Wikipedia talk:Flagged revisions/Quality versions

Latest comment: 16 years ago by Mrg3105 in topic QA process

Who reviews what?

edit

Okay, so to rehash my previous question from the old proposal: who determines which editors review which articles? Because it would be nice to see subject experts reviewing those subjects, rather than a random editor who may or may not be equipped to notice more subtle errors or omissions in the content itself. If we're going to go out of our way to flag an article as "Quality", there needs to be some reasonable guarantee that not only has it been checked, but that a Quality-Editor with sufficient qualifications will be publicly identified on the article as having signed off on the version. I simply don't see the point otherwise. Girolamo Savonarola 20:13, 31 August 2007 (UTC)Reply

This needs attention. OTRS volunteers as a self selected group who lack accountability and transparency to the Wikipedia community are not the appropriate group. Admins would be a better starting group if an initial group is needed. GRBerry 21:01, 31 August 2007 (UTC)Reply
I oppose both. One of our worst editors was a subject matter expert with a professorship and an extremely strong point of view, which he used random articles from scholar.google.com to support. He is now happily editing at Citzendium, where he belongs. Wikipedia's decisions should be made on argument and evidence.
Admins lack accountability for this purpose; disagreeing with a reviewer would be the quintessential content dispute, so ArbCom will not intervene - and frequently desysopping would be excessive anyway. There is no other check on admins on a toot. Septentrionalis PMAnderson 02:53, 18 September 2007 (UTC)Reply
I don't see how that stands to reason - all editors still have the right to challenge material if there is contrary evidence. I'm only stating that it would probably be for the better of the proposed-quality tag if people who could actually read and understand the articles due to prior experience were actively engaged. This could just as easily be coordinated through the relevant WikiProjects. Girolamo Savonarola 03:22, 18 September 2007 (UTC) (Prior statement is not referring to admins, so not relevant.) Girolamo Savonarola 04:57, 18 September 2007 (UTC)Reply
The assumption that all admins can or do read articles is unsupported by evidence. An admin need not be a competent editor; he merely needs to get lucky at RfAr and not seriously offend ArbCom. Septentrionalis PMAnderson 04:17, 18 September 2007 (UTC)Reply

"certain standard"

edit

What is the standard going to be. Featured article is an unreasonable standard. Even a "Start class" article could well merit a quality version, and most "B class" articles will deserve them. Frankly, any article that doesn't have a degree of incompleteness that amounts to either being a stub or being an NPOV problem in coverage probably can qualify, so long as it is reasonably sourced. GRBerry 21:05, 31 August 2007 (UTC)Reply

Yeah, FA standards are largely superficial. Good sourcing and basic completeness should be the focus here. Voice-of-All 02:22, 2 September 2007 (UTC)Reply

Proposed process

edit
  • Complete fact-checking of an article should be the minimum standard.
  • To smooth that process, excerpted versions should be allowed:
  • Create a version containing only material with sources that have been checked.
  • Revert that version so that the "live" page contains the full version again.
  • As additional sources are checked and/or provided, update the quality version with the newly-checked stuff added back in.
  • If the excerpted version would be too incomplete, defer review.
  • The fact-checking should be a pure-wiki process, based on trust, not editcountitis:
  • A minimum of two editors pick a version and vouch that they've checked all its facts and sources;
  • Other editors are also encouraged to check, especially if the first two editors are new or were involved in writing the article;
  • The candidate "quality version" should be presented on the article talk page for a week or so. If there are any reasonable objections to it, it should be rewritten (with any new facts re-checked) until a consensus version is developed;
  • One or more "Reviewers" assess that the candidate version has consensus and vouch for the work of the fact-checking editors based on their prior fact-checking contributions;
  • Quality version is passed.
  • The same process should be followed when updating an existing quality version, but only the diffs need to be fact-checked.
  • Articles that have problems beyond needing fact-checking (i.e., neutrality concerns) should not be passed until a consensus version is developed by participating editors, which must also be fact-checked.

--Father Goose 21:48, 31 August 2007 (UTC)Reply
--Added "consensus" steps. Father Goose 20:22, 15 October 2007 (UTC)Reply

What I left out of the above proposal:

  • Should admins and Reviewers be the same thing?
  • Should Reviewers have no special status aside from explicit community support, and only be allowed to recommend new versions, with admins doing the updating?
  • Should Reviewers be a special class of user, with an "update quality version" button?
  • If not admins, how are "Reviewers" selected or promoted? "Requests for Reviewership?" A more informal nomination process?

--Father Goose 22:00, 31 August 2007 (UTC)Reply

    • This is impractical. FA does not, in practice, completely fact-check an article, although it probably should, and might if it didn't spend so much time on trivialities from the Manual of Style. Any rate of flagging approaching the present rate of FA nominations, which would be absurdly low for a process intended to cover much of Wikipedia, would make this unworkable. Septentrionalis PMAnderson 17:10, 18 September 2007 (UTC)Reply
      • What makes it practical is that only a subset of each article has to be fact-checked. Whatever can be readily checked will be included in the "quality version", with everything else excluded until those parts can be independently checked. But whatever is included in the version we certify has to be fully fact-checked. That has to be the absolute minimum standard. Fact-checking FAs shouldn't be too difficult, as they do insist on complete sourcing, so it's just a question of checking the source material.--Father Goose 00:31, 19 September 2007 (UTC)Reply

I'd say that reviewers would be community appointed via WP:RfR or something. They would just measure consensus (like Bcrats). If on a talk page, there is a consensus of at least three editors to flag a revision, the a Reviewer that is not one of those users or a frequent of the page, should review it. Talk pages can be tagged with "needs review" tags so that reviewers can just scan the category. Auto-reviewing for quality version can be disabled (so that reviewers can make big edits when stable=current without it being reviewed). Voice-of-All 05:54, 14 October 2007 (UTC)Reply

Hmm... you somehow managed to summarize the entire consensus version proposal in one paragraph by removing the instruction creep. :) --Merzul 12:04, 15 October 2007 (UTC)Reply

More concretely about the proposed process above, the only thing I'm not sure about is the gradual review process with excerpted versions... Perhaps, it wouldn't be practical otherwise, but if the focus is on quality, then fact checking should simply be given more emphasis. I don't see the problem that for a short period editors of an article really focus on fact-checking the article. Thus, the review process would be like above, but initiated when editors of an article wish to have the article flagged, then they must undertake the rigorous fact-checking process, and as above give a detailed on wiki-discussion, and proof that hey have evaluated all sources. A reviewer then looks over the discussion, etc, and based on this discussion flags the page. --Merzul 18:15, 15 October 2007 (UTC)Reply

As I keep saying, fact-checking must be the minimum standard. But by pairing it with consensus, I think other standards of quality will be incorporated as well. People will raise objections to any version they think is inferior in quality to a prior version, and will presumably object if there are any other glaring faults, such as non-neutrality or terrible writing. I think we don't need a formal spec for these types of "quality" -- people will speak up when they feel something is amiss. Every quality version will have been scrutinized by at least two fact-checkers and one reviewer, so I think truly poor articles are unlikely to slip through process as outlined above. For articles that receive a lot of attention, the volume of interest will likely ensure a high degree of quality.
I've amended the process outlined above to include the "discussion" stage. Let me know if you think it needs more changes.--Father Goose 20:22, 15 October 2007 (UTC)Reply

Clarrification

edit

Reading Wikipedia:Flagged_revisions I don't think that "For logged-in users the option to display the current (default) or quality version of the article would be a user preference." should be part of this proposal. I don't think that flaggedRevs can do this. Can someone clarrify? Adam McCormick 19:09, 2 September 2007 (UTC)Reply

No it can't. Though the hooks in the core seem usable by the extension if such a thing were to be added. Voice-of-All 03:05, 11 September 2007 (UTC)Reply

Will this raise liability?

edit

At the moment, we have a fairly open editing/viewing model, which includes explicit disclaimers that things are, in a nutshell, on an "as-is" basis. In other words, legally we stand by nothing. Will the introduction of "approved and quality checked" versions which are potentially given preferential viewing treatment above other diffs potentially create legal issues? Because this would make fact checking into a necessarily overwrought process. I'm just worried that quality issues that nonetheless escape with errors intact may in fact create even worse problems than the Seigenthaler incident, particularly if labeled exceptionally. Girolamo Savonarola 00:53, 19 September 2007 (UTC)Reply

Legally, the Wikimedia Foundation is only a provider (not a publisher) and cannot really be sued. The people who "sight" and "approve" the versions are not selected by the Foundation itself, but by the community. So this really shouldn't make us legally liable. Cbrown1023 talk 22:56, 6 November 2007 (UTC)Reply
The Foundation has a good legal defense, but the person who assigns the quality rating to an article is in a weaker legal position. She is responsible for her action. Since her action is not exactly making or publishing a libelous statement, but rather not removing one, it seems to me (IANAL, TG) that the most that could be found is that she is guilty of negligence. (On the other hand, the quality rating could be viewed as endorsing or supporting the libelous statement.) The original inserter of the libelous statement remains primarily at fault. -R. S. Shaw 00:46, 7 November 2007 (UTC)Reply

This issue is patent nonsense. Can anyone cite a legal basis for this, or is this just amateur lawyering? Marc Shepherd 14:36, 7 November 2007 (UTC)Reply

There has been some prior discussion of this at Wikipedia talk:Flagged revisions/Sighted versions#Liability. IANAL, but I continue to believe what I said there - if this is implemented there will sooner or later be a court case testing this, and lawyers for the plaintiff will argue that this increases liability. I make no prediction on what a jury would do. But to be clear about it; liability concerns are a primary reason (among others) that I will not participate in any fashion in supporting this system. GRBerry 14:50, 7 November 2007 (UTC)Reply
Other than your personal unsubstantiated feelings, do you have any verifiable sources to back this up? If not, then it's basically "original research." Or worse, no research at all. Marc Shepherd 16:25, 7 November 2007 (UTC)Reply
I wasn't aware that WP:V and WP:NOR applied to the Wikipedia talk namespace... In any case, all I am doing is raising the issue. I don't think it's disingenuous to consider the larger implications - this is a huge change to the way that content purporting to be encyclopedic will be displayed on one of the most-accessed sites online. I'm happy to be proven too cautious on the matter, but I don't think that raising the issue is unreasonable; I think trying to ignore it completely is. I just want some informed opinion, that's all. Girolamo Savonarola 16:48, 7 November 2007 (UTC)Reply
Strictly speaking, WP:V and WP:NOR don't apply in the talk namespace. But if liability is offered as the reason to be concerned about the proposal, that concern needs to be backed up at some point, especially where the people raising it admit they are not lawyers. Marc Shepherd 16:54, 7 November 2007 (UTC)Reply
One can file a lawsuit for just about any reason, so ultimately I think GRBerry is right that sooner or later this (and Wikipedia in general) will be tested in the legal system. But I'd be surprised if it amounted to anything. How much legal responsibility do traditional publishers have to provide unfailingly accurate information? Since we avoid "how to" information, I'd be really suprised if one could claim much more than "you unintentionally gave me incorrect information which did not cause me physical harm" -- how strong a suit could that be?--Father Goose 18:50, 7 November 2007 (UTC)Reply
I'm more concerned about the publishing/endorsing libel and slander area damaging a person's reputation. We already know of real cases where a person has been confined (bordering on arrested) because of content in a Wikipedia article. So we know of real cases where there have been negative consequences to a real person because of Wikipedia content. Someday, someone will sue because of this sort of thing. If problematic content only became visible because someone flagged a revision, all of a sudden the flagger is also at risk because they are exercising real editorial oversight.
Marc, no, those of us concerned don't need to get a lawyer and prove that we are right. We can oppose the system because we are unwilling to take a risk. If you want to convince us that we are wrong, try to get a competent legal authority to explain how we as editors are protected from liability. If all you want is to leave the matter in doubt for the next editor to consider the proposal "unproven" is good enough - but if you want to convince those who have already reached a conclusion you need some affirmative evidence for your side. GRBerry 19:57, 7 November 2007 (UTC)Reply
Obviously, you're free to oppose any idea for any reason you want. Heck, you could even say, "It was proposed on a Friday, and I think Fridays are bad luck." But if you expect that argument to be taken seriously, you need evidence, and you've offered none. On top of that, as you're concededly not a lawyer, your advice on the matter has the same standing as medical advice from non-doctors, or accounting advice from non-CPAs.
You asked me for affirmative evidence. How does one go about proving a negative? You haven't offered a prima facie case, so there is nothing to refute. You just seem to have a loose "feeling" about it, which I'm sure is very real to you, but which is unburdened by any actual facts. What is any other person supposed to do about it? Marc Shepherd 20:11, 7 November 2007 (UTC)Reply
(outdent) Many lawsuits are frivolous; nonetheless, it costs a substantial amount of money to defend against them, however obvious the case may seem. You can argue against my concerns all you want, and that's fine, but I don't think my raising the concern is improper. Girolamo Savonarola 20:16, 7 November 2007 (UTC)Reply
I'm not suggesting it's improper to raise the concern. But taking the concern seriously requires more than just a loose "feeling" expressed by non-lawyers. Marc Shepherd 20:49, 7 November 2007 (UTC)Reply
That's not what "This issue is patent nonsense" communicates. ("Aliens will shoot reviewers with their ray guns" would be patent nonsense and would not be welcome here except as comic relief; the prospect of possible legal trouble is hardly nonsense.) You, like anyone, can hire a lawyer to give an assessment here (or try to convince one to give a free one) to show the issue is trivial (if not nonsense). -R. S. Shaw 23:12, 7 November 2007 (UTC)Reply
Marc, I want to hear everyone's opinion, within reason. I understand your position, but it would nice to hear from a broad spectrum of experienced voices. Girolamo Savonarola 23:27, 7 November 2007 (UTC)Reply
Frivolous lawsuits or Wikipedia bashing are at least plausible. I'd advice contacting a lawyer (do we even have a foundation lawyer anymore?). Voice-of-All 23:56, 7 November 2007 (UTC)Reply
Sure do: Mike Godwin (User:Mikegodwin).--Father Goose 02:00, 8 November 2007 (UTC)Reply
Incidentally, see Wikipedia:Wikipedia Signpost/2007-11-05/French lawsuit for some details of actual lawsuits we've faced.--Father Goose 03:40, 8 November 2007 (UTC)Reply

Unofficial demo set up

edit

See here.--Eloquence* 11:38, 3 October 2007 (UTC)Reply

Dead link as of today. Wjhonson (talk) 20:24, 26 January 2008 (UTC)Reply

Nutshell

edit

Page needs a nutshell. Let's start working on it. Wjhonson (talk) 20:23, 26 January 2008 (UTC)Reply

I did some big changes that should be reverted then maybe discussed ;). I did this because it seems en:wiki is more interested in Sighted Versions... but I think this one is "the real deal"! Perhaps, I'm just hurrying too much, and clearly we should start with sighted version, where quality flags are only used by OTRS and other emergencies. --Merzul (talk) 21:37, 19 April 2008 (UTC)Reply

Only crats can dish this out?

edit

Won't this will further increase the workload of crats? Can't this right be given by admins as well? OhanaUnitedTalk page 03:49, 15 May 2008 (UTC)Reply

That is configurable, and therefore a matter for discussion. – Mike.lifeguard | @en.wb 17:25, 15 May 2008 (UTC)Reply
But it's better to discuss it now rather than after it's implemented. Less conflicts and issues would occur this way. OhanaUnitedTalk page 00:35, 17 May 2008 (UTC)Reply

QA process

edit

This is a wonderful idea, but I would like to point out a few things.

  • Wikipedia has two aspects to it: the ongoing processing of articles, and the Projects of completing articles databases in their areas of knowledge - yes, one day in the dim future Wikipedia Projects will be obsolete because all its articles will be completed to as far a completeness
  • The process as it works currently creates enormous friction due to the high number of articles that are either complete rejects from the start (and flagged for deletion), or worse, are stubs,and require substantial editing by more regular Wikipedia editors, preventing them from participating on GA/FA articles. I'm not sure what the percentage are across Wikipedia, but it seems to me that about 80% of new articles are below desired Start status on creation.
  • With this in mind, it seems to me that the QA process needs to start "at the gate" as it were. Its fine for anyone and everyone to create articles on any conceivable subject, but I would like to suggest that no article can be added to Wikipedia without some basic components such as: At least one source, At least the defining article subject introductory statement referenced, the appropriate Project tag in the Talk, and at least one category. The Category:Uncategorised needs to be eliminated because it has just become a dump for articles and it serves no purpose as a category other then to tell the reader that Wikipedia has no idea where the article belongs in the greater scope of human knowledge.
  • It would be good if if QA was coordinated with the Categorisation people.--mrg3105 (comms) ♠22:28, 29 May 2008 (UTC)Reply