CZ:Proposals/Self-Correction Policy

This proposal has been assigned to the Editorial Council, and is now in the Editorial Council proposals queue.

Driver: Larry Sanger

Complete explanation

The Citizendium should adopt a policy that requires that we list all factual errors in previous versions of our articles--including unapproved articles--on a new "Corrections" subpage.

A correction should be made only if:

• The claim is truly a matter of fact. We will not announce correction of bias.
• The claim can be proven to have been in error. A source should be linked to (or quoted) in most cases. A reasonable difference of scholarly opinion must not be identified as an error.
• The claim is reasonably significant, not trivial. Trivial errors include typographical errors that do not affect the meaning of the text, grammar and spelling changes that do not affect meaning, and so forth. Any matter with the slightest tendency to affect a person or other entity's public reputation is significant. Moreover, matters on which students might be tested, and that people might use as the basis for action, are to be considered significant.
• The claim has been on the wiki long enough to have potentially "done damage." As a rule of thumb, this will be 24 hours. We need not announce corrections of quickly-fixed errors, unless the error became public knowledge in the short time in which it appeared.

Generally, it is probably unrealistic to expect us to note all factual errors we have corrected. But we will commit to noting at least those errors described above and which were requested to be corrected either on a talk page or by a member of the public.

Correction notices are especially encouraged when a person who is/was involved with the matter discussed (i.e., a person who might become a topic informant) asks for the article to be corrected for factual error, especially egregious or multiple factual errors.

No person will be made to take responsibility for a particular error, even if he wishes to do so.

Correction notices will be removed one year after being placed on an article, or an amount of time otherwise decided by the Editor-in-Chief or Corrections Editor (if available), commensurate with the impact of the error.

The Editor-in-Chief or, when installed, the Corrections Editor will have the final word on whether a correct must be made, and what form it must take.

Reasoning

In brief, this is the same standard that newspapers and other legitimate periodicals use. We are obligated to adopt the same policy. Doing so will earn us good will from the public and increase our credibility considerably.

Argument 1. Self-correction would encourage people to write a bit more carefully. To avoid being "corrected" (but bear in mind that no individual person will be made to take responsibility in a correction notice!), many of us will be more likely to double-check our facts. This will have a material effect on the credibility of the results.

Argument 2. Self-correction would increase our credibility. The Citizendium has already hoisted a flag over new territory: we say we stand for credibility and expert knowledge, as well as openness and dynamic collaboration. If we own up to our errors, we are in effect declaring: we care so much about our own reputation for accuracy that we will take responsibility when we get things wrong. Lists of errors are embarrassing, but they are also impressive in that they reveal commitment and responsibility on the part of the authors.

Argument 3. Encourages public feedback and hence involvement in the project. Particularly if we provide an e-mail address or form that people can use to post factual corrections to us, no doubt some people will act as fact-checkers for us. Some of those people will get involved. This can do nothing but good.

Argument 4. Self-correction would sharpen our vision and leadership for a new, more responsible Internet. Wikipedia has very famously made some mistakes, and if we grow large enough, it is only a matter of time before the Citizendium is caught in some howlers. (One hopes this will not happen so often and that they won't be so bad; but I do expect it to happen.) If Wikipedia had adopted a policy of posting corrections of errors, and apologizing for them, the public ill-will its scandals generated would have been considerably moderated. If the Citizendium differentiates itself by committing to publicly correcting and apologizing for its errors, we will at once increase our standards and create a reputation for higher credibility than the usual open Internet website. That can do nothing but good.

In fact, we would be differentiating ourselves not only from Wikipedia, but from the vast majority of wikis, blogs, and other "crowd-sourced" websites. We would essentially be taking the leadership of this issue of owning up to errors. If the issue became an important public issue (see next argument), this could raise the profile and importance of CZ as a whole.

Imagine a Citizendium-sponsored logo or badge:

This website is

self-correcting.

Argument 5. Announcement of a self-correction policy might spearhead a public debate--at which we would be the center. The media, knowledge professionals, and a large part of the general public are understandably and increasingly bothered by the amount of sheer garbage and error online. The typical responses to this are: (1) yes, the Internet isn't not perfect, but look at how much information there is! And (2) we're just regular people exercising our freedom, and there's nothing wrong with that. But suppose we put the issue of self-correction into the arena of debate about the Internet. Then there are natural replies: (1) sure, it's not perfect, so why not do your share of improving it and own up to your errors? And (2) you can more responsibly exercise your freedom if you commit to admitting your mistakes. Knowing the Internet media as I do, I am inclined that some of them would jump on this as they did over Tim O'Reilly's Bloggers' Code of Conduct. O'Reilly did say, "And when things go awry, acknowledge it," but he did not ask bloggers to own up to their errors when they are shown to be in error. I would like to do so, on behalf of the Citizendium as a whole. A paragraph of a press release could read:

"The 'publish then filter' Internet desperately needs to improve its credibility. There is no reason that Wikipedia, blogs, and other websites should not adopt the same standard of self-correction that newspapers use," Sanger said. "I urge everyone with a website, no matter how small, to explicitly adopt a self-correction policy. I also urge users of popular wikis, blogs, and other websites to demand that they adopt such a policy."

Argument 6. If adopted on the Internet generally, self-correction might improve the tone and value of Internet discussion. One of the features in most (legitimate) old-fashioned publishing as well as face-to-face conversation is that one can immediately correct errors in what others say; our readers and friends make us own up to our errors. On the enormous, anonymous Internet, posting online means never having to say you're sorry. Well, we are ideally positioned, as a cutting-edge Internet project, lead the rest of the Internet toward a culture of self-correction. Imagine an Internet in which people lose credibility (and readership) for being proven wrong, but not owning up to their error.

Implementation

A practical "to do list" type explanation of how the proposal will be implemented, and who will implement it. If there is no one to implement the proposal (as, for example, with many technical or recruitment proposals), then it is automatically declined.

Discussion

I am totally against this policy. First, we shouldn't have to do the extra work this would entail, particularly for first drafts. Second, I have never picked up a newspaper and found errors displayed for the public from their first draft of a news story. They may keep internal records, but no public ones at the bottom of the actual newspaper article. Also, it should be taken as a given that draft articles will have a few mistakes. If you want to keep a record of errors, it should be on a separate page that only CZ people can read. David E. Volk 09:04, 28 March 2008 (CDT)

David Volk is exactly right. We of course already do have a complete file that contains all back versions and all their mistakes. We are not at all like newspapers in this regard (they are telling about their errors in original research, which we do not engage in.) Richard Jensen 09:17, 28 March 2008 (CDT)

But unlike newspapers, we have decided to present our information for public consumption before it is "complete" and fully vetted. Obviously, again unlike newspapers, we are committed to the whole philosophy of publish then filter. But these are additional reasons to adopt a self-correction policy. After all, newspapers have self-correction policies, I assume, because this improves their credibility and to give their readers a sense that they really are committed to accuracy, and are not above correcting themselves publicly as needed. I can see creating a new Corrections subpage, but it would defeat the purpose of the policy entirely to make that page hidden from the public. --Larry Sanger 09:24, 28 March 2008 (CDT)

Although newspapers make corrections, they often hide them on page 36, hoping that nobody reads them. They also lump them into one heading as another means of hiding them. Science journals tend to do the opposite, they would say something like Correction:We cured cancer in mice, by XXX and YYY. David E. Volk 10:06, 28 March 2008 (CDT)

A good consequence of this (I am still against it in general) is that all authors are more likely to write and verify on Sandbox pages or personal computers before posting. David E. Volk 10:28, 28 March 2008 (CDT)

After more thought, I have to agree, David, that at least putting the corrections on a subpage is a good idea. I just imagine an article having that "spot" on it for years. So I'm inclined to make this time-sensitive. We will remove a correction notice after one year, and we will post corrections only if an error has appeared for more than 24 hours...well, you'll see the language. Also, I agree with the good consequence, but one of the objections I want to address is related to that: won't this make people even less "bold" about contributing? It's worth considering. --Larry Sanger 10:34, 28 March 2008 (CDT)

I can't possibly understand why anyone would be against this. There are so many reasons why this is a good idea:

• content accuracy over time - if something is incorrect or errored, then it will surely be fixed
• possibly expediting dispute resolution - creates a channel for corrective input
• enables legitimate feedback - enabling bonafide experts to have input
• can be a recruitment tool - may convince authoritative figures to sign up and or contribute
• enchance reputation - let us be known as "the source" for accurate information
• adds a level of transparency - shows that as human beings, we are not beyond infallibility, and are willing to go the distance to make it right.

--Robert W King 10:40, 28 March 2008 (CDT)

Thanks, Robert. Some good points there. --Larry Sanger 12:19, 28 March 2008 (CDT)

My main objection was to having the errors right on the main page. As for the points made by Robert,

• this won't effect content accuracy over time, errors can be reported and fixed, but not published in an error list.
• this proposal has no mechanism for expediting disputes. Facts and references are needed, just like now
• feedback is already a proposal, not related to this one

Initially, this will:

• Prod people to report on the number of current errors we have
• Entice people to go looking for errors we haven't found yet, lead to some bad press as well as good.

The good points in Roberts arguments are:

• Recuitment will go up, after we first take a hit or two on the chin
• Does enhance reputation
• Does show integrity

David E. Volk 11:07, 28 March 2008

Comments on David's points:

• this won't effect content accuracy over time, errors can be reported and fixed, but not published in an error list.

Obviously we'll catch many errors without reporting them. But that doesn't mean that we (and the rest of the world) won't catch more errors, and more quickly, if we report them.

• this proposal has no mechanism for expediting disputes. Facts and references are needed, just like now

Well, this proposal has named decisionmakers (the editor-in-chief or, possibly in the future, chief area editors). I could see adding a Corrections Editor instead--actually, that might be better. Note that the disputes that might arise here will be relatively tractable compared to many content disputes: do we describe a particular sentence as an error and own up to it (and does it fit the other self-correction rules), or not? It's "yes" or "no," and in case of dispute, the decisionmaking entity states the decision, and that's the end of the matter.

• feedback is already a proposal, not related to this one

The two proposals definitely impact and support each other; not sure why you include this among your objections.

Larry, the feedback comment was meant to suggest that that particular point is mute, not a statement of objection regarding the proposal. You will note I have made points both for and against the proposal. If we decide to put error liss as a subpage, I am agreeable to the proposal. David E. Volk 12:56, 28 March 2008 (CDT)

OK-dokey, thanks for the clarification. I was simply writing to elaborate the proposal further, not to attack you... --Larry Sanger 12:59, 28 March 2008 (CDT)

• Prod people to report on the number of current errors we have

Which is a decided advantage to us. Those people will be more likely to get involved, and they're doing fact-checking, a crucial service to us.

• Entice people to go looking for errors we haven't found yet, lead to some bad press as well as good.

Even the "bad press" will be good. More impressive than the errors, which will be easy to dismiss, is the fact that we own up to them. Wikipedia doesn't own up to its errors in this way; hence CZ joins the narrative of how to deal with Wikipedia's lack of reliability. "Why doesn't Wikipedia have a self-correction policy, like Citizendium?" people will say.

--Larry Sanger 12:30, 28 March 2008 (CDT)

There are some parts to this I really like (the basic concept, and having them on a subpage), some parts I wildly disagree with (listing errors in drafts), and some parts I'm basically neutral on, although I have a lean (deleting them after a year). To tackle my main disagreement, I think it's not good to publish errors in draft versions because, well, they are drafts! It no more makes sense to list errors in drafts than books should list errata in the pre-prints (I forget the publishing jargon for them) that are circulated in limited numbers. Also, if we list errors in drafts, it's going to greatly increase the amount of paperwork involved in working on articles, which we don't need. J. Noel Chiappa 12:00, 28 March 2008 (CDT)

I actually like that idea. Since our aim is to approve articles anyway (declaring them to be our "best"), it would probably be wise to only deal with corrections on those alone; since "draft" or non-approved status is constantly available for edits (until the next approval). --Robert W King 12:09, 28 March 2008 (CDT)

So you would be opposed to Wikipedia listing corrections? After all, they don't have approved versions; it's all drafts. Well, I would think that they should list corrections. If so, then we should list corrections for our drafts.

While we might call unapproved articles "drafts," and while we might label them as such, we are still putting our drafts out there for public consumption. If Wikipedia's experience is any indication, the public will take what it gets from us as probably correct. They are using our drafts as if they were finished products. We can't stop them from doing so short of closing the project, which would be a disastrous error, I think. How the public uses our work is what should determine whether we list corrections or not, I think.

Think of it this way. CZ is part of the whole "Publish, then filter" movement. I am proposing a tweak to this movement: "Publish, then filter, but admit your errors." If you can agree with that in general, it implies that we should post corrections of our drafts, not just our approved pages. --Larry Sanger 12:19, 28 March 2008 (CDT)

Well, we should also consider that perhaps making a record of errors could end up being a full-time job in itself, frankly one that I don't think anyone would volunteer for. There's only so much "red tape" people can stand. And isn't our primary goal to have approved articles anyway? --Robert W King 12:23, 28 March 2008 (CDT)

A full-time job? Maybe, maybe not. How many errors will we catch? Not many, I suspect; I think there aren't that many factual errors in CZ. But if there are, that's another reason to have a Corrections Editor, and possibly a whole Corrections Workgroup. OK, I'm going to add that in. No one will volunteer for it? Oh, I think you're very wrong there. I think there will be plenty of people who would like to hold our feet to the flame in this way. :-) --Larry Sanger 12:41, 28 March 2008 (CDT)

"you would be opposed to Wikipedia listing corrections"? Jeez, talk about rhetorical straw men! :-) But to answer, I am put in mind of the parable about 'sewing a piece of new cloth to an old'! Asking your question is, to me, akin to asking whether I'd be opposed to improving bleeding of patients who have the flu by using leeches instead of knives! :-) The whole Wikipedia process is so broken I don't think it's useful to try and compare with it. Now, to be serious...

Your point about drafts is a good one, but I'm still concerned about the amount of work we're letting ourselves in for if we include drafts too. I'd prefer to use an even larger and more aggressive warning on drafts (or perhaps even block non-Citizens from seeing them, although that's probably too aggressive), before including drafts in this policy.

The problem is we're discussing this is a bit of a vacuum, data-wise. Maybe there aren't that many errors in drafts, and it won't be an issue. Maybe (on the other far end) there are a heck of a lot, and as a result groups of authors will take to collaborating on new versions of articles in other places, so they don't have to bother with this paperwork (which well remove the interesting data in the history on how the various others contribute to the final shape of the article). The thing is, we just don't know...

Don't get me wrong, I highly applaud the basic concept, it's just this one aspect that's a concern to me. J. Noel Chiappa 13:04, 28 March 2008 (CDT)

Actually, this is very interesting. I can't gainsay you when you say that we need more data, or that we might catch a lot of errors. My guess is that we won't, but of course we might. I'm not entirely sure that we could be caught off guard, though. Well, like this: a person who corrects a clear factual error should also be motivated to record the correction--but, of course, in some cases, people might be happy to correct errors without recording the correction. Indeed, the whole question is how we decide, and when, to record a correction. How about this: when a factual correction is explicitly requested on the talk page or from outside the project, the correction is noted. If someone just happens to come across a factual error and fixes it while editing an article, we have no means whereby to expect that the person record his work on the Corrections subpage. Of course, if someone simply volunteers on the spot to record the correction, he may (if it is within the rules). Then the workload question becomes: will there be so many correction requests on talk pages and from outside the project that we cannot keep up with both the corrections and the logging thereof on Corrections subpages? That seems very unlikely indeed to me, but--right--we should do a pilot project first to see if it's feasible. --Larry Sanger 14:56, 28 March 2008 (CDT)

Is it possible that we're thinking too far ahead on this? Are we creating a catch-all for something that doesn't necessarily need catching yet? I mean I'm all for this, but now that I think about it, it seems premature. --Robert W King 15:01, 28 March 2008 (CDT)

Well, if it's not needed, that turns out to be excellent and low-cost good PR for us. --Larry Sanger 15:15, 28 March 2008 (CDT)

Which is, in part, why I'm in favour of it for Approved articles (mostly it's because I think it's plain just the right and intellectually honest thing to do for them, with all sorts of precedents in books, journals, etc). J. Noel Chiappa 16:01, 28 March 2008 (CDT)

I don't understand the need for this on unapproved articles. My understanding was that Citizendium makes no guarentees about articles which have not been approved. Only for approved articles do we stick our neck out and say "this is correct". I would regard it almost like preprints vs. refereed papers. Journals will often print errata for mistakes in refereed papers whereas the un-refereed arXiv preprint server rarely (ever?) has errata. Roger Moore 15:31, 28 March 2008 (CDT)

I agree with Roger here. I don't see any "damage" an unapproved article could do, because that's exactly what this disclaimer was written for. Maybe, we should make it a little more eye-catching, similar to the approval notice with a different color, but that's another discussion. Otherwise, I'm in favor of this proposal. Jens Mildner 16:00, 28 March 2008 (CDT)

The main opportunity is not to fix a problem, i.e., our failure to acknowledge damage that an unapproved article has done. It's correct that this isn't really a problem, at least now. Mind you, it could be a problem--it would be a mistake to think that we won't be taken to task for errors in our unapproved articles, as Wikipedia has been. The main opportunity, rather, is to solidify in the public mind that we really are committed to accuracy. Taking responsibility for clear factual errors in articles that we publicly posted is the opportunity I have in mind. --Larry Sanger 10:42, 29 March 2008 (CDT)

I think the approval process itself is a very good answer to criticisms of inaccuracy in articles. If we are taken to task for inaccuracies we can point out that it was a draft article and has not yet been approved. Nobody takes arXiv to task for inaccuracies in preprints. However I agree for this to be feasible there needs to be a clear way to distinguish the two. One idea would be to have an option to only view approved articles. The other problem with errata for drafts is how do you define an error? Supposing I am editing an article and accidentally flip a sign in an equation or put 'GeV' instead of 'TeV' etc. Those are factual errors so would I need to report them? Supposing I catch them with 24 hours or even 1 hour do I need to report a correction then? So there are both philosophical and practical reasons to restrict the errata to approved articles only. Roger Moore 12:46, 31 March 2008 (CDT)

The way CZ is running right now, errors are flagged very clearly in the discussion pages. I haven't seen anybody correcting somebody else's error yet without a polite explanation on the associated talk page. Outsiders consulting a CZ article can read the talk page and if they are too impatient for this, they probably won't look at the page with corrections either. So, if citizens stay as polite as they are now, there is no urgency in a correction page, unless such page becomes a "mea culpa" page where authors/editors confess their own mistakes.--Paul Wormer 15:46, 28 March 2008 (CDT)

Anthony.Sebastian here: I'd start the 'self-correction' program by vetting and periodically re-vetting the approved articles, corrections proposed on the Talk page, at least two editors agreeing the correction(s) valid, an assigned editor for each article putting corrections on a subpage, purging all older than one year. I'd also mention the 'self-correction' program on the article's approval banner, and invite reader input on factual errors (on the Talk page) but requiring readers to join CZ before allowing input. Perhaps that way we can motivate ourselves to expedite the approval process. BTW: I think we need a more comprehensive (non-public) 'quality control' program, with a definition of quality. --Anthony.Sebastian 16:41, 28 March 2008 (CDT)

I think we should record errors for approved articles. I don't think we should report errors for unapproved articles unless:

• The erroneous text was substantially insulting to someone or
• The erroneous text was cited by a third party.

Corrections living for one year on the article page sounds a bit extreme; how about 1 month on the article page and forever on the corrections subpage? Warren Schudy 19:56, 28 March 2008 (CDT)

• I think this is unnecessary complication as applied to unapproved articles-- we edit on a wiki, so we can make corrections as we go. The earlier versions are by definition working versions that have been supplemented or corrected. DavidGoodman 22:03, 28 March 2008 (CDT)

I generally agree that this sounds like a good idea for approved articles, and a bad idea for unapproved articles. If someone wants to see our mistakes in unapproved articles, they can go to the talk page or comb through the article history. I think Larry's arguments for this apply far more to approved pages than to unapproved/draft pages. Anthony Argyriou 18:11, 1 April 2008 (CDT)