Photo: Flickr user lifeontheedge

Thursday, December 29, 2005

Wikipedia is not a cesspool of poison-penned libelists. It is not a tabloid. Should they chance upon vandalism, most people will be smart enough to figure out that "alex smith = gay!" isn't the fourth step of the Krebs cycle.

But Wikipedia does have a problem (well, a few) : while bushels of isolated facts stick to an article like bugs to flypaper, the regular wear and tear of wiki life (revert wars, adding detail, etc.) tends to atrophy an article's readability until, like beach footprints, it's reduced to a barely-differentiated indent. (Well, not an indent. I should stop being so metaphorical.)

Articles that have the most frequent edits are the ones most likely to suffer. (In a cruel twist of fate, these tend to be articles that are (or were) the best -- once an article passes a certain quality threshold, it draws visitors from all over.)

If you've been editing wikipedia long enough (and I have) it's a familiar feeling: you go through an article, nipping and tucking until it's brilliant and clear and readable, then you check back 3 months later and it's descended back into the fog.

So what to do?

Stable Pages

There's been a lot of talk about stable versions lately. They'll probably be implemented soon, which is a good thing. Here's how they work:

The very best wikipedia articles are doubled up: one version remains in the Standard Wikipedia, subject to immediate edits by whomever, and the other is preserved, temporarily, under glass, viewable but not editable. This preserved version is updated all at once, periodically.

As someone in the mailing list put it: as long as we're forever pushing the boulder up the hill, why not put a wedge behind it once in awhile?

One of the more important functions of a stable page will be serve as a label for the temporary high water mark -- instead of having to dive back into the history to see whether an article's gotten worse since being featured on the front page, you just have to compare the live and stable versions side-by-side. And there's a definite, well-defined goal for each page, a version number (as it were): Good changes can be periodically reincorporated en mass when the stable page is updated.

With enemies like this, who needs friends?

Tuesday, December 27, 2005

Wikipedia-watch has kicked up a bit of a firestorm by tracking down and posting the names and addresses behind Wikipedia administrators' screennames. The mailing list reaction is predictably tinged with outrage and incomprehension (I share the former), but it's touched off a good discussion about wikipedian anonymity in general.

Steven Ericsson Zenith:

Hidden identity provides no basis for authority since the landscape of individuals is unknown and the changes to that landscape impossible to track. Such that, even if a group of anonymous admins is able to command respect for a period, there is no guarantee, no way to judge, that a group of admins have the same capacity in the future. Indeed, if the current group of admins do manage to establish public confidence then the public is immediately at risk since that group can be opaquely usurped.

Zenith is arguing on some rarified abstract plane, but more concretely: will accounts ever be covertly bought and sold? This seems unlikely -- wikipedia is too large and decentralized for something like that to stay secret (and who would do it? Paging Stalin's airbrushers and the Mossad.).

Fastfission argues that this whole line of thinking is moot from the get-go, that transparency is compatible with anonymity (and that seeing for yourself every edit an author's made on a project makes up for not knowing whether he's living in a bunker in Berlin):

Wikipedia is one of the most transparent enterprises on the entire internet -- it is easy to see in an instant everything a contributor has done, everything they have ever squabbled about, every time they change their mind and any place they might have been discussing a change before it was actually made. It is an easy task to show ever omission, every purposefully false addition, every bit of slander, as it went down, who did it, at what time, at what place.

A couple people add that anonymity is necessary, at least online:

What makes admins so special? A lot of them are mostly janitors. Non-admins do the majority of content edits, which is what Wikipedia's credibility is based on...There are a lot of wack jobs on the Internet that could use that information for nefarious someone who is not yet secure in an academic career I would loathe if people could Google my real name and come up with results of me bickering with cranks online about any of the various subjects I end up bickering with cranks over.

Monday, December 26, 2005

The Doomsday argument is a probabilistic argument that claims to predict the future lifetime of the human race given only an estimate of the total number of humans born so far. The argument predicts, with 95% confidence, that humanity will disappear within 9120 years.

An incredibly detailed history of aviation.