Wikipedia talk:Stable versions now/Archive 2

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

Archive 2

Archive 3

Archive 4

Default to 'stable' or 'editable'

I think most people agree that some sort of 'stable version' system is a good thing. Whether it is located in a separate namespace, specially named article-space page, talk sub-page, history link, or where-ever and how the process is administered are details which need to be worked out but which should not present great difficulties to resolve.

However, I think there is strong disagreement about whether the 'stable' version or the 'editable' version should be the 'default' when someone types in a term and clicks 'Go'. I think the ideal would be for the user to be able to choose to either always get the 'editable' version or default to 'stable' and get the 'editable' only if there isn't a stable version available. Unfortunately, any such option will need to wait on code changes. In the meantime, if we are going to move forward with 'stable' versions we need to agree upon what page is displayed by default.

I am strongly opposed to 'stable' being the unalterable default because it completely changes the way Wikipedia is supposed to work. If a correction is needed or new information about a subject becomes available it should be possible to update the page and display it immediately... not wait to gather consensus that the edit (and all other edits over a span of a few months) is acceptable for display. Yes, the 'editable' version of the page updates immediately... if the user knows where to find it. If a casual user goes to fix an error and gets blocked because the page is protected they may not even know that an editable version exists. Likewise, you shouldn't have to guess whether to type 'Articlename' or 'Articlename/development' to get to an article you want to update. When you click a wiki-link you should get to a page you can correct rather than one which is static. Uneditable by default just isn't the way Wikipedia is supposed to work. Editable versions should be the default display with 'stable' versions as a 'clean backup'. --CBD 13:13, 12 July 2006 (UTC)

I agree that users should be able to choose. Hopefully the software upgrade allows people to pick in their preferences which they want to appear as default. For those wanting static versions, the development version would appear only if there isn't a static version available. In that case, I feel that anonymous editors should be pointed to static versions by default, and that those who have accounts would be able to switch the default to the editable version.

But as of right now we don't have a software upgrade. I think it's unnecessary to assume the worst case all the time (that it will always take several months for any changes to be made to stabilized versions), and there will be links at the top of every page describing how to make changes, but I agree, this issue of which version is the default is significant. An editable default doesn't do too much good at this point, so maybe we ought to test the other idea (static default) now and see what happens. Why don't we, as noted in a section above, do a test of 20 articles—implement this on 10 and leave 10 as a control group, and see what happens? --Spangineer^es (háblame) 13:40, 12 July 2006 (UTC)

I can't honestly see the point in having static versions, if they're not going to be the default view for the casual visitor. It would be quite simple, I think, to set up preferences to allow logged-in users to choose which they display, but for readers, it seems to me to make perfect sense that they see the version which is guaranteed not to have been vandalised, and has been reviewed and deemed worthy of a static version. This way, we also provide a large incentive for editors to improve articles to the point where they can be made static. Worldtraveller 14:07, 12 July 2006 (UTC)

I'm curious as to what you expect to happen. How would we know whether the test had been a success or not? What will be the measurables? --OpenToppedBus - Talk to the driver 13:59, 12 July 2006 (UTC)

Vandalism is one thing, and lag time is another. I'd expect the amount of vandalism to drop significantly, since vandals wouldn't be able to edit the page that appears to all readers. Vandalism patrol would still have to whack vandalism to development versions, but it wouldn't be as critical. Lag time, as you and others have pointed out, is a potentially serious problem, however. I'm not sure how it would turn out—best case would be that insignificant improvements are added to the article within hours and that good, substantial edits are added to the article within days. Of course, there's the potential that such additions would be delayed (backlog, not enough admins, inability to get a group together for consensus). That lag time would then have to be compared to the bad edit lag time of the articles in the control group--how long it takes for vandalism and less than helpful edits to be reverted.

There is the potential for bias--there are alot of admins supporting and opposing this idea on this page, and thus it'd be pretty easy for all of them to add the articles being tested to their watchlists. We'd end up with restabilized articles every few hours if necessary, no problem. Similarly, the extra attention would make people revert vandalism of articles in the control group more quickly than they otherwise might. So perhaps to simulate real work conditions, pick one or two admins who are allowed to restabilize versions. Pick a category of articles that they aren't familiar with, so that they have to work to judge the value of contributions (just like they'd have to do if this were implemented on a wide scale). Pick some FAs, some GAs, and some B-class articles. Make a control group with a similar composition of articles, by both quality and subject, but don't tell more than a few people which articles are in that group until after the fact to prevent it from getting extra anti-vandalism attention during the test. Let the test run for a month, and see what we learn.

We wouldn't be able to judge the effect on our image ("what do you mean my edits won't appear instantly!"), but we already have that with existing protected and semi-protected articles, as well as no page creation for anons. But if we were able to, say, show that stable versions helped the Britney Spears article cut 90% of total vandalism edits while maintaining an average good edit lag time of 5 hours compared to a 3 hour bad edit lag time on the non-stabilized Mariah Carey article, we'd at least have something real to talk about rather than just saying "I do/don't think it will work". Shall I put together a proposal page/essay for something like this? --Spangineer^es (háblame) 14:39, 12 July 2006 (UTC)

I have serious doubts as to whether you can effectively eliminate the bias issues you identify above - the very nature of Wikipedia makes blinded experiments difficult and double-blinded virtually impossible - but I'm open to being convinced. In the meantime, though, I much prefer the "stabilized featured article proposal" being discussed above. --OpenToppedBus - Talk to the driver 16:10, 12 July 2006 (UTC)

I think the main measurable I am interested in is whether stabilizing an article results in less contructive edits being made to the delvelopment version than would be made to a regular article. Mainly I think the proposal is good, but we need to be sure minimize the side effects. I imagine the exact wording of the templates will have a large effect on how "encouraged" people feel to find the delvelopment article and edit it. If people try to "game" the trial run it should be easy to tell. We can disreard those results, but I do not think it should be a problem. Surely everyone interested in the proposal (for or against) would be interested in seeing some examples of what would happen in practice.--Birgitte§β ʈ Talk 17:09, 12 July 2006 (UTC)

I'd have a problem with anons (let alone everyone) being forced to 'stable' versions unless we had some very specific / high standards for stabilization. For instance, articles on 'current subjects', should then IMO never be stabilized no matter what their quality... all articles on living world leaders should remain unstabilized because they could die or do something else noteworthy. Stabilization should be a counter to vandalism and deteoration... never an impediment to needed editing. Any page on which there is or may likely be more to say should never be protected from immediate edits over a long period of time. That'd be counter to the entire premise/foundation of how Wikipedia works. If stable = not immediately editable then it should have standards even higher than FA - basically requiring that the article be of professional caliber at least the equal of what could be found in a major printed encyclopedia covering the topic... and then only on subjects unlikely to require significant change/expansion. I'm talking about people going through and verifying that every reference is on point for the facts it is supporting, that the writing is professional quality, that all stable articles follow consistent style standards, et cetera... exacting review to verify that the version being made stable is something that could be released in a printed encyclopedia and still be an excellent article after sitting on someone's shelf for a decade. On the other hand, if the 'stable' version is more like a 'backup for reference', and not the default displayed, then I think any reasonably good version on any article could be 'stabilized'. The standards currently described fall more towards the 'reasonably good' end of the scale... and I cannot agree with permanently protecting every article which achieves a 'reasonably good' version or anything close to that. If 'stable' includes protection from immediate editing then it should only be applied to articles which we can reasonably expect not to need immediate edits. --CBD 11:14, 13 July 2006 (UTC)

One of the main benefits of stable versions is that they prevent slander from showing up in the articles on living people (and thus eliminate alot of legal headaches; hence the above support of the foundation leader/lawyer). There are articles with moderately high vandalism rates that aren't semi-protected, and as a result, there's some small percentage of time that the article says something ridiculous. Then of course there are the high quality articles that degrade—just look at any FA that doesn't have its author watching over it, and you'll see what I mean. People add sections of badly written prose, failing to realize that the same point is summarized two sections below. Or they'll happily insert unsourced and dubious information, and there it will sit.

If we say that an article has to be perfect to get a stable version, there's no point in having stable versions because they won't be used. We're not suggesting here that articles should be protected forever. As I said above, we could implement something like WP:AIV so that within minutes any article could be destabilized or restabilized in an emergency (like someone famous dying). It's obvious that more improvements can always be made to any article (even if it's at the "professional" level you're talking about), but that's OK. We create a system that allows them to be updated quickly but that doesn't permit vandalism or content degradation. Yes, a delay would exist, but people use answers.com all the time and they don't seem to mind that the information is a week old. Personally, I think the average reader would be willing to sacrifice a few minutes or hours of new information if they were able to have greater confidence in the accuracy of the information they did see. --Spangineer^es (háblame) 12:49, 13 July 2006 (UTC)

I don't see the turnaround times you suggest as feasible. Any 'unlocking' of the page to receive new edits is going to require admin action and that will create a significant bottleneck. Take one of those 'living people biographies' as an example... assume it was stabilized and then the development copy received those standard edits where people "add sections of badly written prose", "insert unsourced and dubious information", et cetera... and then the person dies. How do these admins notice that the 'request to unlock', one amongst hundreds (thousands?), for that page has switched from one where 'Joe Editor' has added 'unsourced and dubious information' which he wants to see updated to the actual article and a request to add the fact that the person died? Once found, does the admin just 'unlock' the page with the 'bad' edits to that point intact? Are admins now empowered to police the content and unlock to the >stable< version rather than the development page or review the changes and decide which to keep?

This 'sacrifice of a few minutes or hours' was unrealistic, and one of the fatal problems, when such a structure of 'review for accuracy before approving' was used on Nupedia. The results here would be the same. Admins should not be in the role of deciding what to incorporate and what not... but even if they were it does not seem possible for a thousand admins (or whatever) to review the edits of hundreds of thousands of users to 'biographies of living persons' and other 'protected' pages in a timely fashion - and even more implausible to think that such decisions could be quickly made by consensus.

As to 'preventing slander from showing up'... this proposal wouldn't do that. The slander could still be added to the 'development version'. It could still be viewed, copied, and indexed by search engines. IMO that goal could only be achieved by policing all edits to every page before they were applied... which is clearly impossible and contrary to the fundamental nature of Wiki collaboration in any case. You aren't talking about just having 'stable versions' of pages available... that just requires a link to the history. Rather, you are talking about a vastly more restrictive form of 'page protection' which would require some sort of consensus or admin approval of further edits. --CBD 10:57, 14 July 2006 (UTC)

Re turn around times: perhaps a distinction like the one that exists between WP:AIV and the content dispute reporting system could reduce the issue. Only "emergency" reports would be added to the AIV equivalent, while consensus-backed changes would be listed somewhere else. People do a pretty good job of clearing content disputes off of AIV, so maybe the same would be possible for a parallel system for stable versions.

As for destabilizing to an inferior article, so what? We're no worse off than we were before stable versions existed. In fact, we're better off because the bad edits don't appear in the article immediadetly—they sit in the development version until destabilization occurs. Once the inferior version is the primary version, it will presumably get alot of attention (since the reason for destabilization was that it needs major edits) and the problems will get fixed just like they are currently (via the wiki).

Re Nupedia: Nupedia was different because all content, from start to finish, was created in a controlled environment. As others have pointed out, a Wiki is a valuable tool for developing articles but not for maintaining their quality. Thus, only articles that have already taken advantage of the wiki's strength and reached a certain level of quality (which would have to be determined) would be able to be stabilized. It is a significant change, and it is more like Nupedia, but think there's still a significant difference between the two.

You're right about the vandalism to the development version, but consider: how many vandals will stick around when they know that their work isn't going to appear on the actual article that people see? It becomes an exercise in futility (what fun is there in vandalizing the sandbox?), and I suspect that numbers would drop significantly. The time saved by vandalism patrol could then be applied to other activities, like, say, stabilizing articles =). Unfortunately, the only way to find out how much vandalism will go down is to implement the system. A controlled test run (or at least as controlled as Wikipedia can allow) might give us an idea though.

I don't want "stable versions" as links to the history, because that's useless. I want to be able to create a great article and not have to worry that it'll degrade into a mess of inaccurate and unsourced garbage if I go on vacation. I also want Wikipedia's reputation to improve. I don't think that these desires are incompatible with this being a free encyclopedia—I don't support locking down the wiki and monitoring all edits, and I don't want to protect pages indefinitely. I'm optimistic that we can set up a system that allows us to lock articles and yet be able to make consensus-backed updates frequently (on the order of hours or days, if necessary) and unlock them immediately (on the order of minutes). --Spangineer^es (háblame) 13:28, 14 July 2006 (UTC)

To reword part of what you wrote just slightly (one word), "Consider: how many editors will stick around when they know that their work isn't going to appear on the actual article that people see?" The satisfaction of seeing their work included drives our contributors just as much as it drives the vandals. The idea of 'stable versions' can be implemented with anywhere from 0% impact on vandalism to near 100% reduction of vandalism 'tacked on' separately... but the greater the reduction in vandalism the greater the impediment to editing will be. Wikipedia exists and has been successful because it goes so far to not impede editing. When 'semi-protection' came out there were news stories about how Wikipedia is no longer 'allowing anyone to edit'... even though semi-protection is actually less restrictive than full protection. The system you describe would clearly be orders of magnitude more restrictive, and IMO is a bad idea both for "Wikipedia's reputation" and the actual viability of the project. I'm all for identifying 'stable versions' that have been reviewed for accuracy... but that isn't what this proposal is about. That can be done without the rest of this. The proposal here is really to sacrifice ease of editing in order to reduce vandalism (with 'stable versions' as a backdrop)... which has been consistently and soundly rejected every time it has come up in the past. --CBD 12:50, 15 July 2006 (UTC)

(de-indent) It's a question of balance and article maturity. A near complete article will almost always need something removed for each thing added, so a discussion or consensus process should happen before articles are changed on such articles anyway; meanwhile, the chances of any given edit being either vandalism or good faith but not worth keeping begin to overwhelm the improving edits. I think there are many articles that would currently benifit by stable version, both by reducing vandalism and making editors more thoughtful and less impulsive in their edits.--ragesoss 12:59, 15 July 2006 (UTC)

An example of linking to stable versions using FA protocols and the system we have now

Guys, I was talking above about using the article history of good and featured articles to link to permanent versions now, without having to juggle the main page. I've created an example using a randomly chosen featured article. As you can see, I've put a subtle banner across the top saying that there's a static version available, and linked to it; you can even see the difference between that article and the current one. This way, we have our static articles already, we have 1,000 Featured Articles and who even knows how many good articles that we can set this up on right away, and we have our "certified good" version that people can look at, while still editing the main page as always. Any and all feedback would be appreciated, here or on the talk page of the example I've set up. JDoorj a m Talk 05:48, 12 July 2006 (UTC)

Personally I like the idea of having the editable version as the default, with a stable/static version for those requesting it, but the main problem with the above method is that there is nothing stopping a vandal changing the target of the stable version to a vandalised version (or editors edit warring over which is the "approved" version). Regards, MartinRe 07:54, 12 July 2006 (UTC)

I greatly prefer this proposal. The FA process would have the additional charge of applying the label explicitly to a single revision, but given that, I would not worry too much about the links to that version being vandalized. Unlike this page's proposal, JDoorjam's maintains our successful philosophy in full, does nothing to discourage editing, and does not add bureaucracy (which would be confusing to novice editors, and extremely time consuming for others). ×Meegs 08:18, 12 July 2006 (UTC)

I prefer this proposal too, but feel it would have difficulties in implementation. Until we have the option of "protected section" or similar, there will be the risk that the target would be changed, and also if someone followed the stable/featured version link, and spotted a mistake, it would be too easy for them to edit it, miss the "out of date warning" and wipe out the development version (as it's not protected). That said, the SVN proposal also has difficulties, my main concern is that it expands the protection policy to cover highly edited pages as well as highly vandalised pages. Also, I don't think the "de-stabilisation" aspect would work in practice, if a stabalised article turns controversial, the proposal says that it should be updated to the development version and unprotected, but in practice, I think it would be more likely for an admin to say "work out the issues on the draft/talk page, and in the mean time the article is remaining protected (at the stable version) to stop edit wars. Regards, MartinRe 12:31, 12 July 2006 (UTC)

I think that the banner is ugly, but I'm sure that can be changed :) Otherwise good idea. - FrancisTyers · 11:56, 12 July 2006 (UTC)

I much, much prefer this proposal. It's taken a while for me to work out why the original proposal made me unconfortable, and I think I now realise why. It's entirely to do with which version someone comes across when they go to an article. It is a huge change to the whole mentality of Wikipedia if, when someone goes to Linus Pauling (for example), they get an article which they cannot edit. Effectively, what we're saying is, "you can submit changes, but they won't really be part of Wikipedia until someone else has looked at them and reviewed them". That's not Wikipedia any more, as far as I'm concerned. The default setting should always be that you get an editable version - the stable version should be the one that takes an extra click. --OpenToppedBus - Talk to the driver 12:05, 12 July 2006 (UTC)

Ditto. This method would be acceptable... highly beneficial even. The changes suggested on the project page, most notably the editable version being on a sub-page rather than the default article displayed, are not remotely acceptable to me. Having links to a 'stable version' in the history for articles would be great. They could be reviewed and updated from time to time, could include a link to comparison against the current article, et cetera. --CBD 12:12, 12 July 2006 (UTC)

Doing so would not reduce any incentive to vandalize (in fact, it becomes a target for vandalism itself), it would not provide any incentive to editors to keep the stable version moving forward, and it would not be used by readers... a majority of whom miss the fact that they can edit, even though a typical article has a half dozen edit buttons. I don't have any objection to doing this, but I wouldn't expect to achieve any substantial gains from it. It's also simple enough that I don't think we need any discussion for it... if you think doing that would help something: just do it. --Gmaxwell 12:34, 12 July 2006 (UTC)

Of course it would help if FAs were factually verified, but pigs might fly :) - FrancisTyers · 12:38, 12 July 2006 (UTC)

Actually, we easily can protect the link to the protected article. All that needs to happen is to put the banner on the top in, say, template space, and protect it. So, for instance, this one would be Template:Featured article/Linus Pauling. The template gets locked, and while, yes, a vandal could remove the template, that'd be rather obvious, and anyone (or any bot) could notice and fix its removal. Again, I acknowledge, it's one more teensy layer of work, but it's a simple fix to what seems like the only objection I've heard so far. JDoorj a m Talk 14:25, 12 July 2006 (UTC)

There's also the other issue that MartinRe raised above, that prominent links to a version in the history could lead to inexperienced editors unintentionally wiping-out later changes if they miss the "You are editing a prior version of this page" warning. I'm having trouble estimating that risk. ×Meegs 15:11, 12 July 2006 (UTC)

Ah, now that is a concern (though, of course, "wiping out" is easily unwiped out by a revert). I have a couple suggestions about mitigating that, too (though with the current software, you're right, it's not possible to wholly prevent). I'm going to put something together at Wikipedia:Stabilizing Featured Articles which will outline my proposals on the matter; I'll let y'all know when that goes blue. JDoorj a m Talk 15:15, 12 July 2006 (UTC)

I much prefer this proposal, much along the same lines as OpenToppedBus. I think having the primary version be protected is counter to what Wikipedia has been up to this point, and having your changes shoved off into some corner where you don't know when it will be implemented (a day? two days? a week?) is an unsettling feeling, and I really think it would discourage editing. I also think that protecting the main versions of a large amount of articles would garner us quite a bit of bad press- I mean, we got bad press because of semi-protection for crying out loud. Anyway, I think the aim to have stable versions that readers can trust is a great one, and I think at this point in time JDoorjam's suggestion is the best way to do it. EWS23 (Leave me a message!) 03:47, 13 July 2006 (UTC)

I do not prefer this proposal; or, more specifically, I do not believe this proposal addresses the same problems the original proposal does. As far as I can tell, this proposal offers benefits only for republishers (mirrors, CDs, print versions). Very, very few interactive browsers of the encyclopedia -- you know, the readers? -- are going to bother to click on these little links to make sure they aren't looking at a vandalized version. I'm a lot more interested in benefitting the millions of Wikipedia readers and boosting the overall reputation of the encyclopdia than I am in making life a bit easier for the republishers. —Bunchofgrapes (talk) 04:01, 13 July 2006 (UTC)

The "readers" of wikipedia.org are not, and cannot be, our primary audience, no matter their numbers. Wikipedia.org is a website with one central purpose - "to write an encyclopedia". That's what we do here, that's what we are for, that's the purpose of all 150 servers the Wikimedia Foundation has scattered around the world. We write an encyclopedia here. We do it by means of volunteers, most of whom joined us because they came across the site looking for information. Information is our hook - it's what we use to get people interested enough to edit. Wikipedia.org is not here to provide information - Wikipedia.org is here to assist people in creating (well, really summarizing and organizing) information. In a very real way, the whole point of displaying articles at all is simply as bait to convince people to click the edit button - clicking the edit button is what we really want them to do. We only provide a public, web-based distribution of our content to entice people to edit. As one of the goals of the Wikimedia Foundation (and, presumably, of most of the members of the Wikipedia community) is to distribute knowledge to the world, it may be a good and important thing for Wikimedia and the community to set-up, pay for, and maintain a public, web-based distribution of all the knowlege we can (primarily the material we've created through wikipedia.org), but that project is distinct from the purpose of Wikipedia.org. As I said above, Wikipedia.org has one purpose, and that above all else - to write an encyclopedia. We should not sacrifice it for anything, even for distributing the results of that work. JesseW, the juggling janitor 01:39, 14 July 2006 (UTC)

Interesting perspective -- I somehow haven't run into or been made to think about the distinction before. Under that view, what is the point of any of these stabiliation or identification plans -- what's the value in tagging consensus version at all? —Bunchofgrapes (talk) 01:54, 14 July 2006 (UTC)

Oh, I can explain that! The benefits to editors of identifing a consenus or reviewed version include: 1) The ability to easily identify changes between the reviewed version and the current version; at this point, there's no easy way to group "all-the-changes-that-haven't -been-reviewed-yet", making identifing subtle vandalism or mere innocent mistakes harder. 2) The ability to feel more confident that they can work on an article, get it well arranged (good grammar, well-sectioned, fully verified, etc.), and have that good arrangement remain avialable and identified, even while work on expanding the article continues. Those are the top two that come to mind, but I'm sure there are others. Those reasons are why I'm interested in and supportive of this proposal, not for any puported benefits to readers, or decrease in vandalism. JesseW, the juggling janitor 02:17, 14 July 2006 (UTC)

Yes, this is an approach I was going to suggest myself after reading the Signpost story. Support JDoorjam idea, oppose the proposal as current. The main article must be editable in the spirit of Wiki.--Piotr Konieczny aka Prokonsul Piotrus ^Talk 16:47, 15 July 2006 (UTC)

Ditto Bunchofgrapes, that's a perspective (JesseW's) that I hadn't considered before. I'm not sure though. What's the point of a encyclopedia? I think we can agree that we want it to be read; otherwise we're wasting our time. And I think that regardless of what happens, our primary audience will always be online. Even though we are gaining more and more offline opportunities to spread our content, the internet continues to grow and reach to the corners of the world. Thus I'm skeptical; it seems counterproductive to not consider our online readers except when trying to lure them in. And it seems inherently backward to use the internet to create information and paper to distribute it, especially as newspaper profits are plummeting and they are scrambling to add online content. Or are you suggesting that mirrors like answers.com ought to be the primary distributors of our content? --Spangineer^es (háblame) 19:32, 15 July 2006 (UTC)

Wikipedia Mantra in Jeopardy

I'm feeling a bit uneasy about this. It looks as though the proposal is advocating the protection of articles, for weeks or months at a time, because the article is just that good. Yes, it improves the reliability of the article, but whatever happened to the encyclopedia anyone can edit? I'm worried the articles anybody can edit will be hidden behind this perfect version (which remember does not exist) and it could take weeks or months for changes to actually be incorporated into the "perfect version". I fear this dystopian end to Wikipedia as we know it when, ten years down the line, only administrators can edit the articles and other users must request changes in development versions. So, I'm a bit torn; I would like Wikipedia to appear more reliable, but I'd also like to adhere to the anyone can edit principle. Perhaps keeping the anyone can edit version as the primary version and making the stable version secondary might work (i.e. mention that there is a stable version available in a header template). Or perhaps we could maintain a location for all of the stable versions. But, please, don't take the anyone can edit away; that's all we have left... for the people of the tomorrow... for the children. Think of the children... the children... joturn e r 19:34, 10 July 2006 (UTC)

Anyone can still edit, and there will not be any merging. The proposal doesn't permit forking. The stable version should track the development version, although you are correct that there will be some delay... If a stable version isn't synced to the development for months then the article should be destabilized, and the proposal says exactly that. The idea of making the stable version non-primary has, of course, been considered. But there are a couple of issues with that... People would fail to update it (because it doesn't matter to almost anyone, since readers wouldn't bother to check it, the same is not true the other way around because you are supposed to update the stable copy, and people care about the content the readers get being accurate), and making the stable non-primary would remove the primary advantages of this proposal (that content unseen by any editors other than it's author is being handed to readers with the full authority of Wikipedia behind it). :) If I ever have children, I think I'd like it if the chances of Cheese being replaced with pages of Image:Penis.jpg were greatly reduced. :) ... When you think about dystopian futures, consider an alternative where all pages end up semi-protected. Would that really be better? Consider an alternative where people fork Wikipedia and the most popular version is a completely uneditable, but reviewed for accuracy version, would that be better? The ability to edit is critical, but we can't forget quality. I think we all need Wikipedia to *be* more reliable, not merely appear more reliable. We can't assure reliability in advance, because we don't control our contributors... Today we catch reliability after the fact, and we're pretty good about it... but that doesn't happen until wrong information has been distributed to potentially large numbers of people. --Gmaxwell 19:54, 10 July 2006 (UTC)

Something I like a great deal about Wikipedia is that it encourages people to ask questions and dig into the history and talk pages when they have a question...rather than rely upon notions of "arguments from authority". Plus I sell people on participation in the project by pointing out that even fixing a spelling error is helping. So as a matter of philosophy, I'd agree with joturner that—if anything—the existence of the stable version should be mentioned as an aside while displaying the "development" version. This would be similar to existing tags like "a recorded spoken version of this article exists from a prior snapshot". Metaeducation 21:18, 10 July 2006 (UTC)

I don't have a particuarly strong opinion on wheather the stable version is at a subpage, and the development version at the page title, or vis-versa. What I like about this proposal is that it allows us to distinguish versions with some consensus from whatever-the-most-recent-editor-put, which we can't easily do at present. The location question seems like a symptom of the constant tentsion between "Wikipedia is a project to write an encyclopedia" and "Wikipedia is an encyclopedia". I tend to lean more on the first, but I don't feel particuarly strong about it. In any case, it's important to remember that articles remain freely editable by anyone; there is just an alternate stable/released/consensus version available. (this was written somewhat earlier, I forgot to sign it at the time) JesseW, the juggling janitor 23:42, 10 July 2006 (UTC)

Joturner wrote "I fear this dystopian end to Wikipedia as we know it when, ten years down the line, only administrators can edit the articles and other users must request changes in development versions." But this is exactly what I think Wikipedia needs, and not ten years from now, but now, right now!

Well, not quite. I can see the advantages in both the classic anarchowiki model and a Britannica type centrally-organized development model for building an encyclopedia. What I'd like to see at Wikipedia is an intelligent and well thought out compromise between these two extremes of sociopolitical structure.

One crucial point which I think long-time Wikipedians (I have been here for 14 months, so not very long-time but rather active) seem to often overlook is that open-source web projects tend to evolve very rapidly, and any tendency to cling to outmoded ideology tends to put a project in danger of earning Darwin's award: extinction.

I have been talking (on my own user talk page) for some time about a (vaguely formulated) scheme in which both editors and articles would have grades. Higher grade editors could perform more actions, such as editing higher grade (more stable/reliable/reviewed) articles, but in general would only have elevated privileges in their areas of demonstrated interest/knowledge/ability.

Editors would increment their grade gradually by persuading some number of higher grade editors that they have demonstrated sufficient knowledge and writing ability. (For example, I feel that one of the biggest but least discussed problems with the traditional wiki model is that well-intentioned but inexperienced editors frequently insert new material without regard for previous organization, notation, terminology, spelling conventions, paragraph structure, verb tense, general style and balance, segues into the next paragraph, and so on, sometimes even turning nearby sentences into nonsense. Higher grade editors would, I hope, be sensitive to such issues and able to recognize similar awareness in an up-and-coming contributor. Similarly, I hope that higher grade editors would be able to wisely and competently assess the factual knowledge and library research abilities and tenacity of up-and-coming contributors.)

In my scheme, some number of editors of sufficient grade could decide to increment or decrement the grade of a specific article. All grade changes (both for editors and for articles) would be incremental and, I hope, would occur at a measured pace. This part of my thinking, at least, is close to Jimbo's insistence on slow and conservative changes, although as noted above, in a rapidly evolving collaborative and global web project like Wikipedia, I think that one should avoid clinging too long to elements of ideology which have outlived their relevance and may even have become detrimental to furthering the goals of the project. My scheme attempts to encourage the rapid improvement of stubs while preventing the rapid degradation of mature and high quality articles, such as "feature articles" and "good articles".

One thing such a system would address is the profound instability of Wikipedia-- a phenomenon which, as noted on one of my user pages, has been decribed by quite a few outside reviewers of the Wikipedia. Gmaxwell's proposal would also respond to such criticisms. I'm not really here to discuss my own proposal (but please hop over to my user talk page if you have any thoughts to share!), but to applaud the most notable feature of Gmaxwell's proposal: his templates could be used immediately. I am glad to see that everyone appreciates what an advantage this is, particularly at WP, where the pace of policy change often seems glacial to those of us who desire immediate and effective action to correct some of Wikipedia's most lamentable deficiencies.

I'd certainly like to see the WP community give it a tryout in the very near future, perhaps in some specific area or areas of subject interest with a dedicated WikiProject base, like WikiProject Mathematics, whose membership already includes a number of active admins.---CH 05:35, 16 July 2006 (UTC)

Bad idea.

<brion> i'm working on stable versions. target is later this month/next month.

<avillia> Well, that's a bit troubling, considering the very shortcut way the policy wishes to implement things, and that it's gotten mentioned in the signpost.

<brion> don't trust software news in the signpost :)

Considering this, it's a horrible idea to start this quick hack. There is not a urgent need for stable versions; We've lived for years without them, we can wait another month under scrutiny. The effort is appreciated, but farrrr off. --Avillia ^{(Avillia me!)} 03:20, 11 July 2006 (UTC)

And why wouldn't this "quick hack":

Let us test out the policy-type infrastructure we'll need once a software solution is deployed anyway? The software isn't going to solve the hard problems like deciding who gets to say when a version is stable.
Be easy enough to dismantle and/or port to the software solution when need be? —Bunchofgrapes (talk) 03:28, 11 July 2006 (UTC)

Duke Nukem Forever, world peace, stable versions, Wikimedia unified login... I'm not blaming the developers, but this is long overdue. Stable versions is one of the things that have been, in one form or other, "only a few [weeks/months/...] away" for years now. Our reputation is suffering. There's no reason not to start this now, and we can move on to a better solution once it exists.

I agree with Duke Nukem: experiments like this proposal (SVN) are terribly overdue and WP urgently needs to try out those which are easiest to implement quickly on an experimental basis (SVN certainly qualifies for serious consideration on these grounds alone!). I hope SVN is given a good hard try over the next few months, and that if successful it will be widely adopted. Won't solve all the problems but it is urgently required to begin immediately to seriously address the most seriously problems. ---CH 05:41, 16 July 2006 (UTC)

Thoughts

We really really do need some sort of stable/static version, separate from the wiki versions of articles - without it, we can simply never be regarded as authoritative. The wiki is a means, not an end, and while it's perfect for collaboratively writing articles, it's no good at all for managing articles which have reached high quality. I like this proposal, as far as it goes, but have some ideas on how it could be better.

It's beyond the scope of this proposal, but having stable articles at a visibly different web address would, I think, enormously increase the PR value of having them. Something like static.wikipedia.org/Article_title or en.wikipedia.org/Static/Article_title. Then the reader is more immediately aware of whether they are reading a 'finished' article or a development one.
On the whole I think there's an argument to be made for using the word static instead of stable. Stable implies that the rest of the encyclopaedia is unstable - not really accurate as it's not about to collapse or change drastically, but more likely to be in the usual process of evolution and enhancement.
This proposal would give editorial control to admins. That would make adminship a much bigger deal than it ever has been. Better, maybe, would be a new class of editor to adminstrate this side of things, selected on the basis of high quality contributions like writing FAs.
See also a similar proposal at WP:STATIC. Worldtraveller 09:30, 11 July 2006 (UTC)

Interesting ideas. Point one I feel is pretty clear, but you're right, that's something that can't be done at the moment. No reason not to go through with this proposal in the meantime though. Using the word "static" makes sense to me; "stable" is just more ingrained right now.

I'm not sure about point three. At first, this "new class of editor" must be a subset of the admin class, because only admins can protect pages. That might not be necessary in the future, but it might be tough to build something more complicated into the software. While I've written several FAs myself, I'm not sure that we would want to limit it to people like me. How many people have written more than 2-3 FAs? Not more than a several dozen I believe. That doesn't leave too many people to be making all these decisions. Thus it might not make sense to restrict it to FA-writers, but rather to have some sort of voting/discussion process where a candidate gives several examples of his/her work and others judge his/her understanding of what makes a good article. This adds another level of process though.

What about responsibility? The point of these static versions is to prevent errors from appearing in the article, but what if one gets by? If it's just an admin judging consensus, it's not really anyone's fault. If it's someone with some level of "editorial authority", would that person be responsible for the mistake? What are the implications of that? --Spangineer^[es] (háblame) 12:12, 11 July 2006 (UTC)

While I think many of us agree on some of those points, Worldtraveller, there is considerable value to a process that can work now. Rather brilliant in fact Gmaxwell, it's obvious in hindsight, but never got going before. Meets a good definition of smart innovation. After a little time to iron out kinks (ie does anyone have an answer for the subpages considered bad issue) we should go ahead with a few articles to test, then slowly build this out. I think the advantages will be obvious. Lets not let something that's not perfect keep us from moving forward. - Taxman ^Talk 15:23, 11 July 2006 (UTC)

Why is this so imperetive to roll out now? --badlydrawnjeff talk 15:39, 11 July 2006 (UTC)

We've been discussing the need for stable versions for a long time, and now someone has a decent idea to implement something that could work immediately. What better time to test it than the present? --Spangineer^es (háblame) 15:43, 11 July 2006 (UTC)

Because this isn't a very good proposal? Because a lot of major issues - such as how this is actually going to encourage editing - haven't been dealt with? Because, judging by the stability fliers taken before, we're simply not ready for it? We can't even get featured articles right, why are we doing this? --badlydrawnjeff talk 18:13, 11 July 2006 (UTC)

I'm not saying "let's implement it permenantly, right now", I'm saying let's test it and see what happens. It's all speculation until we know what the problems are. I'm afraid I don't know what you're referring to when you say "stability fliers"—have processes like this been pushed hard in the past and then failed miserably? I haven't seen any of that. And what relevant problems do you see with the FA process? --Spangineer^es (háblame) 18:50, 11 July 2006 (UTC)

WorldTraveller suggested a separate website. I think the seperate website should be called wikispeech.org and it should be like an international version of Campaign Wikia, only not-for-profit, and stressing securely anonymous discussion of possibly controversial views on sociopolitical and other conentitious issues. See my attempt on my user page to distinguish between two conflicting missions: building a better Britannica and building a better blog, where I argue that these are both laudable goals, but they are very different, require very different policies, ethos, and even differing software solutions, so it makes good sense to split up these missions which I feel tend to get confused in current "Wikipedia" culture. In contrast, I think SVN solidly belongs to the mission of building a better Britannica and therefore it should be experimented with at wikipedia.org. However, I do think it would be a good idea to write the templates so that the stable version gets a distinctive, recognizable, attractive, but unobtrusive background (light pastel?-- would have to test with many complicated articles to make sure it wouldn't conflict with very many figures or whatever) or banner. ---CH 05:48, 16 July 2006 (UTC)

This might be getting off-topic, and might also tend to conflict with the most attractive aspect of SVN, the "Now" in the title (Brion would know more!) but I tend to think that from a social aspect it might make sense to set up an "inventory" of "capabilities" for each registered user. My hope is that this concept would be immediately understandable to gamers and other initiates of the World Wide Web. The idea is that nonadmin users could be awarded by existing admins certain specific capabilities. For example, an admin who happens to be familiar with my own work could award me the capability to carry out SVN actions in my own area of expertise/interest, Math and Physics. Again at the risk of adding to the technical complexity, I tend to think that it might be good to at least consider the possibility to awarding stabilization/destabilization capabilities by the basic subject areas rather than "globally". ---CH 05:57, 16 July 2006 (UTC)

Is anyone else worried about a mentality change here?

Article stabilization causes major changes in Wiki mentality that seem detrimental to me:

The first is a micro-scale change: an article that has been stabilized seems "done". Why work to make even minor improvements in an article when you don't know when or even if they're ever going to get included? Will the article be restabilized in a week? A month? Besides, if the community has put the "Good Enough" stamp on it, why should I put more work into it? Stabilized articles will have a significant drop in the number of editors working on them.

Article stabilization separates administrators even more from other editors. In theory, administrators will go through the same process others would to make changes to a stabilized version. In practice, administrators would Be Bold and make minor changes that they deemed necessary -- missed spelling errors at first, then, hey, maybe a grammar change here or there. If news breaks on a story, if somebody is born or dies or is elected or impeached or wins an Oscar, admins will probably feel pretty comfortable going in and updating that information. Nothing to get de-sysopped over... except it will mean that, in order to really hold the keys to the castle, one needs to be an administrator. Sysoppery will no longer be "not a big deal." It will be a huge deal.

This is my biggest concern about stabilization, this process implies to the outside world that the vast, vast majority of our articles cannot be trusted. It is an admission that the Wiki process is so flawed that you cannot trust an article that has not been certified, stamped, signed off on. Do you know how long it would take to certify the articles we have now, ignoring the 1,800-some articles created every single day, especially considering that any significant content dispute would apparently cause the stabilized version of an article to be deleted? In the interim, we are saying, "we've created this other class of article because you simply cannot trust the rest of the work that we've done." This smacks of a retreat to Nupedia. If you think we're doing ourselves a PR favor by saying that 99% of our work -- that's assuming we can stabilize 10,000 articles -- cannot be trusted, you're mistaken.

Worse, this opens Wikipedia up to more criticism, not less. What happens when an egregious error gets "stabilized"? Or worse, a hoax? This week, we saw that Wikipedia misreporting the cause of Ken Lay's death for six minutes made it into news outlets around the world. What happens when, in a hurry to get over that 1% bar I've mentioned above, a well-meaning administrator "stabilizes" something libelous about a public figure? How many news articles will there be with the headline "Wikipedia's Attempt At Oversight Fails"?

If we want to put more of our weight on our "certified" articles, as I've already said above, we can simply make accessing "certified good" and "certified featured" versions more accessible without making Wikipedia even slightly less dynamic. We already have a process in place to say "we trust the content of this article." Let's keep the Wiki process the dominant, first-to-load feature of this project, and allow editors to see the "seal of approval" version if they're not as trustworthy of the "straight-off-the-wikiroom-floor" version. There is absolutely no reason why we should turn this project into Nupedia when we've come so far by having faith in building an encyclopedia anyone can edit. JDoorj a m Talk 22:29, 11 July 2006 (UTC)

First point - may or may not be borne out. The more complete and finished an article, the less people need to work on it anyway, so I don't see this as a problem. A timescale over which new changes are integrated from development into stable versions ought to be established for clarity.

Second, I agree, it would make adminship a big deal. I don't have any problem at all with a class of trusted, capable users to maintain static versions, but agree maybe admins aren't automatically suited to the task. I don't think it would be too hard to establish a set of non-admin but trusted users.

Third, well the perception among our critics is that no article at all can be trusted because it might just have been vandalised - a perfectly valid criticism of a wiki, and one that can't be rebutted without static versions of articles. I think it's far more of a PR blunder to insist either that all articles can be trusted, or that we don't need to do anything to counter the belief that a wiki encyclopaedia is inherently unreliable.

Fourth point, well I just disagree here. Errors exist in all encyclopaediae. Static versions would still be easily editable and any error identified would probably be corrected within minutes, compared to the years our rivals often take. I cannot for the life of me see how identifying articles that are guaranteed to meet high standards and have not been recently trashed could open us up to criticism.

General comment - would you mind not bolding so much? It looks a bit shouty.

Further general comment - allowing anyone to edit is great for writing articles - perfect in fact. But keeping articles at a high level of quality in a wiki is like spinning plates. The more you have, the more difficult it becomes. Articles such as Sun, if not watched incredibly carefully, degrade substantially over short timescales because people add to them in well meaning but detrimental ways. Experts wrote it, and allowing continued unhindered access to anyone and everyone means that article will decline in quality without serious effort being expended on watching it. Why is that a good thing? Static versions could solve so many problems. My own essay on this is at WP:STATIC. Worldtraveller 00:30, 12 July 2006 (UTC)

We have static versions: the versions of good and featured articles that were certified "good" or "featured". We can link directly to these static, stored versions from good and featured articles. In fact, my proposal appears similar to a couple of the suggestions in WP:STATIC, except it uses the mechanisms we already have to decide on static versions. No new classes of user are needed. (No new designated space would be needed either, though per suggestions on your page it might be useful.) What is your opinion of linking to the static good and featured versions of pages, leaving the dynamic version as the default? JDoorj a m Talk 00:42, 12 July 2006 (UTC)

It's a viable solution, but I'd far rather the static version was the default view for viewers who aren't logged in (preferences could be set to change this for people with accounts). Making the assessed version remote from the reader seems odd to me, although it's certainly a step in the right direction to at least have a link to it. Worldtraveller 00:51, 12 July 2006 (UTC)

I think now we might be on to something. Make the GA and FA versions of articles the default for logged-in users and for people with accounts; people with accounts can switch it so the default is the dynamic version. For G/FA articles, at the top of the page for IPs it would say,

~~"you are viewing the static version of this article, which has been reviewed and deemed a good article/featured article. Click here to view the dynamic version of this article."~~

This would mean keeping the GA/FA system we have now, and not creating any new user classes. There wouldn't even really effectively be different name spaces for each. It's more like we'd have two different modes, "Editor Mode" and "Reader Mode", with the default in the latter. While it would mean our poor developers would have their work more than cut out for them, the casual reader wouldn't even notice. Now, this still leaves the Increasingly Powerful Admin Dilemma, as, again, admins should probably be able to get under the hood to make minor changes to static pages. It would also mean that, for this small but growing set of articles, anon IPs new to the scene would have to get out of Reader Mode, view the Editor-Mode article, see whether the change they were considering is even relevant given the different state of the article, and then make the edit, rather than simply being able to click the tab and dig into the code. I suppose this could be [[cookie {computer|cookieable]] so that anons could at least stay in "editor" mode for the entire session. It's unfortunate that all of these scenarios seem to be adding layers of obstacles between the newest editors we have and the articles they're trying to edit; I'm also concerned about any move that give sysops even more relative power. But this move would protect our very best articles and put them on display without really removing any editing access from anyone (just making it slightly trickier).

.... ~~or maybe you hate that idea. Thoughts? JDoorj a m Talk 01:16, 12 July 2006 (UTC)~~ On my way home from work I decided this is overly complicated and requires an overhaul of all the Wiki code... I decided to fall back in love with my original idea, and have links from the pages to the stable GA/FAs. We could do that right now, as it wouldn't take any system changes whatsoever (just minor tweaks to FA/GA guidelines and templates) and would probably be a good warm-up step to get people used to the idea of fixed articles. JDoorj a m Talk 02:46, 12 July 2006 (UTC)

I have some comments of JDoorjam's four points:

"The first is a micro-scale change: an article that has been stabilized seems 'done'." Not to me, in fact, if I might use my own expository activity as an example, I find that I have been most enthusiastic about further improving articles which I have already worked on extensively. Typically, I can see room for significant improvement after intense work, but some time must pass before I am ready to implement changes which previously were only a glimmer in my eye. Unfortunately, in several recent instances, I was frustrated by ill-thought radical changes to articles I wrote which I thought were both valuable and highly improveable. It can take quite a bit of time and effort, first to get something written down in an organized way comprehensible to experts, then to render it truly comprehensible to a general audience requires a new round of concerted effort, and a certain "latency period" must be observed. But other editors (who lacked the expertise to see the value in waiting for coming improvements) were too impatient and replaced all my hard word with mathematically incorrect and otherwise misleading claims.
"Article stabilization separates administrators even more from other editors." I proposed above that admins could grant stabilization/destabilization privileges to nonadmin users they know well enough to entrust them with that specific power without futher direct admin intervention.
"This process implies to the outside world that the vast, vast majority of our articles cannot be trusted." I don't agree, but if I did, I'd say: a damn good thing! Don't you realize how badly it looks to outside commentators that some Wikipedians, confronted with a conflict between widely accepted realities and classic arnarchowiki ideology, choose to insist on denying reality so that they can cling to a clearly counterfactual item of dogma? That is what makes Wikipedia look ridiculous to many, not the fact that the brave assertion that wiki articles are naturally attracted to perfection has been proven not be correspond to reality here at Wikipedia, N years into the project. "It is an admission that the Wiki process is so flawed that you cannot trust an article that has not been certified, stamped, signed off on." Yes, exactly: this is the true state of affairs, a fact which is very widely appreciated by the outside world. The first is why WP must change its ways or die; the second is why Wikipedia community is regarded as extremely foolish when it refuses to change its ways in order to adapt to a rapidly evolving world.
"Worse, this opens Wikipedia up to more criticism, not less." I think you might have missed a key point about SVN: it slows down the pace of major changes to stable versions, which tends to prevent the kind of hasty error (or whatever it was) which briefly occurred in the Ken Lay article. Which is worse: the hypothetical possibility that someone who has granted stabilization privilege by some admin will accidently/foolishly/maliciously stabilize some character assassination, or the demonstrated possibility that currently inflammatory and false statements can appear at any moment in even the most visible WP articles, and even survive for months despite the alleged army of watchful eyes? ---CH 06:31, 16 July 2006 (UTC)

more freedom lost

This is nothing more than endorsing a version. WHat's next? Corperations paying to have their version of an article stabilized? This just another step closer to the end of an "anyone can edit" encyclopedia. American Patriot 1776 03:18, 12 July 2006 (UTC)

Ok, here's the not hot head version. Yes we need a somewhat stable version but it would have to go through a vigorous examination. More so then an FA or RFA. If we start slapping around stable carelessly, then thinga are going to get messy. Try it out for a bit. American Patriot 1776 03:21, 12 July 2006 (UTC)

Do you think reverting vandalism is bad, too? —Bunchofgrapes (talk) 03:22, 12 July 2006 (UTC)

I beg your pardon? American Patriot 1776 21:28, 12 July 2006 (UTC)

Reverting vandalism can be seen as a restriction of a vandal's ability to edit. Taking "the encyclopedia anybone can edit" too literally can lead to some odd conclusions. —Bunchofgrapes (talk) 21:33, 12 July 2006 (UTC)

Ok... restriction of people who add useful content. American Patriot 1776 21:37, 12 July 2006 (UTC)

I strongly support giving SVN a try, but FWIW, I too would be horrified at the very idea of corporations paying to slant articles in their favor. Actually, that is surely already happening due to shilling and guerilla marketing, and probably on a large scale, since manipulating the WP seems to be one of the few known ways of raising your Google page rank. I've been spending quite a bit of effort monitoring bad edits to the physics pages since last September and have uncovered quite a bit of malfeasance, but not much which looks like guerrilla marketing, but that is probably due to the nature of the subject matter. I have seen quite a bit of evidence that "inventors" of dubious "energy from the vacuum" schemes have tried to promote their "products" at WP, which tends toward attempted fraud, in my view. One point here is that the most determined attempts to slant a given WP article need not be paid; not at all. But they can sometimes me motivated by the lure of big payoff. ---CH 06:41, 16 July 2006 (UTC)