To settle upon an approval process, we have to determine two things: what the technical procedure will be whereby an article is marked as "approved," and what people and process will actually result in so marking an article.
=====
I. The technical procedure of approval.
We may choose among at least three procedures for approving articles:
(1) (This option would require that we wait for several weeks or months.) Code, or adapt, a mediawiki plugin that will display the latest approved version, if available, or otherwise the latest unapproved version.
(2) Place an approval template
http://pilot.citizendium.org/wiki/Template:Approved_Articleat the top of an article, which links to an uneditable version in the page history. In this case, if an approved version were available, users would have to click through to it.
(3) "Protect" articles that have been approved, and move ongoing article development to a subpage. For example, the editor-approved version of the "Citizendium" article would live at this URL:
http://pilot.citizendium.org/index.php/Citizendiumwhile the currently-developed version of the article would live at this url:
http://pilot.citizendium.org/index.php/Citizendium/draftAs a result, the most recently approved version of the article would be the version displayed to users, while participants could still work on the article in a "draft" subpage.
Of these three procedures, it appears to me that (1) and (2) can each be ruled out. (1) is unacceptable because we have editors who are chomping at the bit ready to start approving articles. For several of our editors, this is a high priority and it cannot wait for software work that might or might not be coming. (2) is also unacceptable because it contradicts I.3. of the Statement of Fundamental Policies: "If an 'approved' or 'certified' version of an article is available, that version will be presented to the public by default; in that case, viewing unapproved versions will require further mouseclicks for the public, and such versions will be clearly labelled as unapproved, and users will be instructed not to rely upon them."
Perhaps I am not thinking of some other possibilities, but if we are limited to (1)-(3), then we must go with (3). That's my tentative conclusion, subject as always to feedback from everyone. This then brings us to the next main issue, which is *how* articles will be approved.
=====
II. The approval process.
Our Statement of Fundamental Policies also says this: "Among the things that editors will be empowered, singly or collectively, to do are...(2) to approve high-quality articles. Editors will not have the right, except perhaps in very unusual cases, to 'lock' articles and thereby prevent the collaborative process from continuing." Note that option (3) above does not prevent collaboration from continuing, because it can continue on the "draft" subpage.
There is virtually nothing about the approval process in the current version of the "Policy Outline" document, although there are some useful details on a new page, here:
http://pilot.citizendium.org/wiki/Citizendium_Pilot:Approval_ProcessIt seems to me that there are only a few salient questions that inform the shape of the process. They include (but probably are not limited to):
(a) The approval of how many editors is required for a version of an article to be approved?
(b) If more than one person must be involved, through what means do they cooperate and come to a joint decision? Vote, for example, or nomination and seconding?
(c) What mechanism, if any, is used to determine if a person has editorial license to approve a particular article?
(d) Should editors be able to approve articles on which they have worked very much?
(e) Once it is decided, somehow, that an article is to be approved, what happens after that in order to get the approved version marked as approved?
Here then is my proposal, and I hope as always that you will feel free to comment at length and plug for something different. As you know, I do change my mind (sometimes with frightening frequency) and always try to keep an open mind about nearly everything.
As to (a), I think that the approval of just one editor should suffice. There are several reasons for this. First, the added requirement of more editors complicates the process not additively but multiplicatively. I will elaborate this point in more detail if necessary, but one thing I will make explicit is that it is extremely important, if we are to remain competitive and dynamic, that every aspect of our process remain efficient.
Second, the counter-argument is, obviously, that editors working alone can be grossly mistaken, whereas the chance for error greatly decreases with the addition of just one more expert. I have two answers to this argument, namely that the perfect is the enemy of the good, and that there will be plenty of oversight. As to oversight, an editor will not be able to approve on which he or she as worked much; furthermore I propose that we ought not to begin approving articles until there is some minimum number of editors in a given workgroup.
As to (b), while only one editor is needed to approve an article, editors might well want to *consult* with others. Moreover, another editor may revoke the approval of an article, or (pre-emptively) declare that he or she will revoke the approval if it is approved. BECAUSE OF THIS FACT (which I shout out because it's so darn important, you know :-) ), while approval may be done by individuals, the *effect* is that it will be done by groups.
Clarified. In addition, I think that an article about topic T in workgroup W should be approvable only if W has >n editors; we can debate about n, but I suggest n=5, i.e., a half dozen or more editors are needed in a workgroup before any article assigned to that workgroup can be approved. The reason for requiring that number of editors is that, in any group, it's likely that over half of the editors listed won't be paying attention. The hope, actually, is that if there are six editors in a group, then
two of them might be paying attention. This, together with general oversight by non-editors of that workgroup, should be an adequate check against editors approving bunches of stuff that shouldn't be approved. Actually, in some cases it still might not be an adequate check; but this whole enterprise is based on trust, organically-developing patterns, and probabilities.
Clarified. As to (c), I have no special suggestions here. We specify (as we have already done in many places) that a person must be a specialist on the topic of the article to be approved. One *indicator* of this is that the article is assigned to a workgroup in which a person is an editor. But this doesn't even rise to the level of a rule. For all I know, for instance, Russell Potter can be considered an expert on Arctic exploration, although he is listed as a Literature Editor and not in History or Earth Sciences. (He did, after all, edit the Encyclopedia of the Arctic.) And there are many people who are qualified to be editors of, say, Biology, but who lack sufficient expertise to approve of most of all articles assigned to the Biology workgroup.
Beyond this guideline (i.e., that only experts on topic T can approve articles on T), which we will have to codify more carefully of course, I think we will simply have to rely on oversight by everyone, editors and authors alike, to report questionable cases to the relevant workgroup. And as to what workgroups do with questionable cases, well, we haven't faced that problem yet so we haven't worked out the policy yet. But, again, we will have to codify a policy on that.
Amended. As to (d), the answer is No, for the simple reason that individuals are too often poor judges of the merits of their own work. In particular, they often think it's better than it really is. But we may make an exception in the case where there are three or more editors (who are qualified to approve the article) at work on an article. Then any of those editors may approve it, with the expectation that the others will provide adequate oversight of the decision.
Amended/simplified. As to (e), I propose we begin the following process:
(i) On the article's talk page, an editor declares that the article is ready to approve, and waits for a response. At the same time, the editor places a template of this form {{approve|group=W|date=yyyymmdd|url=URL}} on the article's talk page. E.g., {{approve|group=Biology|date=20061220|url=http://pilot.citizendium.org/wiki?title=Biology&oldid=100011126}} . W is the workgroup name. The date given should always be the day after tomorrow, i.e., it will be no less than 24 hours away and no more than 48 hours. However, the date may be today's date if there has been much discussion among at least three editors capable of approving the article, and there is a clear consensus among them. Note that the URL points to a precise version found in the article history (note that the current version of every article does have a stable URL locatable in the page history). The {{approve}} template should automatically place the article in a category such as [[Category:W Articles to Approve]].
(ii) When the date listed arrives, then a sysop (i.e., anyone with sysop rights, meaning any constable or personnel administrator) copies the approved version from the history to the main namespace article page, places an "{{approved}}" template at the top of the article, protects the article, and finally moves the current version to a /draft page (example above). Finally, the sysop removes the {{approve}} template from the talk page (thereby removing it from that queue). Done!
To update an approved article, the procedure is the same, except that the sysop doesn't touch the working copy on the /draft page.
=====
Note, the above concerns the approval *process*. It doesn't touch the question of approval *standards*. Those standards are outlined here:
http://tinyurl.com/yevu93NOTE that I have, earlier today, added a new standard: "Integrated. Articles must be coherent, unified, and integrated. An integrated article is written according to a single coherent and appropriate plan and in a single style. An unintegrated article appears to have been written by different people or at different times, or with different conceptions about the article's proper structure and style. Typically, an unintegrated article repeats information pointlessly and leaves out crucial information where an expert would expect to find it."
Written with compliments to Nancy Sculerati.
--Larry
