User:Geogre/Version System

Over the past couple of years, I've built up some animus, some admiration, and some esteem. Time to spend it on trying to solve the hardest of all Wikiproblems, the "VFD is broken" problem. I will, therefore, fearlessly propose someone else's idea.

(People wanting to admire my impeccable logic and sparkling prose should use the talk page to this subpage. People who want to complain about it should just bide their time, sharpen their knives, and wait for the soft underbelly to roll to the surface.) (In other words, I'm moving talk on the page to the page's talk.)

Preamble and motto

Motto: Probitas laudatur et alget. (Juvenal, Saitre I, l. 740)

Preamble: Few phrases are heard as often in the hallways of Wiki-central than "VfD is broken." I submit that we almost all agree with that, and, if we don't, we will soon. However, once we get past complaining that VfD, like the alimentary canal smells and looks bad and has unsightly overflows, we stop speaking the same language with each other. Even before we get to "the answer is," we begin speaking dialects, then languages, that are incomprehensible to each other except for their tone. "My" proposal (inverted commas to be understood any time anyone refers to this as my proposal, even if it's me), first of all will not end VfD. It will, second of all, attempt to address the fundamental issue of scale that certain Wiki-admins see as a mathematical doom placed on the project at the moment of its conception. Third, it will not be a "Sysop power grab" (a phrase delightful in its nonsense but aggravating in its vehement repetition). Fourth, it won't take the community out of the community decision making, but, at the same time, it won't mean a single extra vote. Fifth, it will teach your dog to speak Linear B.

It really should do four of those things.

More preamble: Authentic VfD!!!! (please to note the sarcasism)

Premises for this argument are numbered. Conclusions are numbered in the stars.

VfD is votes for deletion. It is, therefore, by intent deliberative.
VfD's deliberations are not deliberations on whether the deletion guidelines should be changed or must be changed, but rather whether any one article fits into those deletion guidelines and what to do about it.
VfD's consensus can chart new interpretations of deletion policy as they apply to previously unseen types of content, but they cannot determine that extant policy decisions be reversed or interpreted in one way or another.
VfD is therefore applicable to articles about which there is debate, about which there is uncertainty, where a decision must be reached about this current example (X) fits into an existing set of rules or not.

So, VfD is therefore intended to be slow. It's deliberative and requires votes. Although Wikipedia has not yet instituted the most obvious requirements of a democracy (i.e. quorum), previously administrators closing out VfD debates would re-nominate articles that had not achieved many votes or which had such a confused voting pattern that dispensation in accordance with community views was impossible to determine. It was never intended to result in no alteration of the deliberated article unless the votes were, in fact, to leave the article as-is. Once on VfD, the article had to be changed in some way pending a consensus keep (not a lack of consensus keep). Consequently, it was slow, very slow, and the people who dealt with clearing old VfD's were heroic individuals who had to put in a monstrous amount of work to do the job.

Obviously, the inherent pace of the process means that VfD cannot answer all of its requirements if there is a large number of articles to be considered. Even if we were to assume that 100 articles a day could be viewed, considered, and voted on by at least 5 disinterested voters, then the sluggishness of clearing the backside of VfD would mean paralysis with such numbers. What has happened as VfD load has increased has been inevitable and regrettable.

1. Voters have decreasingly been deliberating or disinterested or sufficiently numerous.
2. "Political parties" have developed that marshal mass votes without regard to article content.
3. Those clearing VfD have decided (and this is new, Tony) that failure to have consensus for delete or merge is the same thing as a consensus for keeping, when consensus-keep, consensus-delete, and consensus-merge are different things altogether, and it is as impossible to find a "rule" saying "consensus = X%" or "failure to achieve consensus specificially for delete or merge is a consensus to keep" as it is to find a specific rule saying "The plurality is honored in the event of a lack of consensus." The reason is that voters and arbiters are increasingly law-gaming, as they see VfD as non-deliberative and, instead, a venue for instant death and immortal life decisions.
4. Speedy delete has expanded to relieve the bulge in VfD.
5. Mis-tagged speedy deletes have expanded to relieve the bulge left in VfD after the expansion of speedy delete cases.
6. "Rogue admin" deletions are freely admitted by virtually every administrator on the project, and very few are apologetic about it.

These responses have been inevitable, silent, and without deliberation. They all arise from the simple fact that (and here I'm going to propose a grandiose statement), Votes for Deletion functions only when it has a small number of articles, when the voters are disinterested in the topics, and when the voters have some knowledge of and agreement with Wikipedia's policies and aims.

The Fix (is not in)

The "Version system" was proposed long ago. I knew it as Angela's idea, but I'm not sure it originated with her. However, it was cooked up around the time that the talk of "Wikipedia Version 1.0" was taking place, so that means that it was more than a year ago at this point. At the time, it was uncontroversial, but it required a great deal of social change on the part of Wikipedians and some slight software tweaking ("slight" to me, because I can't program anything more advanced than a DVD player). Here is the base of it:

Each article contains a rating. This rating is 1-5. The rating is determined not by an editor, but by a software computation of an average rating by editors.
New pages would be automatically certified as a 3.0.
Readers would employ the article Talk page to record a rating in a dynamic form. Administrators would have the ability to override ratings (see lower, later). (See Talk page contents.)
Any article with an average score of 4.0 or higher would be moved to a VfD/deliberation.
Any article with an average score of 4.5 or higher would become a Candidate for Speedy Deletion.
Any article with an average score of 1.5 or lower would become a Featured Article Candidate.
Any article that gets voted for a Featured Article status would have its score lowered to 1.0 or less.
Any article that failed VfD would only have its rating overridden to 4.5 or 6.0.
Articles with a score of 5.0 for more than a week would be immediately queued for deletion.
Articles with a 6.0 would be put instantly on an Articles Needing Action list, where merge/redirects or rewriting or vandal-swatting would take place. (Note these are not deletions. These are "non-delete VfD conclusions.")
Any article entirely rewritten or acted upon in remediation would have its score reset to 3.0.
Since these scores require either an administrator's direct input ("overwriting") or an average score, nothing would happen until a quorum of reviews had taken place. (See Quorums of Time or Hands.)

That is the essence of it. VfD would still take place. However, any administrator could jump up and redirect an article to "Needs help," and anyone actually working on the article can defer its deliberation for death or immortality by doing the work. Obviously, some people are already upset, but just wait until the details, because that's where the devil really lives, as we all know.

1st Circle of Bedevilment (Talk page contents)

The most important thing for any new article, and one of the major software tweaks, would be that all new pages have a talk page already created from the start. This talk page would have dynamic ratings systems already onboard (bigger tweak), and it would have to have a set of criteria delineated to guide voters/evaluators.

The easy part is deciding what text to use. My own view is that the guideline telescope to reflect the seriousness of some numbers over others.

For example:

Rate this article a 3.0 if you believe that it is informative and helpful to those needing more information on the subject, and yet if you feel that there is room to add contributory information.
Rate this article a 2.0 if you believe that it is one of the best articles to be found on Wikipedia and if you believe it exemplifies the kind of writing you could find only in the finest encyclopedias.

For unsatisfactory articles, consider the following:
- Rate this article a 4.0 only if you believe that it currently violates the terms of the Wikipedia: Deletion guideline. Please be sure that you review the guidelines and that you can explain in what way you feel the article currently fails the standard for inclusion in Wikipedia.
- Rate this article a 5.0 only if you believe that it is a legal or ethical violation to carry it, if you believe that it is nonsense, or if you believe that the article is currently a fit candidate for Wikipedia: Candidates for speedy deletion.
- Please note that neither of these ratings is to be offered simply for disagreeing with an article's information or for believing that the topic is not worth covering. You are free to change articles as you wish, and you are free and encouraged to attempt to persuade editors of your point of view on this talk page. Rating an article a 4.0 or 5.0 is not an acceptable alternative to talking, and excessive mis-voting can be used as evidence in a Requests for Comment or Requests for Arbitration on your actions.

End example

The point is that the talk page should explain to readers very, very clearly that the guidelines have not vanished. Note that a rating of 4.0 will not delete the article. It will simply get it listed on VfD, where it will still need the usual deliberation.

2nd Circle of Bedevilment (Quorum: Hands or Time)

One of the most critical vulnerabilities of a system like the version system is the ability of irresponsible army/party votes. Let's suppose, just for chuckles, that there were someone who disliked a group. Let's just say that this person, we'll call him Adolf, dislikes people we'll call Jews. Now, it would be possible for Adolf to go to every single article that had the word "Jew" in it and rate it a 5. We'd catch him, surely! Right. Now, let's say that he sockpuppets. We'll catch them, surely! Ok. Now, let's say that he has a lot of friends. Or, let's knock this down a step and suppose that it's not someone who be a hater, but rather someone who be a lover. Let's suppose that the general view were that, oh, let's suppose high school web comix hosting on free sites were out because they don't have enough fame or significance to need encyclopedic coverage, and then let's suppose that the web comix in the bathroom authors all support each other. They could inform each other of every new endeavor and vote each other a 1.0. Note, they'd vote a 1.0 to offset what they knew was coming -- a series of 4.0 votes. Simple eagerness and activism would thereby destroy community consensus.

So, there are several ways around. One is eternal vigilance. The other is quorum of time (not votes). If it simply takes one month before a ratings would push an article over to a dispensation page (VfD, FAC, CSD, ANA), then that head rush of eagerness from enthusiasts and adversaries will almost surely have died off and, if things work properly, others would have had a chance to vote, and it would even be possible for fishy voting to have been reported and investigated.

On the other hand, time doesn't seem like that keen a solution, if a month goes by, and there has only be a pair of diligent readers who have had anything to say and they were both on illicit drugs. If there is no variety of voice, the odds of a genuinely perceptive dispensation would be very, very low.

Therefore, it is part of this proposal that quorum be a feature of both votes and time. There must be ten votes, and there must be a month, unless the page were moved to speedy dispensation by administrator's intervention. (Administrators have the capacity, in this scheme, to overwrite each other and other users in extreme cases. A libel page can still be just plain deleted by an admin who sees it, even if the author is at a dynamic IP and phones up every 10 minutes to vote it a 1.0. (See "we'll catch him, surely!"))

New categories, new fun (Articles Needing Attention)

One feature of the new system is that every article would have a new category-like tag, and that's the vote score. It would be possible for folks up to no good to cruise the Score:2 categories to see if there are any good articles to read (that's no good, because we know that any article you don't edit is bad, bad, bad, and we should discourage people from writing so well that other people can't add taxoboxes). It would be possible for people to cruise Score:4 categories and attempt to fix things. Voila! A new form of Wikipedia:Clean Up! It would be a very, very good thing for people to cruise Score:3 articles to see what still needs evaluation (all the fun of expressing your VfD vote, but without the "OMG YOU DELETIONIST TROLL" comments).

Most of all, though, it would introduce a Score:6 category. Score 6 would be reserved for articles that need desperate attention. Either they have no clear way forward, or they have split the community. These are the sorts of articles where we can do our best to help out Wikipedia. It also relieves the anxiety and work placed upon a solitary admin closing out VfD.

Of all the parts of "my" proposal, this is the part I like most.

Porky Pig says what?

That's all I have.

Please to note that I have no intention of arguing for this proposal. I've done that before, and I have so little confidence in the rest of the project when it comes to a vote that I doubt I'll even look at how the debate is going. If, one day, I log on and see that the system has been implemented, I will be pleased. If I don't.... What am I saying? If I don't, it only means that I haven't waited long enough, because surely someone will fix it.