Proof versions of articles

[Chris Day]

I like the idea of a draft version.  i think we'll find that sometimes an article has to become worse to become better. That an approved version is static during this time is a huge plus for Citizendium. 

If we really do want to go with a no draft approach we need to figure out how to have the approved version, that will be preserved in the history, show up first with an option to then jump to the most recent version. This may well be posible with a tweek in the code.

In summary, clicking on [[biology]] would lead us to the historical 'approved' version NOT the current version. To get the current version would require a second click.

[Nancy Sculerati]

I agree with Chris' summary above, and basically, it is what we have now, and it does work. If we do not have a static approved version then what's the point of having approved articles at all? Certainly, nothing could be written, responsibly, about medical topics on Citizendium, or much of anything else in which inaccuracy is problematic unless there is a stable version that is reliable. For me, stable versioning is the draw to participation - the key. Further, why is having the system we have now of approved and draft articles at all problematic? An approved article comes up, want to work on it-hit the draft page. Nothing problematic about that, is there? The problem is a concrete means of getting from draft to approved with copyedits. And that problem has to be solved with a specific proposal, a working model- not generalities. I'd like to hear some proposals that give us that, using what we have to work with - at the moment, unless a copyeditor role is assigned, that is- created de novo- we have none. 

[David Tribe]

A heartily agree with Chris'  suggestion above, and also think we have to identify what the problem is. Its not the existence of a draft per se. Its the mechanics of final copy-editing implementation (that is the critical last 24-48 hours of approval), in my opinion.

This can be addressed in a number of ways.

Steven's suggestion of a prominent label is what I had in mind, but compliance with it has to be enforcible, and edits that break the spirit of the"proof" state have to be quickly reversible, else we wait around for days for the errant author to do the job. We don't have that time. Its only other rules that currently  prevent us from doing that. This situation of powerless editors causes us delays, bring all other copy editing to a halt till the author next comes on line). The mechanics of an "illegal" solution are absolutely straight forward. An editor just "reverts" the transgression.

If the Proof label was formal and a short fixed time limit, say 48 hours , as part of a longer Approval stage of say one week,  there no problems. We just end up with a final temporary proofing stage where major content changes are disallowed. They can start again with the  next version.

[Stephen Ewen]

An approved article comes up, want to work on it-hit the draft page. Nothing problematic about that, is there? The problem is a concrete means of getting from draft to approved with copyedits. And that problem has to be solved with a specific proposal, a working model- not generalities. I'd like to hear some proposals that give us that, using what we have to work with - at the moment, unless a copyeditor role is assigned, that is- created de novo- we have none. 

Well, the problem is pointed in this comment of yours, Nancy: http://en.citizendium.org/wiki?title=Talk:Life/Draft&diff=100082799&oldid=100082797

Basically, it seems pretty simple to me. After nomination, the time for content editing is over. Don't nominate until it really is ready, content-wise. after nomination, the time for copyediting --and only copyediting -- has begun. Defining what is and is not copyediting is where things become concrete, and thus enforceable.

[Nancy Sculerati]

That would mean, in my experience, that we would never have had approved a single article. It's a new experience for us all to collaborate on articles, and what is needed is a clear end-pont that helps us co-ordinate the end of content editing and the start of copyediting. In other words, most of the biology editors, anyway (all of them?) are published authors, solo or with co-authors in print media. When a manuscript is accepted for publication, it generally is accepted for content, and then the proof work begins. We need the same process here. A concrete process that we all can understand and follow. Since it's a wiki, otherwise we just keep on going and there's always "one more thing".

Stephen,in the past, when a major article has been nominated for approval, you see, it tends to get "swarmed" (a term Zach coined describing the process for [[Biology]]) with authors and editors who each want to get their last lick in- like an eBay auction. The nomination for approval is not a redlight - STOP, it's a yellow light, with a timer going (or at least a default timer). In an article like Life, yellow speeds up many authors, just like it does drivers.  That means that what was moments ago a polished article, now invariably has rough edges. The process is not terribly problematic- it works, overall, but it works best if once the article has those rough edges, it gets copyedited and, if a new bit of content didn't get smoothed in, that part is presented to everyone and discussed, and a solution for it is worked out on the talk page. So a bit of editing is always needed as well. That's how it was done for the very first approved version of the very first approved article (Biology). I'll find where that was discussed, months ago, right here in these forums.

Found the thread http://forum.citizendium.org/index.php/topic,331.15.html Look at my post dated December 13, 2006. Start there. This issue with Life is the same old story, that's why David Tribe remarks "deja vu", and it has happened with several other articles, too. It needs fixing, it's not our imagination. We really are trying our best here, we love CZ- and we have managed to forge ahead despite these problems but again, they really are problems and they are really only fully understood by the people actively involved of the nitty gritty of the approval process. Now that you one of those people, Stephen, you'll get to understand. Maybe you'll come up with the solution. I think Catherine has hit on it, it just needs refining.

[Nancy Sculerati]

By the way, one of the reasons I love CZ is that its a place where people really do get passionate over ideas. I think that's really exciting. In a way, I'm glad we have this problem, it is a sign of vitality.

[Anthony.Sebastian]

In my opinion, any 'proof' version of an article should have a very short life span, no more than two days, say. 

The reason: CZ articles live.  They continue to grow, mature, learn from mistakes, incorporate new ideas and understanding, new facts, new levels of organization.  Writing = rewriting.  We need to keep up our momentum improving the quality of the article after approval.  That suffers with a long 'proof' life span. 

Let the copyeditors fix the glaring solecisms. No article should get a nomination tag until all it needs is solecism clean up.  Keep the articles alive.  Don't put them in suspended animation.  Don't liken the process to print publishing.  CZ'sa wiki.

In my opinion.

[Larry Sanger]

The system works, and I mean no insult or disrespect to those of you who have made it work--but it does not work nearly as well as I imagine it could work.

I'd just like us to discuss proposals creatively and "think outside the box" (a phrase the use of which now indicates thinking inside the box, but never mind).

Here's what's problematic about the current system: there are only 11 approved articles, and there are a lot of very active editors in fields other than biology who aren't approving anything.  Why not?  Partly because there aren't enough editors in those other fields; but I think another part is that there is a pretty steep learning curve when it comes to approval.  The fact that the biologists understand the approval system, since they helped develop it, explains not only why there are a lot of approved articles in biology, but perhaps also why there are relatively many active biology editors.

It's not like there are many articles in CZ that are going unapproved, when they should be approved.  It's just that, I think, we could have a lot more editors working toward approval in many more subjects with a simpler, more elegant system.  What exactly that system should be, I don't think we've hit upon yet, precisely.

Now, if you'll indulge me a bit: if we can understand (and successfully promote) the object of approval not as a specific version of an article, but instead as a work in progress, then what follows?

Well, first of all, there's the reason this seems like such a nonstarter to you: an approval notice might be placed on text that wasn't specifically approved.  I agree that that could be a problem.  But why, exactly, would it be?  It's a problem presumably because an approval counts as some sort of guarantee of the quality of the information, and there is no such guarantee about new information.

I'd like to point out, however, that it is hardly as though when someone approves a piece of writing for publication, even a very conscientious editor in the most rigorous traditional system, the publisher has examined carefully every single claim in it with the utmost care.  In fact, one main reason that we can continue to improve articles that have been approved is precisely that there is room for improvement.  If you are worried about seeming to approve some text that has been newly added, when it might be incorrect, then why don't you examine as absolutely carefully as possible every claim made in the article?  After all, those claims might be incorrect as well.  Because, in short, that's too much to ask.

In that case, I would say that it might be sufficient for our purposes if we add a few rules and make a few observations.  One of the latter is that, presumably, if we require three editors to approve an article, new changes to the approved article will be closely watched, and undone or corrected if in error.  Another observation is that we can TRY this and then SCRAP it (and return to the present system) if it turns out to be not to be working.  Moreover, we can mitigate a great deal of the risk by using some such rules as these:

(1) Most importantly, if an editor suspects that changes to an article have made it unworthy of remaining approved, he or she may simply remove the approval template.

(2) If any person wishes to make sweeping changes--to do a major rewrite--he may get permission from others on the talk page, and then proceed to remove the approval template while the article is "under construction."

(3) You may not make any very sweeping edits to an article without first discussing them on the talk page.  This is already often the case.

(4) Non-specialists might be encouraged, or even required, to submit substantive edits on the talk page.

(5) We can state very clearly what approval means and provide a suitable disclaimer.  We can, among other things, state the above rules on a page that explains what approval means.

(6) We can, retrospectively, identify particularly good versions found in the archive, and in the approval template direct readers to those versions if they wish to be especially sure that a version has been fully reviewed by at least three CZ editors.

Consider next the potential advantages of this proposal.

It is much easier to understand, teach, and implement in new groups.  We tell new editors: you want to approve an article?  Then simply go to the talk page and declare, "I think the current version (link to it) meets CZ's standards."  Add your name to a list of approving editors.  When three editors have approved an article, up goes the approval template.  As long as the number of approving editors is three or more, and there are no dissenters, the approval template remains on the article.

The simplicity of this system, which is its central virtue, would in turn (I think) result in many more articles being approved, and might very well motivate more editors to get more actively involved in other groups.

This is ONE proposal.  I suspect there are several others possible.

Now, please don't get too excited about all this, or ask about when we're going to implement it--I would not implement this without getting feedback from the Editorial Council, which I'm on the record now as saying I'll set up this week.  :-)

[Larry Sanger]

Extra rules for Nancy:

(7a) Editors may designate certain articles (e.g., on medical treatments; or sensitive Topic Informant Workgroup articles), or perhaps certain sections of articles, as editors-only.

(7b) Alternatively, all new substantive (not "copyediting") edits to such restricted articles might have to be proposed on the talk page, and then require at least one "second" from a specialist on the talk page before anyone can change the text accordingly.

[David Goodman]

No article here will ever be static. This will I hope remain a wiki, and articles will continue not to be personally owed. They should always be open to minor changes, and open to the suggestion of major changes.

I particularly like Larry's addition of 7b. I think a rule for concurrence with edits makes extremely good practical sense in eliminated nonconstructive trivia.

[Nancy Sculerati]

I cannot follow your post Larry. I understand that you think the process is too complicated, and so do we all. Can we stick to a real example, or even a hypothetical example?

PS- Citizendium should work without "extra rules"

[Nancy Sculerati]

Anthony - of course it's a wiki. That's the whole idea behind a proof version- NOT a time period, let an article be approved then RIGHT AWAY there is a proof version, for copyediting and a DRAFT version for continued creative work. Could we just try it and see? Isn't that the spirit of a wiki?

[Larry Sanger]

See http://en.citizendium.org/wiki/User_talk:Larry_Sanger/Approved_in_pagehist for reference in following the following.

Here's another idea (I'll illustrate the earlier idea later, Nancy)--a relatively easy method--something I advanced in the early days.  The approved version of an article is, precisely, a page that appears in the archive; so you link to that page using a template.  If you want to change which version is approved, you simply change the link to a new version.  Quite frankly, I would be willing to scrap the Draft system, and *temporarily* display drafts first (contrary to what the Statement of Fundamental Policies says), and link to the approved version VERY prominently (think: large, bold, colored font; big fat link people can't help but click) from the top of a page.

Again, see http://en.citizendium.org/wiki/User_talk:Larry_Sanger/Approved_in_pagehist for an example.

No doubt we'd prefer to change MediaWiki so that, for non-logged in users, the first displayed article is the most recently approved version in the page history, if any.  For logged-in users, we'd see the version in progress.  I think Greg has already done some coding that would support such a feature.

[Chris Day]

No doubt we'd prefer to change MediaWiki so that, for non-logged in users, the first displayed article is the most recently approved version in the page history, if any.  For logged-in users, we'd see the version in progress.  I think Greg has already done some coding that would support such a feature.

Pro: If Greg can get this done then i think this is good from the sense to not have to juggle to different page histories for the same article (this is similar to what i had suggested above).

Con: copy editing the approved version will be almost impossible unless the article is protected for a set period of time (i think this should be come policy if we run with the idea). However, if at a later data there are still serious copyedit issues, what then?  By that time the current version will have drifted from the approved version.  In my opinon reverting back to the approved version with the copyedits is a non starter, as i mentioned above, sometimes the article has to get worse to get better. But if we keep reverting back to the approved version with minor copyedits we'll lose any subsequent edits. This may well slow down the progress of article improvement.

In summary, the strength of the draft system is that the draft is always available for editing. The more i think of this problem (despite my suggestion of using the history to show the approved version), the more I side with maintaining two copies the approved and draft versions. This idea means that any copy edits made to the approved version MUST also be made to the draft version.  While a bit of a pain for the editor doing such edits, it is the best solution to allow the articles to continue to evolve.

[Larry Sanger]

Question to the biologists who have tracked many draft pages: in your estimate, what proportion of edits made on a draft page actually make the article worse?  Surely it can't be a high number: why else would people be making the edits, if not to improve the article?

If a high enough proportion of edits actually do improve an approved article, and if the edits are rarely if ever so bad that you would want to revoke approval, then perhaps we should be allowing people to edit the approved versions of articles directly.