User talk:Doc glasgow/wiki-musings

A page to record my wiki-thoughts - comment as you will:

See also: my (halfbaked) proposal for deletion reform.

Deletion process and growth of the wiki
I conclude that the deletion process is irreparably broken. This is not simply because it is faction-riven and occasionally nasty, but because it structurally incapable of dealing with the number of articles that the community deems worthy of deletion. It is not currently bearing that load, and there is simply no possibility of it bearing that load as Wikipedia grows.

I have been at times involving myself in New Pages patrol - particularly examining pages below 250 on 'special new pages' (i.e. those that have made it through recent changes patrollers). I examine only a very small fraction of new articles. That often causes me to nominate for afd/vfd. Indeed it was pointed out that I was fifth highest nominator over the last few months (shame of it!).

However, I am aware that I could easily have multiplied my nominations threefold, just from the small sections that I was examining - confident of an 80%+ success rate in deletions. I am not talking here of articles on the margins of notability (schools etc.), but stuff that typically attracts unanimous ‘delete’ votes.

Afd is not working. We are straining on gnats and swallowing camels. For every article that we agonize over, two others (and the number is growing) slip through unchallenged. We often spend far more time debating articles than lazy, vain, or frivolous contributors spend writing them.

As the community grows (10%? a month), this problem (if, indeed, it is a problem) will only increase. Afd (for all its glorious noise), and even NP patrol, are fast becoming statistically irrelevant to Wikipedia’s content.

Can it be 'fixed'?
Short answer, I don't think so.

We could increase speedy categories. However, for this to have real impact, we'd need to increase them dramatically. Inevitably, this would lead to fairly arbitrary deletions - as admins exercise personal judgements within inevitably vague guidelines. The VFU process will be unable to cope with the number of 'appeals' against deletions, thus any effective scrutiny will end.

'Trust the admins' is often the response. But this only works because the community knows its admins, and admins generally cooperate (although we currently have 500 admins, in practice Wikipedia is managed by probably only 20% of them - an effective 'cabal', who to some degree know each other) - as it grows, that personal knowledge will end. Just to keep pace with growth we need 50 new admins a month (+10%). (As community grows, even selecting admins may become problematic, as most will be unknown to many editors, and few will be ‘old-timers’.

Does it matter?
Perhaps not, but we should, at least, consider the logical consequences.

Adding every nursery school, university society, internet community, small-time publication, minor academic and garage band may not be in itself a ‘bad-thing’. Perhaps verifiability not notability ought to be the sole standard.

But what are the implications for verifiability, and the reputation of wikipedia, of an effectively un-weeded wiki?

Wiki works on the principle that, over time, POV and error will be eliminated by the sheer number of contributors. That theory certainly seems to work: factual error or unattributed opinion will soon be challenged on articles like George W. Bush. But already much on the wiki has not been scrutinised by anyone save the creator. NP patrol deletes/nominates the patently implausible and the obviously nonsensical - but the plausible hoax has a fair chance of slipping through. If the subject is in a minority interest area - it may never be checked. If wikipedia gets a reputation for factual inaccuracy and unchecked articles, will good editors be driven off and the lunatics take over?

POV is another issue. Unlimited, unscrutinised, page creation may give reign to POV forking; and a disincentive to NPOVing existing articles. Why debate with neo-cons on George W. Bush if you can go off and create Reasons why New Englanders hate Bush and get your own views on the web untroubled by facts or compromises.

What could be done?
1. Nothing (the default)

2. Split the wider community into sub-wikis in various subject areas – thus creating smaller manageable communities of interest (probably unworkable)

3. Restrict ‘New Page’ creations to seasoned editors (forcing editors to log in first is more plausible, but is unlikely to make much difference in the long-run). (not wiki-like)

4. Create some form of ‘Articles for creation’ process first. (drastic, and not at all wiki-like)

5. I can't think of any more right now (unacceptable)

Questions


 * Is anarchy the default or inevitable condition of a wiki?
 * Are the goals of an encyclopaedia (verifiability, NPOV, accuracy) finally incompatible with the anarchic operation of a wiki?
 * Should we delete afd, and give up NP partol, as being as canute-like wastes of time?

Comments
(Toot if you agree)


 * Firstly, I agree that AfD doesn't scale. It seems to reach a pain threshold at around 160 listings per day, which may be appropriate for the current 1500 new articles per day pertaining at present1 but will probably not be appropriate for 2000, 3000 and so on.  Over the past three months the actual number of daily AfD nominations has averaged 112, according to a recent automated survey2.


 * Secondly, many articles listed for deletion are not deletion candidates. For instance, according to the deletion policy articles on minor branches of a subject covered by an existing article should be merged.  If there is no consensus for such a merge then it should be left as it is.


 * Thirdly, speedy deletion will probably scale just fine for the foreseeable future. Deletion of copyright violations will probably scale at the same level, and I suggest that probably hoaxes should also have their own page similar to WP:CP where an article is deleted as a hoax if not verified within 30 days.  Verification is not a matter for consensus, the major subject of an article either is or is not verified.


 * Finally the culture that has evolved on AfD, whose principal effect is to randomly delete articles on grounds not related to the deletion policy, can probably be ignored. It will probably continue but its influence will decline as the size of Wikipedia continues to grow at an accelerating rate.  The people who will influence the future shape of Wikipedia are its editors, not those who try to enforce their conception of the appropriate form of an encyclopedia by deleting articles.  --Tony Sidaway Talk  13:58, 21 September 2005 (UTC)

1 This figure is an estimate from several recent samples of the new pages list at a distance of seven days. Such samples produce a hard figure for the number of new pages created in article space on a given day which have not been deleted in the intervening seven days.

2 See AFD 100 days


 * Afd's purpose is to 'randomly delete articles on grounds not related to the deletion policy'. You're talking about 'notability'. I'm not going to take sides on that one. But interestingly, in Snowspinner's deletion spree of afd's, I thought it significant that he only deleted 16 of 130 nominations as invalid - and some of those (as his edit summaries show) could have been made on other grounds. That suggests, that actualy the majority of afd's deletions are per the deletion policy - they are hoaxes, unverifiable, POV's, forks, or just less than patent nonsense. Kill notability as a criterion and you still need afd - or an alternative. --Doc (?) 00:04, 7 October 2005 (UTC)

More comments

 * 1) Doc Glasgow - your analysis is well thought out. I see Wikipedia as anarchy in spite of founders' comments that it is not. The only solution I can perceive is requiring registration and login before any editing is allowed. I agree that this is not wiki-like, but many vandals have learned how to game the system by logging off their ISP and logging back on to force a new IP address to be assigned. This would not solve all the problems, and it would likely result in thousands more sockpuppets for blocked users, a distinct disadvantage without verification of identity for issuing login, even less wiki-like. There is no good answer. Especially since it appears that more growth is happening with vandals and newbies (who don't want to bother with such things as actually reading policies and guidelines before they edit). It appears to me that we need broader criteria for Speedy Deletion as well, to reduce the AfD load. -WCFrancis 00:33, 23 September 2005 (UTC)
 * 2) Tony - toot! although I would make the hoax/verification deadline much shorter, along the lines of five days. Not chosen randomly, I use that number since it is the theoretical time for the AfD process. This has the possibility of getting (most) hoaxes off of AfD, thus lightening the load somewhat. (put signature in wrong place: I did this at the same time as the above.) WCFrancis 03:27, 24 September 2005 (UTC)
 * 3) Nice analysis, and I enjoyed reading it. To me, the question that Wikipedia will have to answer, and will have to answer soon, is whether Wikipedia is a wiki first, or an encyclopaedia first. If it's an encyclopaedia first, then those un-wiki-like policies you suggest might just have to be swallowed in the name of building an encyclopaedia. If it's a wiki first, then the inevitable failure of AfD to be anywhere near adequate might have to be swallowed in the name of sustaining a good wiki. When the argument comes, it should be a doozy. Lord Bob 02:07, 23 September 2005 (UTC)
 * 4) I think restricting new page creation to trusted users and expanding speedy categories (BTW, see Criteria for speedy deletion/Proposal/Non-existent Internet entities) are good ideas. Also, maybe AFD could be switched to a system more like RFD or IFD, where almost nobody ever votes "delete" - redirects/images are deleted by default after 7 days if nobody has objected. Or the AFD time could be shortened, or articles that get more than X votes for deletion with no "keep" votes could be automatically deleted. N (t/c) 21:18, 27 September 2005 (UTC)
 * 5) I would like to see more discussion of pure wiki deletion, since I am a strong proponent of it. --Ryan Delaney talk 23:09, 6 October 2005 (UTC)