included on Wikipedia. I have no problem with the first limb of that, but I take issue with the second.—S Marshall T/C 09:06, 8 August 2011 (UTC)

Re: the policy wants you to believe that the minimum criterion for inclusion is verifiability and the truth has nothing to do with whether something should be included on Wikipedia. I disagree... the policy wants readers to understand that 1) the minimum criteria for inclusion is verifiability, and that 2) truth is not the minimum criteria for inclusion. The reason for this is that truth has nothing to do with verifiability. Truth can (and should) play a part in inclusion... but the part it plays comesafter we have established verifiability. If a statement is not verifiable, it does not matter whether it is true or not... we can't add it. If a statement is verifiable, then we can look at other criteria for inclusion... such as how much weight to give it, should it be stated as fact or as opinion, etc. It is appropriate to discuss the accuracy/truth of the material once we get to this secondary stage. But before we get to such discussions we must first cross the Verifiability threshold. Blueboar (talk) 13:05, 8 August 2011 (UTC)

That's not what it says.—S Marshall T/C 18:43, 8 August 2011 (UTC)

Points 1 and 2 are exactly what the policy says, using other words. The rest comes from understanding what our other policies say. Blueboar (talk) 21:25, 8 August 2011 (UTC)

Agreed. The lead clearly says that the three content policies "jointly determine the type and quality of material that is acceptable in articles. They should not be interpreted in isolation from one another, and editors should familiarize themselves with the key points of all three." Someone recently removed that the three work in harmony, which should probably be restored to underline the point that none of them can be understood or applied without reference to the others. SlimVirgin ^{TALK|CONTRIBS} 21:40, 8 August 2011 (UTC)

It's still not what this policy says. Find clearer language.—S Marshall T/C 07:07, 9 August 2011 (UTC)

Why don't you suggest come clearer wording? I don't see a problem with the current wording myself, and if you're not willing to work with other editors, I see no point to continuing this discussion. --Nuujinn (talk) 09:55, 9 August 2011 (UTC)

I suggest amending the first sentence to read:- "The minimum standard for inclusion on Wikipedia is verifiability." It's not necessary to talk about truth in WP:V at all.—S Marshall T/C 10:09, 9 August 2011 (UTC)

Well, since that isn't going to happen (we have long since established that there is no consensus to change the first sentence), perhaps you could come up with another suggestion? Blueboar (talk) 13:28, 9 August 2011 (UTC)

Sure. "The phrase 'not truth' does not excuse introducing inaccuracies into the encyclopaedia." How's that?—S Marshall T/C 14:56, 9 August 2011 (UTC)

Now, that's really good. North8000 (talk) 14:59, 9 August 2011 (UTC)

No, that's really lousy. It will be widely understood as meaning that I get to delete anything that I personally believe is "inaccurate", e.g., that climate change is real, or that vaccines don't cause autism, or that Barack Obama was born in the US, or anything else you want to name. That sentence is nothing more than a gift to POV pushers and crackpots.

We could legitimately say that "The phrase 'not truth' does not excuse misrepresenting the contents of reliable sources", but the major point of this policy is that that your personal beliefs about what's true/accurate/real are irrelevant. WhatamIdoing (talk) 17:54, 10 August 2011 (UTC)

With respect, that's a serious misunderstanding of the policy. The major point of this policy, to the extent that it has a major point, is that before information can be included in Wikipedia it must be verifiable. The business about "not truth" is a fossil, a remnant of an edit originally made to a draft version of NOR, and now stripped of its context by multiple subsequent edits. The unsupported opinion statements that "it will be widely misunderstood" needs to be backed up by evidence of diffs that show where the phrase "not truth" has led to the resolution of a misunderstanding—or has, in any other sense, been unambiguously helpful—but the other wordings proposed could not have worked.—S Marshall T/C 18:26, 10 August 2011 (UTC)

WhatamIdoing is right. The point of the first sentence is that what editors believe is irrelevant. What we do is offer readers an overview of the relevant literature, period, including when we strongly disagree with it. SlimVirgin ^{TALK|CONTRIBS} 18:42, 10 August 2011 (UTC)

I disagree; the statement that you and WhatAmIDoing are making is so broad that it includes things that are and are not policy. The latter unless you count the roaming mis-guided chants derived solely from the infamous two words as policy. North8000 (talk) 18:49, 10 August 2011 (UTC)

And thus we come to what lies at the heart of why we can not reach a consensus on language. We are never going to agree on wording if don't even agree on the basic principles that lay behind the wording. Personally, my understanding of the policy is a lot closer to that of WhatamIdoing and Slim Virgin than that of North and S Marshal. Blueboar (talk) 20:26, 10 August 2011 (UTC)

The onus is on North8000 and S Marshall to show that their understanding of policy is one that is widely shared; not just state that, but show it. SlimVirgin ^{TALK|CONTRIBS} 21:00, 10 August 2011 (UTC)

Wait, what? What I said was: The major point of this policy... is that before information can be included in Wikipedia it must be verifiable, and you lot are actually disagreeing? Is this some kind of surreal practical joke?—S Marshall T/C 21:01, 10 August 2011 (UTC)

Um... you might want to read your comment again... you said a lot more than just that (the bit about "not truth" being a fossil for example). Blueboar (talk) 21:19, 10 August 2011 (UTC)

S Marshall and I are saying is wp:ver means what it says (the 2 ambiguous infamous words aside) and, perhaps unknowingly, the statement that you three are supporting is saying that in addition to that it means something very different which it does not say. Specifically that it mandates INCLUSION of material if it is RS'd. So, S Marshall and I are saying that wp:ver means what it says, and you are saying it means something it doesn't say. And WHO did you say the onus is on? ! ? ! ? North8000 (talk) 21:26, 10 August 2011 (UTC)

It doesn't mandate inclusion of everything that's sourced. It says verifiability is the threshold for inclusion, i.e. no inclusion unless a source exists. This discussion has reached the point where you're assuming people don't understand English, and I can't see the point of that. SlimVirgin ^{TALK|CONTRIBS} 21:37, 10 August 2011 (UTC)

Slim, roughly speaking, you just did a 180 from a few hours ago, and now are agreeing with us. If we wanted to really get to the bottom of this, we would logically analyze the statement that you were supporting a few hours ago. It may not be apparent that, roughly speaking, it included the "mandating inclusion". North8000 (talk) 22:49, 10 August 2011 (UTC)

North, with respect, it's just that you're taking what people say way too literally. No one is counting simple errors by sources. No one is saying that, when the Professor of Mathematics at Cambridge says two plus two equals five, we must include it, or even waste time discussing it. We ignore it, because all agree it's a mistake. SlimVirgin ^{TALK|CONTRIBS} 23:54, 10 August 2011 (UTC)

If that was the problem, Slim, nobody would be wasting any time on it. But what about when an accredited university publishes a book on baraminology and someone wants to use it as a source? WP:V as presently written positively encourages them to do that. But, as we've exhaustively demonstrated above, our current practice, in writing articles, would be to exclude such a source on the grounds that it's pseudoscience. What North and I are looking for is wording that documents our current practice.

What we've discovered is that when we mention the existence of these technically-reliably-sourced pseudoscientific claims, is that Wikipedians frame their answers in terms of which source is more reliable. Can you see the logical problem there?—S Marshall T/C 07:03, 11 August 2011 (UTC)

I sure can't. What's broken? Our current practice is governed by consensus, informed by policy, and it appears to be working fine. Otherwise, we'd be overrun by the mad hordes trying so desperately to add untrue material to WP. Also, we could use the work you cite as a source in some contexts, just not in scientific articles. --Nuujinn (talk) 10:20, 11 August 2011 (UTC)

Do you accept that policy should document good practice? My position has always been that our current practice is working as intended, and that policy should record it. And if we can't use the work I cite as a source in scientific articles, then shouldn't WP:V say so? (My position is actually that the baraminology text would have a place only in articles about baraminology, creationism, or related fringe theories, but I'm running with your view for the moment.)

As for what the logical problem is, it's this: when we decide which source is "more reliable", what we're actually deciding is which source is more likely to be true. Aren't we? And doesn't that make a nonsense of "not truth"?—S Marshall T/C 10:37, 11 August 2011 (UTC)

No, I don't. Policies prescribe required practice, they do not and should not document good practice. And no, V should not be so specific as to say which sources are reliable for specific topic areas, that is more appropriate in my view for guidelines, which are informed by policies. I don't know if you have noticed, but we decide everything here--if we decided to change V to say that only squirrels were qualified to verify sources, that would be our policy. And I agree with you that the baraminology text you cited would have limited value as a source--you asked "But what about when an accredited university publishes a book on baraminology and someone wants to use it as a source?" My answer is, we would do what we do, come to consensus about the usage of the source and so long as we did not violate policy, we're fine. That's normal operations around here. Now, if you need some help fending off someone trying to use that book in an inappropriate context, I'm happy to weigh in, but the value of Gedankenexperimenten is limited here. And no, when we decide which source is more reliable, we are explicitly not deciding which source is more likely to be true, we're deciding which source is considered more likely to be considered more accurate by experts in the field, which is a much different matter. --Nuujinn (talk) 10:55, 11 August 2011 (UTC)

(making more room for this) --Nuujinn (talk) 10:57, 11 August 2011 (UTC)

• I'll set aside the question of when policy should document the practice we already use, and when it should be used to effect a change of practice, for the moment. It's an interesting one, and central to my position, but even more crucial is the second point. I said, "when we decide which source is 'more reliable', what we're actually deciding is which source is more likely to be true", and you disagreed. You said it was more a question of which source would be "considered more accurate", by "experts". But I don't think this really does constitute disagreeing with my point. "Considered more accurate" is a synonym for "considered more likely to be true", and "expert" is a synonym for "one who has studied the subject and is likely to know what is true". I think that all you're doing is inserting a semantic layer between your position and the word "truth".—S Marshall T/C 11:19, 11 August 2011 (UTC)

It is may be a simple semantic layer, but it's an important one that affects what we can and cannot do. Everyone can edit here. Everyone can participate in discussions of content. I am reasonably well educated, but I am competent as an expert in only a small number of fields. Making a judgement about which source is more likely to be an accurate reflection of what experts hold to be true is fundamentally different than deciding which source is considered to be "true". For the former, I can use my knowledge about writing, research, journalism, academic sources, religion, and any number of indirectly related subjects. To do the latter requires that I have expertise in the subject area. That's a very important difference. --Nuujinn (talk) 13:51, 11 August 2011 (UTC)

I'm afraid I don't think that position holds any water at all, because it's hiding behind the undefined term "expert". I can only think of two objective definitions of "expert" that would include a lecturer in evolutionary biology but exclude Dr Timothy R. Brophy of Liberty University. The first is by defining an expert as someone who knows the truth, which is unfortunately rather circular, and the second is by defining an expert in terms of the mainstream scientific consensus (or "most common academic view", or whatever), which is what Blueboar and I discussed at such length earlier.—S Marshall T/C 14:31, 11 August 2011 (UTC)

IMHO once you try to define the finish line and how to get there (regarding potential exclusion of false material) things get complicated. Right now the prevalent mis-interpretations of wp:ver basically say that any such conversation is illegitimate, and say that the falseness of a statement may not enter into any conversation which would be determining possible exclusion of the material. So the remedy is much simpler than trying to define the route and end result for excluding material based on falseness, we just need to essentially get wp:ver to neutral ground on that topic. And simply let the conversation occur, an dlet considerations enter into exclusion conversations. Which basically means mitigating the unintended (-by-most) consequences of "not truth". North8000 (talk) 15:35, 11 August 2011 (UTC)

yet another arbitrary break

(edit conflict) In its original context, VNT read: "One of the keys to writing good encyclopedia articles is to understand that they should refer to claims that have become an accepted part of the public domain. In this sense, Wikipedia is about what is verifiable, not what is true." When first added to this policy, the VNT phrase was the fifth subsection of the page. Subsequent edits have emphasised the "not truth" aspect, stripped it of its original nuanced context, and placed it as the first sentence of the policy. The current version is, in this sense, a relic, or fossil, of the version in its original form. I really do think I'm writing quite plainly and simply here.—S Marshall T/C 21:30, 10 August 2011 (UTC)

Status So it looks folks liked proposal "A" (which I served up) 6:1, notably with the folks with the strongest views on both ends of the spectrum weighing in. Nuujinn created proposal "X" which I like even better. Are we to the point where we should weigh in on proposal "X"? North8000 (talk) 15:22, 8 August 2011 (UTC)

About "status": actually there is also the previous discussion, that got archived while people were engaging in discussion of meta-issues. Personally, I'm OK with leaving the page unchanged, or simply adding a footnote. But if the users who want to change the page actually want something to happen, as opposed to having a club for endless discussion with no endpoint, they should be prepared to compromise, and to focus on the actual wording of the page. --Tryptofish (talk) 16:00, 8 August 2011 (UTC)

Of course, plus about 30,000 words of other thoughts. I was more trying to address concrete compromise current proposals, and seeing what folks thought about methodically weighing in on Nuujinn's proposal X. North8000 (talk) 13:57, 9 August 2011 (UTC)

The ground rules here currently include that:

1. There is no consensus to change the first sentence.
2. There is no consensus to leave the first sentence unchanged, or at least the unintended meanings in the first sentence.
3. We are working on an additional sentence or sentences after the first sentence to clarify the intended meaning and/or remove the unintended meaning of the first sentence.
4. WP:Consensus is a policy that all editors should normally follow.
   

This having been said, Proposal X is outside the current scope of discussion.  Unscintillating (talk) 03:19, 11 August 2011 (UTC)

OK, so you are saying that Nuujinn's Proposal X is off the table now because it changes the first sentence. With support for "Proposal A" being 6 of 7 of the folks with the strongest opinions from both ends of the spectrum, I'm thinking of being semi-bold and putting it in. Comments? North8000 (talk) 14:10, 11 August 2011 (UTC)

What's Proposal A? I see a Proposal X, but no Proposal A. A Quest For Knowledge (talk) 16:13, 11 August 2011 (UTC)

I've been watching this discussion from the sidelines. The version I would accept is the one posted by WhatamIdoing at 17:56, 10 August 2011 (UTC) - "The phrase "not truth" means that "truth" your belief that something is true is not a substitute for verifiability." I feel the "your belief that something is true" phrase is critically important. The core purpose of this policy is to exclude editors' opinions and beliefs from article content. Roger (talk) 16:48, 11 August 2011 (UTC)

That's what the phrase "verifiability, not truth" means. And the first sentence of the policy already explains what it means: "whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true."

Indeed you are correct SlimVirgin, the proposal seeks to add a statement which is in fact already there - phrased differently, but meaning the same. Consequently I withdraw my initial (conditional) support. Roger (talk) 21:21, 11 August 2011 (UTC)

I think what's happening here is that editors opposed to change are making concessions in the hope of bringing the discussion to an end. I see that as a mistake, because it will lead to contradictions or repetition being added to the policy as a desperate measure. The best way to stop the discussion (if that's what people want) is to stop taking part in it. SlimVirgin ^{TALK|CONTRIBS} 18:48, 11 August 2011 (UTC)

The best way to stop the discussion is, clearly, to reach consensus.—S Marshall T/C 19:13, 11 August 2011 (UTC)

I guess I'm a bit guilty myself of what Slim called making concessions in the hope of bringing the discussion to a conclusion. I like the existing first sentence, but I was hopeful that a clarifying second sentence would be a constructive way to resolve the concerns about "not truth". I think I was right on the merits, but wrong on the dynamics of this talk page. I've actually come to the conclusion that some of the editors who say they want to change the page are unwilling to accept any change that could actually achieve consensus. They, some of them, want either to overturn the concept of verifiability, or, if they can't achieve that (and they can't), to discuss the issue endlessly without ever agreeing to anything. For the rest of us, there is no best way to stop the discussion, but there is the option of ignoring it. --Tryptofish (talk) 20:13, 11 August 2011 (UTC)

I think we already have it, we just lost track. North8000 (talk) 20:23, 11 August 2011 (UTC)

I'm definitely in the camp that Slim refers to, not to bring discussion to an end, but to try to address some legitimate concerns brought up by North8000 and some others. At this point I believe very firmly that there are a couple of editors who are unwilling to work towards a consensus, and are in effect, if not by intentional, stonewalling discussion. --Nuujinn (talk) 22:07, 11 August 2011 (UTC)

I strongly concur with SlimVirgin's position on this issue in full, including the position that no compromise is possible on this issue. It appears S Marshall and North8000's position is not informed by a thorough understanding of social engineering and hacking (in which hackers boost their egos by attacking anything perceived to be important). The longstanding "verifiability not truth" language fits together with many other components of Wikipedia's core content policies to control Wikipedia's risk of getting hit with more embarrassing scandals like the Wikipedia biography controversy. Deleting that restriction will foreseeably lead to another Seigenthaler-like scandal which will critically if not fatally damage the encyclopedia's credibility. --Coolcaesar (talk) 07:27, 15 August 2011 (UTC)

First to clarify, in a large RFC, about half of everybody wanted the "not truth" removed. Second, I'm assuming that you are discussing situations where bad material got in. EVERYBODY involved wants to keep wp:verifiability at full strength. The core argument is that the words "not truth" are not needed for that, and do more harm than good via. unintended consequences. Vs. folks who feel otherwise. North8000 (talk) 10:36, 15 August 2011 (UTC)

Time to try to end it?

The version of Proposal A which received the 6 to 1 consensus, notably amongst the persons from the farthest ends of the spectrum on both sides was to add a second sentence which says: "The phrase "not truth" means that "truth" is not a substitute for verifiability" I am going to be semi-bold and try putting it in based on that consensus. Maybe that will put an end to this.

I thing that WhatAmIDoing's version farther down under "A" is even better which is to add a second sentence which says: The phrase "not truth" means that your belief that something is true is not a substitute for verifiability. However it did not go through that consensus process. We could change it to that later, and IMHO we should.

Finally, I think that Nuujinn's proposal "X" is the best of them all, but that it not on the table because it involves changing the first sentence.

So here goes!.....maybe this will settle it!  :-) Sincerely, North8000 (talk) 20:46, 11 August 2011 (UTC)

I did it. Now I'd like to propose changing it to WhatAmIDoing's version. North8000 (talk) 20:57, 11 August 2011 (UTC)

Well Slim, who was the "1" opposed in the above-described consensus took it out, saying "no consensus". Now what? North8000 (talk) 21:09, 11 August 2011 (UTC)

You've basically got two choices:- dispute resolution (in an attempt to enforce consensus—good luck with that, but I think Slim will successfully argue that six people doesn't constitute sufficient consensus to change a policy)—or back to discussion.—S Marshall T/C 21:33, 11 August 2011 (UTC)

Here we go again! If we're going back to the drawing board, I rather it be with Nuujinn's proposal "X". North8000 (talk) 21:51, 11 August 2011 (UTC)

Third choice... accept that you are not going to be able to change things and move on. Blueboar (talk) 22:09, 11 August 2011 (UTC)

Why on earth would we accept that? I mean, we would if we were sure nobody's mind could be changed through rational discussion and nobody new would join the conversation. But surely neither of those conditions obtain.—S Marshall T/C 22:19, 11 August 2011 (UTC)

I think you should be forced to accept it. Endlessly plowing the same ground can be considered disruptive on wikipedia, and I think that this could definitely be nearing that point. The discussion has remained civil, but really this was started in April or maybe even March. It would have been appropriate to table this all the way back in June since I don't think any progress is being made. Quale (talk) 23:03, 11 August 2011 (UTC)

I don't know why it is necessary to repeat both that (1) there is no consensus to change the first sentence, and (2) there is no consensus to not change the first sentence.  However, given that it is necessary to do so, I agree with Blueboar.  I'd suggest a four-week hiatus, at which time it is up to the involved editors to show that they can move forward without the involvement of informal mediation.  Unscintillating (talk) 01:08, 12 August 2011 (UTC)

Quale, if you believe it's disruptive to continue a very long discussion in the absence of any kind of consensus, please feel free to start a RFC/U on me. My position is that it is not my fault that no progress is being made. Unscintillating, how will your proposed hiatus help?—S Marshall T/C 07:41, 12 August 2011 (UTC)

My position is that it is your fault that no progress is being made, along with North who shares a roughly equal amount of blame. I am content to let people draw their own conclusions based on the record, although you'd have to be a masochist to wade through the four months and counting of wheel-spinning this page has seen so far. Quale (talk) 23:10, 12 August 2011 (UTC)

So, about half the folks at the large RFC want to completely remove "not truth" , half don't, so there is no consensus to keep or remove. The discussed middle ground is keeping "not truth" with an added sentence which says/clarifies/reinforces that the meaning of "not truth" in this policy is only to reinforce the verifiability requirement. So of the three possibilities, the one at the extreme end of the spectrum (keep "not truth" and add no wording) is the one that is currently in there by default. I am SICK AND TIRED of people who like that no-consensus-status-quo-from-one-end-of-the-spectrum villainizing even everybody seeking the middle ground as committing some type of misbehavior!! That's ridiculous, and certainly does not help develop any middle ground. Sincerely, North8000 (talk) 11:39, 12 August 2011 (UTC)

As far as a middle ground solution, we had one proposal that received a 6 out of 7 approval from the main combatants. And the "1" (Slim) did not claim it was a policy change, just that it was redundant or had wording problems or something like that. And the "1" reverted it saying "no consensus". Sounds like an ownership problem enacted through a massive double standard; the "double standard" has been a pattern here. North8000 (talk) 12:00, 12 August 2011 (UTC)

Should we try to

1. Enforce the consensus
2. Take the same idea out for broader comment
3. Villianize and stonewall the people seeking the middle ground until they just give up
4. Craft a new proposal which just adds a sentence
5. Float Nuujinn's proposal "X" for broad input. (possibly with minor tweaks prior to that)
6. Other "move forward" idea.

North8000 (talk) 12:08, 12 August 2011 (UTC)

Comments on ideas

• 2.—S Marshall T/C 12:09, 12 August 2011 (UTC)
• I prefer #5, #2 as a second choice, #4 as a third choice. #1 is good in principle but let's not. North8000 (talk) 12:10, 12 August 2011 (UTC)
• It's not a good idea to say, in effect, "not truth actually means something else". If we are going to have a "definition" of "not truth", I would prefer that it be in a footnote, not the main text. At the bottom of Wikipedia talk:Verifiability/Archive 49, I suggested a way to formulate it differently, so that wouldn't be a problem. (It's listed along with the current wording on the page, and I'm just fine with leaving the page as it currently is.) --Tryptofish (talk) 19:07, 12 August 2011 (UTC)
• Please, let's not talk about enforcing a consensus when there isn't one. Abandon #1. Septentrionalis PMAnderson 21:42, 12 August 2011 (UTC)
• None of the above. No consensus for the change you want, so just drop it for now. Take it up again in three to six months if you want to test whether consensus has changed. Quale (talk) 23:10, 12 August 2011 (UTC)

Breaking the sentence up into its component parts

The sentence that is causing all the commotion reads: "The threshold for inclusion in Wikipedia is verifiability, not truth—whether readers can check that material in Wikipedia has already been published by a reliable source, not whether editors think it is true." This sentence has four component parts:

1. threshold for inclusion = Verifiability.
2. threshold for inclusion ≠ Truth.
3. threshold for inclusion = readers can check
4. threshold for inclusion ≠ editors think it is true.

I strongly agree with all four component parts. Does someone disagree with them? (and if so, which)? Blueboar (talk) 21:33, 12 August 2011 (UTC)

I'm probably not the target respondent, but I agree with all four too. I think the disagreement is, instead, with the way that #2 can be misconstrued to mean inclusion of untrue material is acceptable. --Tryptofish (talk) 21:39, 12 August 2011 (UTC)

#1 seems to me to be the point of this policy. #2 is certainly the main problem for me—particularly, its interaction with WP:FRINGE and WP:HOAX. I have other concerns as well, but I appreciate that you're trying to simplify here.—S Marshall T/C 22:30, 12 August 2011 (UTC)

Among other things, #2 actually does mean that untrue material is acceptable, as long as it is verifiable. The issue is rather that, as we can't decide here whether something is true or not, we all hope that the share of untrue material is lower in the subset of verifiable information than its share in all information (including, among other things, information made up by Wikipedia editors).  Cs32en Talk to me  23:25, 12 August 2011 (UTC)

But that's not how it works in practice: untrue, but verifiable, information is typically relegated to articles about fringe theories or notable hoaxes. It's typically not included in the main articles for each topic. Thus, species doesn't mention baramin; pyramid doesn't mention ancient astronauts; and Apollo program doesn't mention moon landing conspiracy theories. Because in practice on Wikipedia, untrue material is only acceptable in quite specific circumstances. My position is that policy should document practice here.—S Marshall T/C 23:59, 12 August 2011 (UTC)

Yes, the policy talks about "the threshold for inclusion in Wikipedia"... it does not talk about inclusion in specific article X (be it Species, Pyramid, Apollo program.) This has nothing to do with the Truth or non-Truth of the material. All verifiable information needs to be weighed in the context of the article topic... that is part of DUE weight. Blueboar (talk) 00:26, 13 August 2011 (UTC)

As usual when we have this conversation, my reply is that the policy doesn't say that. We could add a sentence to explain what you've just explained, though, if we can find a phrasing that won't give the Anti-Change Party a conniption.—S Marshall T/C 01:12, 13 August 2011 (UTC)

Blueboar, I agree with the statements of all four parts. But that saying #2 is unneeded and does more harm than good because it is so prone to misinterpretations. North8000 (talk) 00:06, 13 August 2011 (UTC)

I disagree... I think it does far more good than harm. Perhaps you were not around before we added it, but I was... we used to have all sorts of unverifiable information being added on the basis that "it's true". Now we don't. (Or at least we now get a lot less of it... and we have a solid statement that allows us to remove it when it does happen). Blueboar (talk) 00:26, 13 August 2011 (UTC)

OK, if we could just keep that effect 100% while reducing the misuses. I think that that is doable. North8000 (talk) 01:36, 13 August 2011 (UTC)

Timestamp to prevent autoarchiving—S Marshall T/C 18:31, 16 August 2011 (UTC)

Machine translation

The guidelines for non-English sources (WP:NONENG) include the following statement: "Translations published by reliable sources are preferred over translations by Wikipedians, but translations by Wikipedians are preferred over machine translations." I propose that this guideline should be revised. Translation is a complex cognitive task, and while machine translations may sometimes be accurate, assurance of quality requires a human agent. As the relevant article correctly states "Relying exclusively on unedited machine translation ignores the fact that communication in human language is context-embedded and that it takes a person to comprehend the context of the original text with a reasonable degree of probability. It is certainly true that even purely human-generated translations are prone to error. Therefore, to ensure that a machine-generated translation will be useful to a human being and that publishable-quality translation is achieved, such translations must be reviewed and edited by a human." The policy should state that fluently bilingual Wikipedians may use a machine translator, provided that they evaluate and correct the output, but that unassisted machine translations should never be permitted. I'd also suggest a new requirement for all translations to be independently checked by another editor. Rubywine . talk 02:23, 5 August 2011 (UTC) On second thoughts it's impractical to impose a proofreading requirement. Rubywine . talk 11:43, 6 August 2011 (UTC)

Unassisted machine translation pretty much has to be permitted, because we don't always have another good option, especially for less-common languages. A machine translation is more friendly to the readers who don't read the other language than presenting them solely with a completely unintelligible quotation in a foreign language. We have non-English quotations and phrases in all kinds of languages, and it is unreasonable for us to assume that any of our readers know anything other than English—much less that any single reader will be able to understand العربية and 한국어 and Македонски and 文言—and in the course of reading the English Wikipedia, they might easily encounter all of those languages at least once.

When and if we get a competent translator to that article, then the machine translation can and should be replaced by a good translation by a skilled human, but that doesn't mean that we have to leave readers in between wondering what it says.

Additionally, WP:Nobody reads the directions, so no matter how much you discourage it, it's actually going to happen. WhatamIdoing (talk) 03:08, 5 August 2011 (UTC)

I don't agree at all. There is no need to include machine translation into the article at all, in fact in doubt you would classify them as an unreliable source. The fact that we cannot expect readers to be versed in all those language is somewhat irrelevant, as all that the reader actually has to do is plugin the text into an automatic translator (such as Google). The notion that we have machine translation of foreign language quotes seems rather unappropriate to me.--Kmhkmh (talk) 06:04, 5 August 2011 (UTC)

Fluently bilingual (and multilingual) Wikipedians may use a machine translator, of course. Using a machine translator but evaluating and correcting the output is what I do when I'm translating, and I don't see a problem with it. As for the requirement that all translations should be independently checked—my question is, by whom? I certainly don't offer to spend my volunteering time checking other people's translations on a regular basis. I'll consider doing so on specific, respectfully-worded request on my talk page. No policy is going to require me to do it.

The basic problem with translations is the misapprehension that any editor is entitled to require other editors to help them check sources. I've seen editors demand that independent editors run around translating texts in languages they don't speak, on various occasions, and the answer is "bugger off".—S Marshall T/C 09:14, 5 August 2011 (UTC)

No policy requires you to do anything at all. Nobody here is entitled to demand anything from anybody. Regardless of how much translation work you volunteer to do, there's no reason for any policy to accommodate your personal disinclinations. There are thousands of bilingual and multilingual Wikipedia users; if translation work is only being undertaken by a very small group, that is undesirable and unnecessary. I am in favour of an active drive to engage a great many more people in this work. Rubywine . talk 11:11, 5 August 2011 (UTC)

I think nobody has an issue with machine translationen being used as tool. The issue here is merely that we should not include unassisted machine translation into articles (as quote or citation translations). Any unassisted machine translations shouldn't be a part of the article itsself due to potential unreliablity. That unsassisted machine translations are used by editors to perform a crude check of foreign language sources or on discussion pages to support an argument is perfectly fine of course. If somebody is using them for assisted translations that's probably fine, though personally I'm a bit skeptical there. However how an editor exactly produces a translation is out of our control anyway and probably not really our busines to begin with (as long as there is no legal issue like a copyright violation). The only thing that matters is the quality of the translations when he enters it into WP and that is responsible for it (rather than some program).--Kmhkmh (talk) 14:04, 6 August 2011 (UTC)

I agree with Kmhkmh in every respect. WhatamIdoing, you seem to be proceeding from an underlying set of assumptions which I think are wrong, and inconsistent with policy. Wikipedia: Translation specifically states: "Translation takes work. Machine translation often produces very low quality results. Wikipedia consensus is that an unedited machine translation, left as a Wikipedia article, is worse than nothing." I see no reason to take the view that unedited machine translations of smaller quotes and phrases are any more acceptable. The worst possible outcome is for readers to be misled and misinformed. Rubywine . talk 11:30, 5 August 2011 (UTC)

Machine translations are vaguely accurate but can introduce problems. At the very least the grammar is usually very poor; at worst the translation is misleading and just plain wrong. It really depends on their usage - if it's copy/pasted without checking then no, it shouldn't happen. However I and I'm sure many others have used a machine translation to inform and suggest content which is then written/copyedited into prose. That is fine. violet/riga _^[talk] 14:34, 5 August 2011 (UTC)

It sounds like you you've all missed the context of this question. Machine translation is only mentioned once in this entire policy, and it is mentioned solely in the context of supplying information about non-English reliable sources. This is not about translating whole articles: This is not about the fact that Bacterial pneumonia is way better over at the Spanish Wikipedia and I'd like to have some competent person translate it for us, or any situation even remotely like that. For that purpose, which has nothing to do with this entire policy, I agree that machine translation would be inadequate.

This recommendation appears in WP:NONENG and is specifically about direct quotations from non-English sources. There are only two reasons why you might be typing a direct quotation from a non-English source into Wikipedia:

1. You've been asked for a quotation from a non-English source, and the person who asked you doesn't speak <fill in the blank>, so you've been asked to provide a translation into English.
2. The article needs to include a famous quotation, book title, or similar short phrase for some purpose.

The policy says that in such cases, you need to supply the best translation you can, which means (in order):

1. A proper translation by a reliable source is best.
2. A translation by one of our competent editors is second-best.
3. If all else fails, a machine translation is better than nothing.

This means, to give a practical example, that at Thomas Corneille#Notes, we don't stop with just "In a letter to her father, shortly before her execution, Charlotte Corday quotes Thomas Corneille: "Le Crime fait la honte, et non pas l’échafaud!"" and leave the 97% of the world that can't read French wondering what it means.

Instead, we add "(The crime causes the shame, and not the scaffold!)", which is a decent translation (since we have one) immediately afterwards. If we didn't have this decent translation, then we would add the worse-but-not-too-awful machine translation, "(Crime is a disgrace, not the scaffold)". What we don't do is leave all the people who don't read French scratching their heads over why that sentence is in the article and what it means. [ip address redacted] User:Fred Bauder Talk 19:41, 6 August 2011 (UTC)

Thanks for your thoughts and comments, and also for pointing us to the relevant policy - that's most helpful. However, you're wrong to think that we've all missed the context. Yesterday, after quoting Wikipedia: Translation: "Wikipedia consensus is that an unedited machine translation, left as a Wikipedia article, is worse than nothing", I said "I see no reason to take the view that unedited machine translations of smaller quotes and phrases are any more acceptable."

Your argument has actually consolidated my viewpoint. Direct quotations are very likely to be idiomatic, so I disagree with WP:NONENG re machine translation. The example you've given is not bad at all, but I'm confident there'll be worse ones. I'd rather provide readers with a foreign-language quotation than with a hit-or-miss, potentially very bad translation of an idiomatic phrase. I think it's ok to leave the readers scratching their heads. They can look elsewhere for a good translation, or, if we really want to be helpful to them, we can source one for them ourselves. Rubywine . talk 18:18, 6 August 2011 (UTC)

Now consider the other situation: Someone asked you for a quotation to show that the source really does support the text. You provide a quotation, and the other editor says, "But I don't know what that means; I don't read ____". And you say, "Well, that's too bad, because I'm no good at rendering it in English, and Rubywine refuses to let me give you a machine translation." Is providing no translation, on the grounds that you can't provide a really good one, going to help that editor at all? Would that not be exactly the sort of rule that is described as "a rule that prevents you from improving or maintaining Wikipedia"? WhatamIdoing (talk) 18:55, 6 August 2011 (UTC)

If the machine translation of the quotation is merely clumsy, then it's probably better than nothing. However, there's every chance that a machine translation of an idiomatic expression will be nonsense, or far worse, completely misleading. A machine translation can convey the precise opposite of the original text. Providing misinformation is not improving and maintaining Wikipedia, it is damaging it. And that is why I believe we should either go to the effort of sourcing a good translation, or we should provide none. It is much better to say "I don't know" than to tell a lie. What's more, I think it is unlikely that we will often be in a situation where a good translation of a direct quotation cannot be externally sourced. Throwing in an unedited machine translation amounts to laziness. Rubywine . talk 19:12, 6 August 2011 (UTC)

It depends entirely on what class of direct quotation you're talking about. Professional translations are very frequently available for famous proverbs (which is convenient, because aphorisms are hard). Good translations are often available for titles of major books and plays.

However, they are almost never available for sentences in non-English sources that we are using to support article content. There are thousands of academic journal articles and millions of newspaper articles that are not written in English, and there are almost never professional translations available for them. If someone asks for a quote and a translation, and the best we can do is a machine translation, then we definitely should provide that.

On the general question, WP:There is no deadline. Translations do not have to be perfect on the first try. If providing a poor translation (whether due to machine translation or a lack of skill in a human translation) is the best we can do, then perhaps the existence of the poor translation will inspire some more capable editor to improve it. WhatamIdoing (talk) 04:05, 7 August 2011 (UTC)

No we shouldn't! Why should we offer unreliable ,questionable translations? Editors should never include content they can't really vouch for and that's exactly what an unassisted machine translation is. If a reader wants to roughly verify the accuracy of some citation or quote he can copy it into a translation machine (such as google) himself, there is no need for us to include such "rule of thumb" reasoning into the article itself. If you're afraid the reader is unable to use or find a translation engine, provide him a link to some online translation machine like google in the footnotes, but do not include that translation into the article or worse create the possible impression to readers the translation was actually produced by somebody who knew what he was talking about.--Kmhkmh (talk) 11:16, 7 August 2011 (UTC)

One answer is that we can offer "unreliable questionable" translations when they ipso facto support the relevant point.  See [Jaakko Pöyry], in which the mostly-incomprehensible machine translation of the name of the book is a strong indication of the notability of the subject.  English-only readers entirely get both the point that this is a machine translation of idiomatic Finnish, and that Jaakko Poyry is a man who has come from a hinterland to a world-wide stage.  Unscintillating (talk) 13:06, 7 August 2011 (UTC)

I would take such a translation to be evidence that an assertion is verifiable, but not verified, and would suggest that they would be useful in AFD, or on the talk page of the article to help others help find a decent translation, but I would be very hesitate to accept such as cited references in the article. Of course, it would depend on how bad the translation is, but comes a point that an editor who does not have the language in question cannot evaluate the quality of the translation. Here I speak from the experience of rereading works in various languages after having improved my skills--my early understanding from 2-3 years of instruction of such were very poor. --Nuujinn (talk) 14:36, 7 August 2011 (UTC)

I certainly agree that it's irresponsible to add material to the encyclopaedia based on a machine translation of a language you do not speak.—S Marshall T/C 17:24, 7 August 2011 (UTC) I don't think machine translation can ever be considered a reliable source. Indeed, the meaning can easily be the opposite of what the original said. If the person providing a machine translation is in a position to check that the translation is adequate, it effectively becomes his/her translation, and the fact that it is (based on) a machine translation needs only be mentioned for the purposes of intellectual integrity. If he/she is not able to verify the correctness of the translation, it should not be used. The editor introducing the translation must assume responsibility for its accuracy, preferably explicitly. Perhaps we could change the text to something like

Translations published by reliable sources are preferred over translations by Wikipedians. If the translation is not the editor's, it must be attributed. If the translation provided by the editor is substantially the result of machine translation, this should also be mentioned, but an editor should never provide a machine translation unless he/she is in a position to confirm its accuracy.

--Boson (talk) 19:25, 7 August 2011 (UTC)

Machine translations, even between major languages, can come up with just the opposite of original meaning of a text. It's better to leave a text untranslated, because a machine translation conveys the appearance of accuracy, and it's even worse if the machine translation is being smoothed by someone who does not know the original language. We may look at the issue in five years again, and maybe machine translators have made further progress until then. In the meantime, I suggest that we advise against using machine translations, and that an editor must take personal responsibility for all translations. An editor who can confirm the accuracy of a machine translation will also be able to translate the text without the help of a machine translator. I also don't think it would be helpful to indicate whether an editor used a machine translation, as it would also not be helpful to indicate whether the editor used a dictionary or not.  Cs32en Talk to me  11:33, 8 August 2011 (UTC)

It depends on the language in question. I notice, without any surprise, that every single person opposed to machine translations, and who identifies a language on his or her user page, speaks German. I've seen some truly lousy machine translations from German (double negatives seem to be a real problem). On the other hand, I've never seen a really poor machine translation from Spanish. It's often a bit awkward, but it's never the opposite of the original. You're basically proposing that nobody be allowed to use a tool merely because it doesn't work so well for your particular language.

And, again, I think you're overlooking the context. This is primarily about providing a translation when you're using the |quote= parameter in the {{Citation}} templates. You'd be supplying both the original language and the English translation, and any person who read the language could correct or improve the translation at any time. If someone cited a source as supporting a statement, based on a machine translation or a plain old misreading, then we need to know this. "Showing their work" by providing the mistaken translation helps us find and correct errors. WhatamIdoing (talk) 16:45, 9 August 2011 (UTC)

Google translate does fairly well with Romance languages. It's not to be trusted for others.—S Marshall T/C 16:53, 9 August 2011 (UTC)

And so you would issue a universal ban. That does not seem appropriate to me, nor does it seem like it reflects the actual practice of the community. WhatamIdoing (talk) 22:33, 10 August 2011 (UTC)

What I'm saying is that from my personal experience, I would trust google translate (but not babelfish) to turn French into tolerably accurate English. I would not trust google translate with German or Dutch, and I definitely wouldn't trust it with something highly inflected like Finnish. If in doubt, I would have to reject the machine translation unless confirmed by a human speaker.—S Marshall T/C 22:51, 10 August 2011 (UTC)

I am German, but I almost never have used translations into German. Translations to German are so bad that I - at least for the last two years - have always used translations from various languages (Spanish, Italian, Japanese) to English. My comments are based on my experience with machine translations into English and have nothing to do with the German language.  Cs32en Talk to me  23:41, 10 August 2011 (UTC)

Under NONENG, all translations are to English, not from English. There is never any excuse to translate something into any language except English under NONENG.

WP:V and NONENG are not the general translation guide. NONENG is about what to do when somebody cites a Polish paper to support some claim in an article, and nobody knows for sure if the Polish paper says anything even remotely like what the original editor claims it says. NONENG says that in such cases, it's best to get a professional translation, second-best to get a Wikipedian to translate, and if all else fails, you can ask Google translate to have a go at the title—and that having that machine translation is better than leaving a citation in a state that is unintelligible to 99% of the world's population.

NONENG always moves to English. You would never translate anything into German under NONENG. WhatamIdoing (talk) 19:51, 12 August 2011 (UTC)

Of course, we are talking about translations to English. If "nobody knows for sure if the Polish paper says anything even remotely like what the original editor claims it says", then, with a machine translation only, still nobody knows whether the Polish paper says exactly what the editor claims it says. It is not sufficient to know whether the papers says something remotely similar to the content of our article, and therefore a machine translation is not helpful in such cases.  Cs32en Talk to me  20:04, 12 August 2011 (UTC)

A proposed rewording of WP:NONENG

Thank you everyone. Maybe it's impossible to reach consensus but I'd like to try. I have the impression that most people favour a change. I don't feel there's much point in trying to formulate fine-tuned rules about the translation process; it's really about whether the editor has the language skills to take full personal responsibility for the translation. Based mainly on Boson's suggestion, but with some alterations, this is a proposal which I think is in line with the majority view:

• Translations published by reliable sources are preferred over translations by Wikipedians.
• A Wikipedian may only use a machine translation if he/she speaks the source language and is able to confirm the accuracy of the translation.
• All translations must be attributed to the translator by name.

Do you agree or disagree? Rubywine . talk 20:53, 9 August 2011 (UTC)

I disagree with everything except the first point. People add information based on machine translations for languages they don't understand (well enough) every single day. If they paste in their machine translations, we increase the likelihood of finding their errors. We also use machine translation for purposes far outside the subject of this policy, e.g., to decide whether the person who says "this five-page magazine article in Chinese is entirely about this company" is telling the truth. That's a perfectly legitimate use of machine translation, since even the most busted machine translation is going to reliably identify whether the name of the company appears repeatedly throughout the five pages. This sloppy statement would ban any and all use, not just uses that are higher risk.

Also, the names of the editors who did the translation have no business appearing in the article under any circumstances. WhatamIdoing (talk) 21:00, 9 August 2011 (UTC)

• Contributor names are always recorded in the usual Wikipedian way, of course. If the translator was a reliable source, then it's not just reasonable but strongly encouraged to attribute them, since not attributing them is potentially plagiarism or breach of copyright. As for your second point, who will police this and who will enforce it?—S Marshall T/C 22:29, 9 August 2011 (UTC)

• Yes of course. On my second point, who can stop dissenting translators, determined to foist unassisted machine translations upon the unsuspecting world? I don't know. The existing WP:NONENG guidelines can't be policed or enforced. Most other guidelines can't. The system assumes that the majority respect consensus. Rubywine . talk 01:03, 10 August 2011 (UTC)

• The existing policy is enforceable by individual editors on talk pages (with backup from uninvolved administrators if necessary). The proposed version would require editors to know which language other editors can speak, before being enforceable.—S Marshall T/C 07:14, 10 August 2011 (UTC)

• No it doesn't, it only requires that other people can speak the source language. If an editor's submissions were found to match unassisted machine translations, and were deemed to be incompetent by a consensus of competent speakers, then they could be asked to stop. That's all. It is unlikely that this scenario would arise. Rubywine . talk 12:18, 10 August 2011 (UTC)

• That means that we have to decide who's "competent" to pass judgment. It would be far simpler for these competent people to simply fix bad translations when they find them, in exactly the same way that I fix bad grammar when I find it.
  Given how few pages I see with any non-English content, I find it hard to believe that we have a widespread problem with seriously misleading machine translations in our articles. This issue may not be worth the time we're spending on it. WhatamIdoing (talk) 17:48, 10 August 2011 (UTC)

• We don't need to decide who is competent. Firstly, because consensus about correct use of language is implicit in every discussion we have on Wikipedia, and there's no real difference here. If anything, it's easier than usual. Any human being can recognise a bad machine translation of a language they speak; there's nothing subtle about it. Secondly, because the enforcement scenario is a red herring. The point of this guideline is not to enforce and police, the point is to guide. There have been instances of bad translations being submitted, which have been mentioned by translators, and that is avoidable and regrettable. It's clear that most people in this discussion say there is no place for unassisted machine translations on Wikipedia. We just need to reach a consensus on the wording of guidance to that effect. Regarding your point that there are few pages with any non-English content - I think that is likely to change. There is a vast quantity of non-English material which has been requested for translation. This issue will be increasingly relevant in the future. Rubywine . talk 22:06, 10 August 2011 (UTC)

No, if you're going to have "a consensus of competent speakers", then you have to decide who qualifies as a "competent speaker".

Requests for translations of articles from other Wikipedias has nothing to do with this policy. NONENG is not about whole articles. NONENG is about what to do when somebody cites a Polish paper to support some claim, and nobody knows for sure if the Polish paper actually says what the original editor claims it says. What you need to do is explain why omitting the translation of the paper's title, on the grounds that we don't know how it was translated, would improve Wikipedia. WhatamIdoing (talk) 22:31, 10 August 2011 (UTC)

• Fair enough re NONENG. But I'm not going to be dragged into further discussion of a red herring, or repeat myself. Rubywine . talk 22:38, 10 August 2011 (UTC)

I suggest two changes:
(a) Change "is able to confirm the accuracy of the translation" to "confirms the accuracy of the translation". Maybe a comment can be added that this can be done either on the talk page or in the edit summary.
(b) Translations should not be attributed to the name of editors, but they should be attributed to real world sources if such sources have been used. If not, as with all other content, the attribution is in the history of the page, plus the confirmation on the talk page (or talk page archive) and/or the edit summary.  Cs32en Talk to me  23:48, 10 August 2011 (UTC)

Thanks, I agree with your suggestions and I'm happy to adopt them. In the absence of an explicit RS attribution it would be implicitly understood that the editor was the translator. I'll leave this a bit longer in case anyone else wants to comment. Rubywine . talk 22:13, 11 August 2011 (UTC)

A proposed rewording of WP:NONENG #2

The revised proposal is to replace this sentence in WP:NONENG :

• Translations published by reliable sources are preferred over translations by Wikipedians, but translations by Wikipedians are preferred over machine translations.

with the following :

• Translations published by reliable sources are preferred over translations by Wikipedians, and should always be attributed. A machine translation may be used only if the Wikipedian speaks the source language and confirms the accuracy of the translation.

Footnote: Attributions and confirmations may be provided on the talk page or in the edit summary.

Rubywine . talk 16:41, 14 August 2011 (UTC)

When we say that something "should always be attributed", we normally mean that it requires WP:INTEXT attribution. Your first sentence is therefore going to confuse a lot of people. Also, the name of the translator is commonly unknown and therefore cannot be "always" provided in any form. Furthermore—who cares who the translator is, except for the translator himself (who presumably would like to build his career)? If the translation is published by a reliable source, then we don't need to say "Translation provided by Bob Smith, who was hired by The London Times to provide a translation for this phrase in their article."

I firmly disagree with the second sentence. A machine translation may be used for many purposes that have nothing to do with verifiability, and your proposal fails to acknowledge that. Furthermore, as admitted above, machine translations aren't always bad, especially for Romance languages. Finally, since WP:Nobody reads the directions, and this rule is directly counter the community's actual practice, it's both ineffective and inappropriate. Written policy must describe the actual policy, and the actual policy is determined by what the community actually does, not by what a couple of people wish they would do instead. (It would be appropriate to say "Be cautious in using machine translations, which are lousy for many languages".) WhatamIdoing (talk) 23:18, 14 August 2011 (UTC)

"Be cautious in using machine translations, which are lousy for many languages" would be a step forward. The rest is ground we've been over before. WP:Nobody reads the directions is a weak argument. People try to read guidelines; long pages of tortuous prose get ignored because they're taxing and not understood. WP:NONENG is not a remotely arduous read, so instruction creep doesn't really apply here. If guidelines were so uninfluential, so universally ignored, you wouldn't spend so much time arguing over them! Rubywine . talk 03:05, 15 August 2011 (UTC)

Holy cow do I have a lot to say on this subject... I'll try to be concise. It touches on several different translation issues, actually. Some exclusive to Wikipedia, some not.

• Machine translation: I do make use of Google's app, which does make readily recognizable glitches. At times producing a translation that is actually the complete opposite of what the original text is saying. Yes, I am fluent enough to recognize this, and I still find the translation apps useful. If I look up a single word, and it doesn't suit the context of the original text, I know enough to do more searching with other tools, look in a thesaurus, etc.
• Professional translation is preferred: Did someone in WP:NONENG say that a professional translation is preferable? I do translation work in my capacity as a professional with the federal gov't of Canada. I've been tested and assigned a formal evaluation of my abilities. Do I qualify as a 'professional translator' by Wikipedia standards? (This is in my second language, French, so I do consider myself qualified to do French -> English translations, and vice versa, up to a level that includes technical language in a handful of fields. I've done editing for an academic journal in both languages, years ago.)
• A specific example: I just did a translation of a biography article (Anita Aubspurg) that appears in the French, Italian, and German Wikipedias. The original article appears to be the German version, all the references are to German books/Websites. The Italian version just copied and pasted the German references, and that's what I did too. No, I don't know German. My Italian isn't as good as my French, but I can read it pretty well. (The French version was all but empty, they didn't translate the Italian or German article in full at all). I used a translation app on both the Italian and German versions of the article, compared them side by side, and edited the article so that it would be coherent in English.
• Citing sources: I've been reading quite a few French Wikipedia articles, and glancing through some Italian and German, and guess what? Articles using inline cites to references are in a small minority. Who's enforcing the Wikipedia reference rules for the non-English Wikipedias? If the articles are acceptable in those Wikis without inline citations, why are they not acceptable in the English one? Should the articles in the non-English Wikis be purged if they do not meet that standard?
• Citing sources, part 2: No, I can't verify the original source texts used by the original article author. But then, I can't do that for an English article if it's citing books, journals, etc. that I don't have access to. I could read the original French books if I had access to them, but if all the copies are in France, or Senegal, or Cote d'Ivoire, or Tunisia, what should I do then? What would you do if you were editor for an English language article that was citing documents in the library of Oxford University, but you live in California? Does the article get purged? If the original article is on a subject for which information only exists in publications in a language for which English translators are rarely available, is the consensus that it not be translated at all? FWIW, I do try to look for some equivalent English sources online to add to articles I've translated, but for many subjects, there's just no information in English. (Doesn't mean the topic isn't noteworthy.)
• My two cents' worth -- I think it comes down to a question of reasonable due diligence. When you read an article in Wikipedia, or the New York Times, or a paper encyclopedia, you do take a certain leap of faith that the article isn't complete fiction with imaginary sources. But I think if an article contains inaccuracies, the open nature of Wikipedia is that it will be found out and either removed or corrected by someone more knowledgeable.

OttawaAC (talk) 01:24, 15 August 2011 (UTC)

OttawaAC, thanks for your comments. There's an awful lot of context here that you'll need to get a handle on. Please read through the previous parts of this discussion topic. This is just a quick reply to your point number 2. In this context, reliable source doesn't mean professional translation; a Wikipedian can't be recognised as a professional translator. A reliable source on Wikipedia means a published, third party, reliable source. Of course you could cite your own externally published work, but only in the same way that you could cite someone else's. Rubywine . talk 02:17, 15 August 2011 (UTC)

OttawaAC, in theory, NONENG has nothing at all to do with interwiki translations... but you wouldn't know that from Rubywine's proposal. Rubywine's proposal would ban the use of machine translation for any and all purposes, even for figuring out if a long article that is alleged to demonstrate notability in an AFD discussion actually talks about the subject. I do not know whether this is what Rubywine intended, but it is what s/he actually hopes to put in this policy. WhatamIdoing (talk) 02:44, 15 August 2011 (UTC)

WhatamIdoing, this is ridiculous. You are misrepresenting my intentions and the scope of the policy. Tiresome. Rubywine . talk 03:14, 15 August 2011 (UTC)

Yes, the proposal as written is ridiculous. Writing policies is hard. If you say "no machine translations unless you understand the language", then people will believe that you actually mean "no machine translations unless you understand the language", full stop. They will not magically know that you mean "no machine translations unless you understand the language, and, by the way, this only applies to direct quotations in the text of the article, because discussing sources on the talk page, translating whole articles from other Wikipedias, and figuring out whether a non-English source indicates WP:Notability in a deletion discussion are outside the scope of this policy, and we know that Wikipedia:Policies and guidelines#Content demands that we maintain the scope of the policy without wandering into off-topic stuff, like what's okay at AFD", which is what you probably mean to say.

Writing policies is much harder than you think it is. WhatamIdoing (talk) 20:26, 15 August 2011 (UTC)

I have amended the text of the proposal to rule out this plethora of unlikely misunderstandings and threats to personal liberty. Rubywine . talk 03:43, 16 August 2011 (UTC)

I actually don't think that the amendment is necessary, because I don't think editors will fear getting blocked for posting a machine-based translation on a talk page. I support both the amended and the original version of the proposal.  Cs32en Talk to me  19:23, 16 August 2011 (UTC)

(original version=#2, amended version=#3)

For the sake of continuity I have restored the original text of #2, have moved the amendment to #3, and have copied Cs32en's comment below. Rubywine . talk 22:11, 16 August 2011 (UTC)

A proposed rewording of WP:NONENG #3

Do you support the proposal to amend the guidance in WP:NONENG regarding the use of machine translations, as given at WP:V:talk?

Please note that the scope of WP:NONENG is limited to the translation of non-English sources for use in English Wikipedia.

The proposal is to replace this sentence in WP:NONENG :

• Translations published by reliable sources are preferred over translations by Wikipedians, but translations by Wikipedians are preferred over machine translations.

with the following :

• Translations published by reliable sources are preferred over translations by Wikipedians, and should always be attributed. A machine translation may be used in the text of the article only if the Wikipedian speaks the source language and confirms the accuracy of the translation.

Footnote: Attributions and confirmations should be provided on the talk page or in the edit summary.

Rubywine . talk 03:56, 18 August 2011 (UTC)

I actually don't think that the amendment [to your previous proposal] is necessary, because I don't think editors will fear getting blocked for posting a machine-based translation on a talk page. I support both the amended and the original version of the proposal.  Cs32en Talk to me  19:23, 16 August 2011 (UTC)

• (where the original version=#2 without the bolded text, and the amended version=#3) Rubywine . talk 22:56, 16 August 2011 (UTC)

No, they won't fear getting blocked, because violating WP:V almost never results in blocks. But we would have some tendentious editors who use such unqualified statements for obstruction. This change dramatically reduces my concerns about including this sentence.

Rubywine, are you prepared to explain what "should always be attributed" means now? WhatamIdoing (talk) 23:43, 16 August 2011 (UTC)

I'd like to remind you that this is an open discussion, and not a 1 on 1. The idea came from Boson. It was strongly supported by S Marshall, and then phrased as an amendment by Cs32en. I have reworded it more concisely, and nobody else has objected. It is clear and unambiguous, since the footnote rules out the possibility of in-text attribution. It won't confuse anybody. Rubywine . talk 01:12, 17 August 2011 (UTC)

The footnote does not rule out anything at all. It rules in additional options. WhatamIdoing (talk) 16:33, 17 August 2011 (UTC)

I have never seen an inline attribution on a reference, i.e. a nested attribution, and I doubt that it is possible to create one. Even so, I have changed the wording so that the only options are talk page or edit summary. Rubywine . talk 17:11, 17 August 2011 (UTC)

• Support: I think the new version is better, but I don't think there will be much difference in practice between the two. I suggest that "speaks the source language" be replaced with "is familiar with the source language". People can sometimes read a language but not be able to speak it. For example, Latin and Classical Chinese, are written languages that few can speak outloud. --LK (talk) 10:11, 18 August 2011 (UTC)
• Weak support If a wikipedian speaks the source language, why not translate it him/herself? That said, I see no problem with allowing the use of machine translation. Perhaps a friendly ammendment, and if it isn't I won't make a big fuzz: still say that a translation by a wikipedian is better? This is to encourage the translators among us to do it and be recognized.

Also, I might misunderstand, but I think what you are saying can be achieved with Template:Ref series templates, and I have seen done - even to borderline copyvio.--Cerejota (talk) 11:05, 18 August 2011 (UTC)

Notability section

Notability section, tweaking

[subsection inserted above] Unscintillating (talk) 03:04, 13 August 2011 (UTC)

Does anyone mind if I tweak the writing of the part in bold? Or do we even need it? I had difficulty understanding it when I first read it.

"Wikipedia's General Notability Guideline says that the subject's notability is established by the publication of independent, reliable sources, regardless of whether the names of any such sources have yet been typed into the article.

I assume it means notability depends on whether published sources exist, not whether they've been added to the article. But that's what the first part of the sentence says already. SlimVirgin ^{TALK|CONTRIBS} 17:00, 12 August 2011 (UTC)

I also believe that's what it means. Considering this is a summary of another policy, if it is to be an actual convenience to readers, it should include the critical points from the other policy. It is a critical point that articles should not be nominated for deletion if nearly everyone knows the topic is notable, even though the independent sources are not cited. Jc3s5h (talk) 17:20, 12 August 2011 (UTC)

In the recent edit history of the page, it went from this to this. What would the "tweak" consist of? It might be best to put the proposed wording here, in talk, first. --Tryptofish (talk) 18:58, 12 August 2011 (UTC)

I was going to tweak it, but then realized it doesn't need to be there in the first place. The first part of the sentence says: "the subject's notability is established by the publication of independent, reliable sources." The bolded second clause just repeats that, but in an odd way. No one would assume that, if I created a stub that said only, "Queen Elizabeth II is the queen of England," that the lack of sources on the page at that moment in time would mean Queen Elizabeth wasn't notable. SlimVirgin ^{TALK|CONTRIBS} 19:07, 12 August 2011 (UTC)

I'd like to hear from the editor(s) who made the change about why they wanted to add that wording, before making a decision. I agree that it sounds odd, as written, but I have a feeling (ie, guess) that the idea was to point out that sources can exist even if they haven't been added to a page, so a page shouldn't be deleted without checking for sources first – as though these editors have had experiences where your hypothetical Elizabeth II stub was nominated for deletion (and stranger things do happen). --Tryptofish (talk) 19:16, 12 August 2011 (UTC)

I agree that it's redundant, but AFD has significant and persistent difficulties with people who do not (choose to) understand this relatively simple point. Repeating it seems to get the point across more effectively than stating it once. The point that needs to be made is that {{unref}} and {{nonnotable}} are completely separate issues. If you want more formal language (which I've been finding less effective recently), then something like "Notability requires only that suitable independent, reliable sources have been published; it does not require the citation of any sources at all in the article" would probably suit.

There was discussion earlier about removing the entire section, since there's no particular reason to for WP:V to say anything about all about WP:N. WhatamIdoing (talk) 19:42, 12 August 2011 (UTC)

I would prefer to go back to the previous text, which simply said: "If no reliable secondary source can be found on a topic, Wikipedia should not have an article on it." That's the only part of the issue that concerns V. It was expanded on August 7, [1] and again on August 8. [2]

If we can't have the old version, I'd like to tweak the new sentence to say: "Wikipedia's Notability guideline says that a subject's notability is established by the publication of independent, reliable sources, not by whether such sources have actually been added to the article." SlimVirgin ^{TALK|CONTRIBS} 20:20, 12 August 2011 (UTC)

Two thoughts... 1) Does not the existence of a tertiary source on the topic (such as an article on the topic in some other encyclopedia or almanac) go towards notability? (if so, we should avoid the word "secondary"). 2) There is a difference between "I can't find any sources" and "No sources exists".

I think the point of this section is to tell editors: "Don't create an article if you (the creating editor) can not find any sources that discuss it"... but it is being misunderstood as saying "I should nominate an article for deletion if I can't find any sources about it". The first is correct... the second isn't. Blueboar (talk) 21:03, 12 August 2011 (UTC)

Well, first of all, the original is one diff earlier, and it says not one word about secondary sources. It refers to WP:Third-party sources (aka independent sources), not WP:SECONDARY sources, and as you know, WP:Secondary does not mean independent. A first-party secondary source (and there are many in the hard sciences) does not demonstrate notability.

"Can be found" has unfortunately been interpreted as "can be found by lazy deletionists who refuse to look any further than a section named ==References== in the current version of the article". I suggested "have been WP:Published" last month, and the discussion was de-railed by a long digression by one editor who wanted to ban unref'd articles entirely, and the distraction of the endless not-truth discussion.

Similarly, I'm concerned that "whether such sources have actually been added to the article" may be more susceptible to misunderstanding than "whether someone has typed the names of such sources into the current version of the article". I had a long and discouraging conversation last month with an experienced editor who seemed to have trouble distinguishing between adding content from a published book (e.g., typing "Only 5% of women die from breast cancer" into an article) and actually WP:Citing the book (=typing the author's name, the title, and the date into the article), so my belief in the average editor's ability to understand plain English is at an all-time low. Someone might well think that "adding sources to the article" referred to including the full text of primary sources rather than to typing up the author, title, date, and publisher for any source that you used. WhatamIdoing (talk) 21:17, 12 August 2011 (UTC)

For years it said "third party sources," and that would be my preference. But someone recently changed it to secondary sources, and I don't mind that either. The only concern I have is extending it to say things covered by the guideline. SlimVirgin ^{TALK|CONTRIBS} 21:22, 12 August 2011 (UTC)

The long-standing version was "If no reliable third-party sources can be found on a topic, Wikipedia should not have an article on it" (e.g. here). I also wouldn't mind: "If no reliable secondary or tertiary sources can be found on a topic, Wikipedia should not have an article on it." But I don't see the need to say more than that here. The details are dealt with in the Notability guideline. SlimVirgin ^{TALK|CONTRIBS} 21:25, 12 August 2011 (UTC)

I think these comments by both Blueboar and WhatamIdoing are very helpful to me in understanding the issue. Perhaps the key issue, then, is the wording about whether WP "should" have an article on a subject. That does, indeed, sound like an invitation to lazy AfD. Instead, maybe the wording should be about whether "content" (as opposed to article) should be "added" (as opposed to exist), with the "added" part what is most important. What I mean is to frame it in terms of adding material, as opposed to responding to material previously added. --Tryptofish (talk) 21:25, 12 August 2011 (UTC)

I think we should go back to third party sources, as the issue of the independent coverage is pretty essential. And I concur with Tryptofish's last observation. --Nuujinn (talk) 15:11, 13 August 2011 (UTC)

How about: What matters is whether such sources exist, not whether the article presently cites them.? Septentrionalis PMAnderson 21:45, 12 August 2011 (UTC)

So, delete the bolded part above, and add that sentence in its place? I like that. I think it's a good idea. --Tryptofish (talk) 19:54, 13 August 2011 (UTC)

Notability section, WP:V is content policy

WP:V is content policy, see WP:N#NNC.  If the purpose of the notability section in WP:V is to repeat parts of the WP:N guideline, the section can be deleted.  Blueboar has mentioned that "the language...originally...was meant to relate to how the concept of WP:Verifiability could be applied at the article level (as opposed to a sentence or paragraph level)."  Given the burden of deletion at Wikipedia, I think we need to be moving in the direction of encouraging more sourcing by article creators, which in the absence of new proposals is best done by leaving the current language.  Unscintillating (talk) 03:04, 13 August 2011 (UTC)

Its a guideline not a policy. Lets not mix the two up. I say eliminate that section altogether. Its also misleading because it only covers the GNG, when if you click on that link it says an article is notable if it meets the GNG OR one of the secondary guidelines listed on the right. Dream Focus 03:43, 13 August 2011 (UTC)

Two points: first, the policy now says that the GNG says something that I can't find it saying. Does anyone have a link to where the guideline says this: "that the subject's notability is established by the publication of independent, reliable sources (regardless of whether the names of any such sources have yet been typed into the article)"?

Secondly, a bit of historical context. This policy contained the sentence about "no third-party/secondary/reliable sources = no article" in some form since at least Feb 2006, six months before the Notability guideline was created. So this sentence of ours—"If no reliable third-party sources can be found on a topic, Wikipedia should not have an article on it"—[3] is a verifiability issue, entirely independent of whatever the Notability guideline might say. We therefore shouldn't say anything here that implies this policy follows that guideline. That's why I think we ought to return to that one stand-alone sentence without elaborating. SlimVirgin ^{TALK|CONTRIBS} 18:11, 13 August 2011 (UTC)

I was wondering where that sentence got to - that's a very critical sentence that outlines fundamental verification policy. That needs to be put back in. --MASEM (t) 21:07, 13 August 2011 (UTC)

That a notability issue was described here when no other suitable page existed does not mean that "we should not have an article" is actually a matter of verifiability rather than notability. It could merely be a legacy of what seemed convenient at the time.

I do not believe that this sentence deals with verifiability. Material can be 100% verifiable without coming from a third-party source—exactly as this policy has said for years. WhatamIdoing (talk) 21:14, 13 August 2011 (UTC)

The point of the sentence is the can be found part. It's find to use first-party, dependent sources as long as they are reliable (SPS comes into play here, for example), as long as we know there also exist third-party sources that discuss the topic; the first party source may be the best source to use in that case (such as using the award organization's list of winners on their webpage than a newsreport reporting the same - the key is that the newsreport has shown interest in it. Without any third-party sources in existence, a topic has no relevance to anyone but those directly involved with the topic, and thus there's appropriateness for an article on that topic. That's a key aspect of verifyability, that someone else has at least considered the first party source as factually correct. --MASEM (t) 21:58, 13 August 2011 (UTC)

I am beginning to think that we made a mistake in trying to tie the sentence in question directly to the WP:Notability guideline. I think the original intent of the sentence was to say:

• If no Verifiable information exists on a topic, Wikipedia should not have an article on that topic.

While that idea is similar to the concept of notability, it isn't quite the same. Blueboar (talk) 13:44, 14 August 2011 (UTC)

Yea, I have no idea why the idea was tied to notability. "Third party" source is one, by definition, would have had to do some research to validate the facts since they were not privy to the first-hand details; the more reliable that source is, the more fact-checking they have likely done. Requiring that some third-party sources exist is a basic metric for verification of a topic, and has little to do with notability which is more about how deeply that topic is covered in sources as to make for a good encyclopedic article. --MASEM (t) 14:00, 14 August 2011 (UTC)

That sentence was never meant to be tied directly to the Notability guideline. It spent a lot of its life under Burden of evidence. I moved it not long ago to a subsection called Notability under the "Verifiability and other principles" header, but the intention was not that it derived from the Notability guideline. Perhaps we should simply move it back to Burden of evidence to break that connection. SlimVirgin ^{TALK|CONTRIBS} 18:49, 14 August 2011 (UTC)

Notability section, proposal to put sentence back under "Burden of evidence"

The proposal is to delete the notability section and move the sentence back to the "Burden of evidence" section.  The sentence is:

“ If no reliable third-party sources can be found on a topic, Wikipedia should not have an article on it. ”

• Support  Unscintillating (talk) 22:49, 14 August 2011 (UTC)
• Support retaining that sentence and not elaborating further about notability; and I have no objection if it's moved back to the "Burden of evidence" section, and the Notability subsection is deleted. SlimVirgin ^{TALK|CONTRIBS} 22:54, 14 August 2011 (UTC)
• Aye.—S Marshall T/C 22:56, 14 August 2011 (UTC)
• Moving it works for me. I still strongly prefer that "can be found" be replaced by "have been WP:Published", to alleviate the problems that AFD is reporting with a small number of editors who apparently want to believe that "can be found" means "can be found without me needing to look any further than the ==References== on the current version of the article". WhatamIdoing (talk) 23:07, 14 August 2011 (UTC)

• Oppose. The only interpretation of "topic" for which this statement is correct (never mind which policy it belongs in) is the article as a whole. But it would be too easy for people to interpret as the topic of a specific addition, or a topic that is broader than the article. Also, the statement only belongs in the notability policy, or a summary of notability in this policy. It is not a statement about verifiability. I could give a perfectly verifiable statement about the location of a state highway culvert, from a first-party source (the state highway department). The reason for not having such an article is that nobody cares, not a lack of verifiability. Jc3s5h (talk) 23:16, 14 August 2011 (UTC)
  • It has nothing to do with notability, that's the thing. A lack of third-party sources means we have no assurance on the reliability and verification of the information on the topic, and thus should limit its coverage from "none" to "part of a larger article". --MASEM (t) 23:20, 14 August 2011 (UTC)

• A Supreme Court decision is a first-party source, but is absolute proof of the decision. It is certainly verifiable, although there are many reasons, which fall outside this policy, why reporting just the decision would make for a bad article. Jc3s5h (talk) 00:02, 15 August 2011 (UTC)

• But at the same time, a SC decision will be covered by third-party sources as well, even though we'd likely to use the actual decision for sourcing statements about that decision. The issue is when only first-party or second-party sourcing is available. --MASEM (t) 00:30, 15 August 2011 (UTC)
• A SCOTUS decision is not a first-party source for everything. It is a first-party/affiliated source for the new decision it contains, but it is is a third-party source for (e.g.) any prior court case the decision describes, laws it mentions, facts relevant to the specific case, etc. WhatamIdoing (talk) 20:33, 15 August 2011 (UTC)

• No, SCOTUS is a part of the US government, so is a first party source for its own decision, and any federal court decisions, quotes from federal prosecutors, federal legislation, etc.

• No, that's not enough to make them first-party. SCOTUS is a first-party, primary source for what SCOTUS says; it is a third-party/independent source for what some independent branch of the federal government, i.e., the Executive or Legislative branches, says. And it's certainly not "affiliated" or first-party with respect to the claims put forward by the opposing sides in the case. When they wrote in Brown v Board that "This segregation was alleged to deprive the plaintiffs of the equal protection of the laws under the Fourteenth Amendment", SCOTUS was not a first-party source for that material. (Primary, yes. Affiliated, no.) WhatamIdoing (talk) 02:09, 16 August 2011 (UTC)

• I would say an appellate court is a secondary source when describing the evidence placed before a lower court, and it's a primary source when discussing the arguments made before it, and its own decisions. SlimVirgin ^{TALK|CONTRIBS} 02:14, 16 August 2011 (UTC)

• If it merely repeats what was said by someone else, then it is not a secondary source. It's not merely a matter of counting up chains in the link. If I quote Masem, and you cite me, then that's still primary material, even thought your step is third-hand. WhatamIdoing (talk) 22:47, 16 August 2011 (UTC)

• Strong support - the removal of this line drastically changes how WP:V is to be applied. --MASEM (t) 23:21, 14 August 2011 (UTC)

• That sentence is about notability, not verifiability, so it does not belong in WP:V. — Carl (CBM · talk) 02:35, 16 August 2011 (UTC)
  • The sentence is not about notability, it is about the existence of third-party sources to provide an unbiased and reviewed facts about a topic, needed for WP:V. --MASEM (t) 03:04, 16 August 2011 (UTC)
    • Here is what the first paragraph WP:Notability says about the sentence.

"... if no reliable third-party sources can be found on a topic, then it should not have a separate article. Wikipedia's concept of notability applies this basic standard to avoid indiscriminate inclusion of topics."

Bob K31416 (talk) 15:32, 16 August 2011 (UTC)

• • • Third-party sources are only needed to show notability. An article written using entirely first-party sources (e.g. a biographical article using only a published autobiography by the subject of the article, or an article about a scientific experiment sourced only to a reliable, peer-reviewed paper where the experiment was published) could meet WP:V without meeting WP:N. There are many reliable first-party sources (and unreliable third-party sources). — Carl (CBM · talk) 15:53, 16 August 2011 (UTC)

• Such an article would violate V, because of the lack of at least one reliable third-party source. The point of requiring this is to make sure someone other than the primary sources (the authors) deem the material worthy of comment; the requirement is also in place to avoid OR. The requirement has been in this policy (a core content policy) for years, since before the Notability guideline existed, and was regularly acted on. SlimVirgin ^{TALK|CONTRIBS} 19:17, 16 August 2011 (UTC)
  • No, WP:V has nothing to do with "third-party" sources, that is purely the domain of WP:N. It may be that this policy had language about notability before WP:N existed, but that is years in the past and no longer relevant. Now that WP:N does exist, we have a distinction in policy between verifiability and notability: it is possible to have a verifiable article on a non-notable topic, and an unverifiable article on a notable topic. These independent considerations are covered by separate policies, and language about notability doesn't belong here. — Carl (CBM · talk) 19:46, 16 August 2011 (UTC)

• N is a guideline. V is a core content policy. It has been the case for as long as I remember that articles had to be based on reliable secondary sources. We are meant to offer an overview of the literature, and to do that a literature has to exist, at least in the minimal sense that a third-party source—someone other than the primary source and the Wikipedian who created the article about it—has deemed the issue worthy of mention. That's not just an N issue. It is a V and NOR issue, which is why this policy has included mention of it since at least 2006.

  You can't (by definition) have an unverifiable article on a notable topic, and I can't think what a verifiable article on a non-notable topic would be. SlimVirgin ^{TALK|CONTRIBS} 20:57, 16 August 2011 (UTC)

• Mostly, I agree with SV's analysis, but it is certainly possible to have a verifiable article about a topic which is non-notable for encyclopedic purposes. Millions of newspaper pages of "society events" and such drivel have been published, rendering the parties described verifiable. Some of these were even written by reliable journalists. But the parties would not on their own merit be notable topics for WP, would they? There needs to be a little more to it if we don't want Emma Smith's 1875 cotillion in New York City. LeadSongDog come howl! 22:17, 16 August 2011 (UTC)

WP:N is only a guideline, not a policy, because if a separate article is created on something that is non-notable but verifiable, no great harm is done (provided other policies, such as NPOV, are followed). Another way of saying this is that if a few facts that really belong as a heading in a broader article are instead a separate article, this is merely a guideline non-compliance rather than a policy violation. Jc3s5h (talk) 22:27, 16 August 2011 (UTC)

Apart from society events (like LeadSongDog mentions) another way to get a verifiable article on a non-notable topic is to take a paper in some academic journal, on a topic that has only ever been studied in that one paper, and write an article that directly summarizes the results of the paper. We often delete or merge articles like that for lack of notability, but nobody argues that they are unverifiable if every claim they make is literally present in the paper being used as a source. Regarding "You can't (by definition) have an unverifiable article on a notable topic, ...", that brings out the fundamental difference between WP:V and WP:N. N is about the article topic, independent of its actual content, while V is about the actual content, independent of the worthiness of the topic. These should be treated separately, which is why we have separate pages for them. — Carl (CBM · talk) 00:10, 17 August 2011 (UTC)

There's no point in tying ourselves in knots, because this isn't rocket science. If someone creates an article on "John Smith's fishing technques," it needs a source that isn't My Fishing Techniques by John Smith, or Smith's wife's blog. That is, we need a reliable, published third party—not John Smith and friends, and not the Wikipedian who created the article—to tell us (a) that this issue is worth mentioning, and (b) what kind of thing we should saying about it to avoid OR. That's the only issue that V has ever commented on. All the details belong in N. SlimVirgin ^{TALK|CONTRIBS} 00:27, 17 August 2011 (UTC)

We both agree that article needs another source besides that one. But I am saying that fact belongs in WP:N, not here, because it is not related to whether the material that is included in the article is verifiable. Knowing whether "the issue is worth mentioning" (that is, whether we should have the article) is entirely a WP:N issue. The other thing you mentioned seems like an NPOV issue. If Smith's book is the only source in existence, then (essentially by our wiki definitions) its viewpoint is the NPOV viewpoint on the matter, and descriptive claims taken from it are verifiable and not original research. My response to the RFC is that this policy should stick to verifiability, not repeat things that belong in other policies. — Carl (CBM · talk) 01:35, 17 August 2011 (UTC)

I do not agree, actually. Per WP:SELFPUB, a primary source is a perfectly good source to V itself and the basic facts about itself. I have a dictionary here. I don't need a third party source to tell me that it exists, nor what year it was published. I can cite the primary source itself. That's neither enough to 1) comment on specific controversies, awards, etc. or 2) demonstrate notability. But the "we need secondary sources to meet V!" mantra is just not correct. No matter what the particular wording of V is changed to, it doesn't make sense to say that a published work is not an authoritative source on its own existence and publication date. Jclemens (talk) 01:39, 17 August 2011 (UTC)

I just meant that article needs more sources to meet WP:N. I agree with what you said. — Carl (CBM · talk) 01:48, 17 August 2011 (UTC)

To take the example of "John Smith's Fishing Techniques", if the only sources are the book and the blog, we have way to consider if this source is even close to reliable, working on the assumption that John Smith here-to-date was a unknown person. Ergo, without any third-party addressing the book, we cannot even assure it meets the basic "verifiability, not truth" aspect. Maybe the entire book is a work of fiction published as a non-fiction title, maybe it plagiarizes Joe Jones's Fishing Techniques, we don't know. We'd not be able to verify that what John Smith has written actually took place; we'd only have John Smith's word on it, which is not sufficient. A reliable third-party source that would at least consider the work, determining if the account is legit or not but without otherwise introducing additional analysis (eg a primary third-party source) would at least provide something towards verifiability. Furthermore, a third-party source would still show the work existed even if only ten copies were made and destroyed without the original text surviving; without a third-party, in such a case, even if you can remember the general ideas of the text in your head, there's no way for any other user to validate it, ergo, it would fail. However, even for a published book, there is nearly always an ISBN number, and therefore some record of the book's existence in a third-party catalog (maybe LOC) even if that is just a primary source.

Note that this is nothing about secondary, transformative sources. Third-party != secondary. WP:N asks for secondary sources as a basis to build a good encyclopedic article. WP:V asks for third-party so that we know we're putting in facts that the reader can check themselves. --MASEM (t) 13:11, 17 August 2011 (UTC)

• Masem commented, "WP:V asks for third-party so that we know we're putting in facts that the reader can check themselves."
• The subject sentence is, "If no reliable third-party sources can be found on a topic, Wikipedia should not have an article on it."
• And here is the subject sentence in the first paragraph of WP:Notability, "... if no reliable third-party sources can be found on a topic, then it should not have a separate article. Wikipedia's concept of notability applies this basic standard to avoid indiscriminate inclusion of topics."

Bob K31416 (talk) 14:31, 17 August 2011 (UTC)

An ISBN catalog is not a third party source, nor is a library catalog. Those help Wikipedians locate the primary source, but that's essentially it. I could probably get an ISBN for my cat (if I had one, which I don't), and then get it cat-aloged (ha ha) somewhere, but those entries wouldn't prove the existence of a work: the ability to look up, purchase, check out, or otherwise acquire the primary source for verification is what's really happening, and catalogs and listings are merely aides to that end in this context. — Preceding unsigned comment added by Jclemens (talk • contribs) 02:00, 18 August 2011 (UTC)

Time to centralize discussion

STOP... it seems that there is a debate at GNG that is centered on what this policy says about notability.... and a debate here that is pointing to what that guideline says. I suspect that both pages are being improperly edited in order to skew the debate at the other page. Please, do not edit one policy or guideline in order to win a debate at some other policy or guideline page. Instead, we need to create a centralized discussion so editors on both pages can reach a consensus as to what should be said on both pages. Given this confusion... I am going to revert both pages back to their last stable versions while we centralize the discussion and reach such a consensus... I suggest that WT:Notability is the better venue for that discussion. Once we figure out what the notability guideline should say, then we can come back and make any edits to this page that are needed. Blueboar (talk) 20:07, 13 August 2011 (UTC)

Assume good faith! First off, different people are editing these articles. The debates for each page should be on that article's talk page, not mixed together. Dream Focus 20:29, 13 August 2011 (UTC)

OK... I suppose it is possible that it is not a deliberate thing... but the problem remains... both discussions are pointing to similar bits of language that exist in the other policy/guideline page. Neither page can have a meaningful discussion or resolution when both sections are being edit warred over. Blueboar (talk) 20:56, 13 August 2011 (UTC)

Policy question with regards to hurricane

We're having a bit of a debate over at the tropical cyclone Wikiproject. The question is whether we can use this source - the Atlantic hurricane database - for adding additional information to certain hurricane articles. The source provides the data points for every known Atlantic tropical cyclone in a fairly raw format. Would it be acceptable, for example, to use that and say a certain cyclone formed further north than any other? It is technically verifiable, but one would have to go through the entire database to prove it. Furthermore, such a statistic is not reliably published anywhere else.

Personally, I feel that it should not be included unless some other independent reliable source said it. However, others feel it should be allowed because it is technically verifiable.

Any input would be great. --♫ Hurricanehink (talk) 20:48, 12 August 2011 (UTC)

My two cents is that it isn't really a verifiability issue, so long as one can follow the link to that database and see the data in question. Instead, it's a matter of WP:SYNTH. I'd suggest that editors who are involved with the content try to decide whether or not it is synthesis, as opposed to common sense reading of the information, to conclude that it was further north than the others. (I have no opinion about that.) --Tryptofish (talk) 21:00, 12 August 2011 (UTC)

This kind of question should normally be posted at the WP:Reliable sources/Noticeboard. I've linked the FAQ at the top of the page so that future people will have an easier time finding the noticeboard. WhatamIdoing (talk) 21:23, 12 August 2011 (UTC)

Well, the source is certainly reliable. It's published by NOAA and the National Hurricane Cener. ♫ Hurricanehink (talk) 21:38, 12 August 2011 (UTC)

WP:RS really isn't the issue at all. --Tryptofish (talk) 21:40, 12 August 2011 (UTC)

What does going through the entire HURDAT database have to do with anything? I personally think it is not a WP:SYNTH, but the policy seems rather confusing to me.YE ^{Pacific Hurricane} 22:02, 12 August 2011 (UTC)

The issue is that we'd be evaluating the data in a primary source to reach a conclusion, eg. that a particular cyclone formed further north than any other. That seem to be OR to me, esp. since the data at the web site cite above is arcane and not easily understandable by the casual reader. --Nuujinn (talk) 22:17, 12 August 2011 (UTC)

How is it 0R? We have a HURDAT article that tells how to read the data set. YE ^{Pacific Hurricane} 23:12, 12 August 2011 (UTC)

(ec) While the data itself is in a machine-readable format, there are user-friendly viewers of the data, such as this one, which can be used to verify the facts included in the database. The question is whether the need for these decoders makes the original source inaccessible for verifiability purposes. I am pretty sure that looking at maps does not constitute WP:OR. Titoxd^{(?!? - cool stuff)} 23:14, 12 August 2011 (UTC)

The place for that discussion is WT:NOR. Since I'm here, I'll just say that I agree with Hurricanehink. If it's significant to the world outside of wikipedia that a particular cyclone formed farther north than any other, you should be able to find a reliable source that states that directly. If interpretation of binary data is required to support that claim then it suggests that this "fact" isn't of great importance outside of wikipedia. If that's the case, we don't need to test the limits of original research to state it either. It think it's fine to include facts drawn directly from that data source, (e.g., this cyclone formed at this point), but not draw conclusions based on what is really a primary source. If the cyclone data were in a sortable table the reader could use it for her own analysis and draw the appropriate conclusions herself. Quale (talk) 06:23, 13 August 2011 (UTC)

It's not a conclusion, though, it's a fact. Anybody with a first-grade education can decide whether "45N" is a higher number than other numbers in the same column. Juliancolton (talk) 14:43, 13 August 2011 (UTC)

I agree it is obvious. Why does it to be in a sortable table? Everyone knows that 11N is further south than 18N, or 135 knts is the highest number in all of the data in the month "7" ( or July).YE ^{Pacific Hurricane} 15:49, 13 August 2011 (UTC)

Er, sorry to display my ignorance, but I don't know that at all, I'm completely unfamiliar with the notation you all are using--I don't know what 45N, that 11N is further south than 18N, or what 135 knts means (but I can guess knots, is that correct?). And I have a masters. I guess you all went to better 1st grades than I ;). This really isn't the venue, but there are comparisons and judgments being made about data in a primary source. Quale said it better than I, to do anything beyond simply listing a datum is pushing the bounds of our policy on OR. --Nuujinn (talk) 15:58, 13 August 2011 (UTC)

(ec)WP:CALC allows certain operations being performed on the data "provided there is consensus among editors that the arithmetic and its application correctly reflect the sources".  You could reasonably make the argument that MAX is a mathematical function on the talk page.  If there are still objections, you could try to convince a reliable source to publish your findings.  Unscintillating (talk) 16:17, 13 August 2011 (UTC)

I do think that this fail under WP:CALC or WP:NOTOR. It had been discussed on WT:TROP, but Hink argues that going through an entire data set is hardly routine. While Wikipedia is not a vote, three of the four project members that have commented here have said that this is not WP:OR. Is this a consensus? YE 17:02, 13 August 2011 (UTC)

To get general consensus, you should bring it up on ORN. I would point out that three uninvolved editors here have voiced concerns that it is OR. Looking over the talk page discussion, I noticed a couple of things. One is that there is some disagreement as to how to use the data, which data are more or less relevant, and how far back the data are reliable. Those disagreements should give one pause. The other is that it seems that the concept of OR is somewhat unclearly expressed. I think the key phrase in this case from WP:OR is "To demonstrate that you are not adding OR, you must be able to cite reliable, published sources that are both directly related to the topic of the article, and directly support the material as presented." You all have directly related nailed. The question is whether these data, as presented in the database, directly support the kinds of assertions you all would like to make. Personally, I think that's a reach, since what we are talking about is doing a query against a database and drawing a conclusion from that query. This is substantially different that representing what a secondary source claims, which is our main activity. Finally, I have to ask, how important are these conclusions anyhow, if they are not available in secondary sources? --Nuujinn (talk) 17:20, 13 August 2011 (UTC)

Can you give me a link to ORN please? HURDAT is reliable reliable sources related to hurricanes. I agree that the big question is whether they directly support the material, you have a good point. You are drawing a conclusion, but it is an obvious conclusion.YE ^{Pacific Hurricane} 17:54, 13 August 2011 (UTC)

It's a shortcut to Wikipedia:No original research/Noticeboard. WhatamIdoing (talk) 21:09, 13 August 2011 (UTC)

(ec)I see what you mean at WP:NOTOR, and the essay identifies that the range of a table is an example of a description of the data that does not advance an argument.  It appears that you need to convince Hink that the database is a table and that it is "quick and easy" to calculate the range, and easy to "detect errors".  In any case you might write to NOAA and ask if they have published this information, or if they would do so.  Unscintillating (talk) 18:21, 13 August 2011 (UTC)

<-- @Nuujinn, that's been my argument the whole time. The information is indirectly gleaned from the best track (but not directly), and I've also questioned how important the information is if no one else said it. --♫ Hurricanehink (talk) 18:00, 13 August 2011 (UTC)

(ec) It should be noted that these conclusions are somewhat important. In Hurricane Karl (1980), without the records, the article is basically a merger. With the records sourced by HURDAT, Hurricane Ekeka and Tropical Storm Omeka are much more notable. YE ^{Pacific Hurricane} 18:11, 13 August 2011 (UTC)

That makes the information sound like fluff, trivia, whatnot. If that's what keeping an article in existence, that's slightly problematic. --♫ Hurricanehink (talk) 18:13, 13 August 2011 (UTC)

I disagree than it is trivia. However, I am starting to agree that HURDAT records is borderline YE ^{Pacific Hurricane} 19:21, 13 August 2011 (UTC) WP:OR. 18:25, 13 August 2011 (UTC)

Yow, that's even worse. I agree with Hurricanehink again. If the only thing that is keeping an article separate and not merged to a list is a single claim based on your interpretation of a large file of binary data, that doesn't work for me. If this conclusion you are drawing from your interpretation of a primary source is the only way to demonstrate notability for a separate article, then it just isn't individually notable. If it were a notable fact, someone in the real world would have actually noted it in a reliable source directly and we wouldn't have to have this discussion. Quale (talk) 03:40, 14 August 2011 (UTC)

In every hurricane season there necessarily must be one that "formed further north than any other", but so what? IMHO only the northernmost hurricane EVER would be notable, but only if it's origin is way outside the "normal" zone for hurricane formation. In that case there would necessarily be news articles or at least academic articles written about it. Roger (talk) 11:15, 15 August 2011 (UTC)

Edit request from 142.177.169.159, 15 August 2011

I am Norman Baglini whose name is erroneously mentioned in the explanation of "deductible". The second sentence that reads, "The first commercial insurance deductible was introduced by Norman Baglini in 1952". is totally incorrect and must be deleted. Please confirm when this has been done by emailing me at redacted. Thank you.

142.177.169.159 (talk) 15:15, 15 August 2011 (UTC)

You are posting your request at the wrong venue. This talk page is for discussing the text of the WP:Verifiability policy. You need to post your edit request at the talk page of the article in question. Blueboar (talk) 15:37, 15 August 2011 (UTC)

The text was in Deductible and had been added by an IP with no source. I removed it. Johnuniq (talk) 00:33, 16 August 2011 (UTC)

RFC on identifiers

There is an RFC on the addition of identifier links to citations by bots. Please comment. Headbomb {talk / contribs / physics / books} 16:00, 15 August 2011 (UTC)

Removing not truth

• Discussion moved to Wikipedia talk:V/First sentence. Please continue there.

Let's end the "Verifiability, not truth" topic

There have been discussions about the "verifiability, not truth" bit for months, and it's getting repetitive. Let's call it a perennial proposal, add a notice about it at the top, and close any new thread on this topic unless it gives some really new perspective on it, not discussed before. We don't even need to list it at Wikipedia:Perennial proposals, because it's already there. Cambalachero (talk) 00:14, 18 August 2011 (UTC)

• I've got a better idea. Let's end the calls to end the debate, at least until we actually reach a consensus either in favour of, or against, "not truth".—S Marshall T/C 00:24, 18 August 2011 (UTC)
• Or, by that reasoning, let's end the idea of keeping "not truth". There was no consensus to keep it, Call it a dead horse. And lets start villainizing people who want to keep it.  :-) North8000 (talk) 00:31, 18 August 2011 (UTC)

• I think a clear consensus here supports a longstanding consensus in favour of "not truth". This is a talk page for a stement of policy to newcomers. It is meant to be useful and easily understood. It is not helpful to flood its talk page with endless debate. I suggest sending further debate to Wikipedia talk:Verifiability, not truth. Working to explore alternative ideas for the project is exactly the purpose of project space essays. --SmokeyJoe (talk) 00:36, 18 August 2011 (UTC)

• I see a clear consensus against Jrincayc's proposed version. I do not see a consensus in favour of "not truth", and in fact the situation with the first sentence is well summarised here. Hiding the problem on a little-used talk page is not going to help.—S Marshall T/C 00:47, 18 August 2011 (UTC)

The way WP:consensus works is that if there is not a consenus to change, then the status quo has consensus. Alternatively, you may take the “policy describes practice”, in which case I challenge you to locate accepted content that is “truth in the absence or verifiability”. Or do you dispute that “verifiability in the absence of truth” is common in the current content?

The place you call “here” is an extraordinary stretch of the meaning of “well summarised”. As for hiding, your “here” has one author and four incoming links.

Things can also be hidden by burying in superfluous content. That is what is going here, at WT:V.

The essay Wikipedia:Verifiability, not truth is indeed little-used, but this is something that should be rectified, not perpetuated. An essay on “verifiability, not truth” is obviously the place to dispute the concept of “verifiability, not truth”. If a alternative viewpoint cannot even be established in that essay, there is no case for repeating past failed arguments on the policy talk page. --SmokeyJoe (talk) 01:47, 18 August 2011 (UTC)

• There is a talk sub-page listing the recent polls on this issue at Wikipedia talk:Verifiability/First sentence polls 2011. Perhaps further discussion could continue there— or as Smokey suggests at Wikipedia talk:Verifiability, not truth—so that this page can be used to address general issues and editors' questions. SlimVirgin ^{TALK|CONTRIBS} 01:26, 18 August 2011 (UTC)

• So many of these terminology issues wouldn’t be here if we instead used Wikipedia:Attribution. That page doesn’t even feel the need to use the word “truth” to explain the “verifiable” in explaining how to add proper content. --SmokeyJoe (talk) 01:54, 18 August 2011 (UTC)

Although THIS is the proper place to discuss issues /potential changes with the policy wording, perhaps it would be good for the folks advocating change to take a few or several week breather to develop thoughts,ideas and presentations in a good fashion. This has sort of turned into trading and parrying talking points.....not that that doesn't have it's value, but I think we're all getting worn out at the moment. Sincerely, North8000 (talk) 01:51, 18 August 2011 (UTC)

• (ec)There is no consensus either to keep or to not keep "not truth".  Suggestions to halt discussion are contrary to WP:Consensus, which calls for discussion when there is no consensus.  "All editors are expected to make a good-faith effort to reach a consensus that is aligned with Wikipedia's principles."  What is appropriate now, however, is a temporary hiatus in discussion, while we focus on the importance of WP:Consensus in future discussions, and that WP:Consensus is a policy that should normally be followed by all editors.  I have proposed a four-weeks hiatus, then a resumption of discussion.  There are various processes such as informal mediation if this future discussion breaks down, but at this point in time we have yet to take WP:Consensus seriously enough to know what the sticking points are.  For example, I suspect that SV and S Marshall could between the two of them come up with a proposal and we could skip a lot of intervening talk.  I find it ironic that policy decision makers don't themselves work to support our policies, can we infer that this is what experience here teaches, that our policies are a tool to be used to maneuver newbies?  What I would suggest is that editors here support our policies.  Unscintillating (talk) 02:13, 18 August 2011 (UTC)

• I consider myself a serious student of WP:Consensus, and as such, I advise that people who want to move from the status quo do so by editing the directly related essay Wikipedia:Verifiability,_not_truth using the advice at WP:BRD. Editing the essay is needed because you should not attempt to advance ideas by directly editing policy pages, and you can’t get very far if the amount of discussion far exceeded direct editing. --SmokeyJoe (talk) 02:31, 18 August 2011 (UTC)
• I know this discussion - and all its relations - are really getting on some people's tits nerves .... but let's all remember that consensus can change. We get newcomers all the time - and some of those newbies can be really quite clueful - especially if their professional field is, for example, communications or wotnots like that. While I'm not suggesting that we go on and on and on about it, I think any decision to stifle this one in perpetuity would possibly prevent a resolution which most of us could be happy with. It's tru there was no consensus to remove "not truth" - it's also tru that there was no consensus to include it in the first place, and the status quo is not, by default, the best place to be. We are all getting worn out, we are all getting ratty (well, some of us are, anyway!) - it's like a really annoying itch that just won't go away. I liked the idea of going towards the attribution page, a bit ... maybe that's an indicator of another possible approach to this long-standing discussion. We all need to remember that (I truly believe) everyone involved in this discussion is genuinely trying to do good, here, so let's not get into bunfights amongst ourselves, yes? (And I still like the idea of wording it verifiable truth, as opposed to verifiability, not truth ..... something along the lines of " ...verifiable truth, not what you personally know, no matter how expert you are." Pesky (talk …stalk!) 02:52, 18 August 2011 (UTC)

There was already an agreement some months ago not to start RFCs on this topic and ruturn here somewhere in September. Count Iblis (talk) 03:14, 18 August 2011 (UTC)

• Yeah, we could stick a "disputed" tag on the problem sentence for now, and talk again in September.—S Marshall T/C 10:35, 18 August 2011 (UTC)

Sidebar note: I started a thread in the previous section which I think was headed towards sorting a few things out on this, but it faded out. (if anyone is interested in continuing it)North8000 (talk) 10:50, 18 August 2011 (UTC)

I'm not for the "let's bury / hide discussion because my preferred version is in there by default right now" ideas. But I'm game for the various ideas along the lines of taking a few week breather while we sort things out / find a different way to approach this. North8000 (talk) 10:55, 18 August 2011 (UTC)

• When dealing with the "let's bury/hide discussion" camp, I'm hoping that the "disputed" notice on the problem sentence will help. The "disputed" notice I suggest can also point to a separate talk page for discussion, which will hopefully clear the decks on this talk page for whatever SlimVirgin et. al. think people should be talking about here, without making the discussion too obscure for good faith users with fresh ideas to find. I'll wait a little while for reasoned objections before placing a "disputed" notice on a policy page, though.—S Marshall T/C 11:15, 18 August 2011 (UTC)

• Attention: I am proposing to place a "disputed" tag on the first sentence of WP:V. Just to make sure people see that and get a chance to respond...—S Marshall T/C 11:20, 18 August 2011 (UTC)

You might put it specifically on "not truth" North8000 (talk) 11:42, 18 August 2011 (UTC)

This is getting beyond the point of disruption... in the previous section we had a clear a 10-2 majority that was opposed to removing the "not truth" language. That is a consensus. North and S Marshal simply do not want to accept it. Two editors, who refuse to accept clear consensus, does not constitute a "dispute". Blueboar (talk) 12:16, 18 August 2011 (UTC)

No, that was for that particular proposal.North8000 (talk) 13:30, 18 August 2011 (UTC)

• I'm curious, Blueboar. What's your understanding of Pesky's position? Or of Unscintillating's position, or of Hans Adler's?—S Marshall T/C 14:31, 18 August 2011 (UTC)

Indeed, this is disruptive. People (many people) are asking for a temporary end to this discussion, and your reaction is to put a tag which specifically instructs peole to continue the discussion... How is that helpful? This thing has been discussed to death for months now, with an RfC and so on, and no better sentence or paragraph has been found. 13:12, 18 August 2011 (UTC)

I think that the people who are disparaging the folks taking their time to represent the 50% who want "not truth" completely out/50%+ who would want a compromise which mitigates it's unintended problems are also fueling this type of thing. Those who are trying to bury the discussion (since their preferred version is in by default, not by consensus) are also fueling this.

And what ever happened to a thread to create a compromise? We had a 6 to 1 consensus amongst the main "combatants" on one. North8000 (talk) 13:30, 18 August 2011 (UTC)

I think you misread the figures: 50% want no change at all, and 50% want some change to the text, but not necessarily the removal of "not truth" (never mind including the opposite with things like "verifiable truth", ugh) Fram (talk) 14:50, 18 August 2011 (UTC)

That's not correct. The biggest RFC was for a proposed change which completely eliminated "not truth". North8000 (talk) 15:14, 18 August 2011 (UTC)

I think it's pretty clear that the most extreme misunderstandings of "verifiability, not truth" are far from having consensus. The problem with the language is that it supports such misunderstandings, but in practice that's harmless as a result of this discussion: The next editor who believes we have a moral obligation to knowingly lie at our readers just because some random, formally reliable source uncontradictedly states something we know to be false can simply be pointed to this talk page. In the long run we should address the problem, but this can wait until everybody has calmed down. Hans Adler 14:07, 18 August 2011 (UTC)

• Do you agree that there's a genuine dispute about the wording going on?—S Marshall T/C 14:31, 18 August 2011 (UTC)

  I think it would be hard to deny there is a genuine dispute. And in fact I am not very happy with the wording either, because I have myself been in several conflicts over the years with editors who were using a fundamentalist interpretation of the first sentence and were totally convinced it was mainstream, which it clearly wasn't. (A lot of editors believe that they take this slogan to be an absolute, inviolable principle, but in the rare really problematic cases they automatically apply common sense. The problem is only those who insist on applying it in situations where it makes no sense.) Hans Adler 15:04, 18 August 2011 (UTC)

How about putting scare quotes on the word "truth"? I'm sure it's been suggested before. The intro makes it pretty clear after that first sentence what is meant: back it up with a legitimate source or it's likely to be sniped as OR/opinion. One person's "truth" does not make anything factually true. Fact vs. fiction: and we have to back up our facts with evidence (i.e. references) to show that it's verifiable. This debate seems a little unnecessary, IMHO. Doc talk 15:14, 18 August 2011 (UTC)

There's little dispute about the intended meaning. The folks advocating change are saying that the poor wording creates unintended meanings. North8000 (talk) 15:42, 18 August 2011 (UTC)

I'm afraid that we're not at the stage where we can talk about remedies. At the moment the question is whether to continue the discussion here, as we've been doing for about eight months now afaik without result, or whether to slap a "disputed" tag on the wording that points interested editors to a separate discussion somewhere else. (Naturally, there are editors who want the discussion to go to a subpage without any kind of "disputed" tag going on their favourite wording, but I think that'll fail. It's pretty clear that there's a genuine and entrenched dispute here.)—S Marshall T/C 15:45, 18 August 2011 (UTC)

Is the "dispute" really about the wording, or is it about the policy itself? Blueboar (talk) 15:48, 18 August 2011 (UTC)

I think that it's absolutely clear that it's just about the two words. I think EVERYBODY involved supports the policy and absolutely requiring verifiability as A requirement for inclusion. North8000 (talk) 15:53, 18 August 2011 (UTC)

(edit conflict) What is the most viable wording that can be agreed upon by those that believe "not truth" needs further clarification? "Not what is offered as truth/claimed to be true"? Bring the word "fact" into the fray? (shudder) I am obviously of the 50% that thinks it's fine the way it is, and haven't looked at the archives. What is the #1 most agreed upon change recommended? A disputed tag would dumb the thing down even further and make the whole policy start to look like a joke, methinks... Doc talk 15:56, 18 August 2011 (UTC)

In all of the flurry, there's hasn't been much organized work on the middle ground. I've been the main one putting stuff out in that area. My feeling in that area is something that says what the first sentence means (i.e. that it just means that verifiability is a requirement for inclusion), such would help get rid of the other imagined meanings. Unfortunately I'll be gone for 4 1/2 days; probably won't be able to discuss/help much. North8000 (talk) 17:07, 18 August 2011 (UTC)

The terminology was inserted by SlimVirgin nearly six years ago, bolded shortly thereafter[4] and moved to the first sentence[5] after the phrase had been removed in a rewrite. Jguk seems to have dropped off of the face of the earth: but SV's addition has been here for a long time. If it ain't broke: don't fix it I always say. This debate will probably be here when you return, however. Have fun on your "days off"! Doc talk 17:39, 18 August 2011 (UTC)

I agree with the idea of adding a "disputed" tag after "not truth", although I'm not sure if it should be added now or such should wait until the last half of September.  I propose adding it on Sep 15.  I'm ok with a new page such as WT:V/First sentence for this future discussion.  Unscintillating (talk) 00:14, 19 August 2011 (UTC)  P.S. Also add "disputed" tags after "threshold" and "true".  Unscintillating (talk) 00:33, 19 August 2011 (UTC)

That appears to be inappropriate. The documentation for {{disputed}} says that it identifies a Wikipedia article as having content whose truth or factual nature is in dispute. Also, the docs say that the template categorizes pages where it appears into Category:Accuracy disputes. Wtmitchell (talk) (earlier Boracay Bill) 01:32, 19 August 2011 (UTC)