Some Figures on Wikipedia Protection Mechanisms

The recent focus on Wikipedia "failing" or being "closed" merit some figures and explanation. On the afternoon of Sept 04, 2009 the English Wikipedia has 3,024,063 articles.

The Special:ProtectedPages for the Article namespace tells us:

  • 5,137 articles are protected (that's 0.17% of all articles).
  • The majority of those, (3,553 articles or 69% of protected articles), are semi-protected, meaning that while they aren't editable by anonymous users, they are by Wikipedians (i.e., those with an account in good standing).
  • Therefore, only 1,583 articles (.05%) are fully protected, and not available to editing by non-administrative Wikipedians.
  • Of all the articles being protected, 1337 of them (26%) are set to expire, most within a month or two.

That's the status quo. Yet, flagging a vetted version of an article has been discussed since 2005. The current widely discussed idea is to conduct a two month experiment in which biographies of living people (402,672 articles, about 13% of the English Wikipedia), or some subset, are "flag protected"; this means anyone can still edit but the public (not Wikipedians) see the last reviewed version. This doesn't necessarily replace the existing protection mechanisms, but could be a good alternative to semi-protection. The experiment will helpfully give guidance on who should be a "Reviewer," and answer the questions of whether this limits disruption, furthers quality, and how long does it takes to review and flag a newer version of an article. Another part of the experiment is "partrolled revisions" which would apply to a wider swath of articles and permit vandalism fighters to bookmark a known good version so they can easily evaluate subsequent contributions, but it won't affect who can edit or what the public sees.

The goal of this, and other features, is to maximize the benefits of open collaboration while minimizing the damage from disruptive edits. In my opinion, this has always been the case and Wikipedia continues to experiment with achieving the best balance.

Ported/Archived Responses

Waldir on 2009-09-04

Interesting insights, thanks! I also think it would be nice to point out that apparently only 20 articles, of all 3M+ in enwiki, are fully, indefinitely protected:

I said "apparently" because, by your numbers, we'd have 1,583 - 1,337 = 246... am I missing something here?

Joseph Reagle on 2009-09-04

Oh, and part of it is you didn't include semi-protected indefinitely in your figures.

Joseph Reagle on 2009-09-04

Hrmm... those 20 indefinite are quite interesting. However, I expect that no "expires" does not equal "indefinite." By that, I mean when you look at Special:ProjectedPages, you see entries like: "Diane Dixon‎ (1,519 bytes) ‎(semi protected, expires 00:00, 19 December 2009 (UTC))". That's what my count is: 1,337 of the entries have an expiration date. However, I supposed that doesn't mean the other 3,800 (5137 - 1337) are necessarily indefinite.

Comments !