More than is usual for this kind of work, the prose is readable and at times engaging. It prompted a few fond chuckles to see the following:
There are over thirty distinct ways to irritate other Wikipedia users, including being incivil, disruptive, or tendentious; researching the wrong way, attacking others’ gender or race...Ah yes, but aren't there much more than thirty? Surely an enterprising spirit would generate new ones. Otherwise life might get boring.
On the whole, though, their analysis is structurally flawed. David A. Hoffman and Salil Mehra write about arbitration as if it were the only means of banning editors, but of course there are more. The period January 2005 through Septemer 2007 was a critical one in Wikipedia's development of community-based remedies, none of which are mentioned in the study.
To highlight the developments:
- May 2005: Wikipedia formalizes its banning policy.
- July 2005: David Gerard provides a definition of community bans: "Some editors are so odious that not one of the 500+ admins will unblock them." In slightly less colorful phrasing this becomes the de facto standard for community sitebans.
- September 2006: Wikipedia formalizes a disruptive editing guideline.
In the time since then the community has become increasingly proactive in enacting, reviewing, and lifting sanctions. Unfortuantely these are not easily studied because the documentation of these sanctions is extremely diffuse. A page exists to record full sitebans, but not for any other type of community-based sanctions (topic ban, article ban, single revert restrictions, etc.), and the definition of community banning is itself diffuse enough to be disputed: when is an editor banned by the community, as opposed to placed under a block of indefinite duration? Discussion of bans (which may or may not require consensus discussion, depending on who you ask) has roamed across at least three noticeboards. Although automated search tools have been developed in attempt to compensate, they can search only for specific instances where the editor's username is known and the tools may fail to turn up the appropriate result.
Additionally, although the community enacts bans and other sanctions of indefinite duration, it has almost no articulated standards for reconsidering an indefinite sanction. Generally the blocking administrator is held responsible and should be consulted, but there is very little provision for what to do if that administrator is unavailable or under what circumstances sanctions should come to an end. The results of that lack are predictably chaotic.
So although it would be fair to say that a majority of sanctions were enacted by the Arbitration Committee or Jimbo Wales at the beginning of 2005, by September 2007 the minority of editor sanctions were coming from these sources. The nature of disputes heard by the Committee was also changing substantially as the community adapted to handling simple and obvious cases, so by the end of the period under study the character of cases before the Arbitration Committee had shifted toward complex and intransigent disputes for which no easy solution was at hand.
So, setting aside other criticisms (I had originally intended to mention the absence of analysis on wheel wars and other causes of administrative desysoppings, and a few smaller points), Hoffman's and Mehra's attempt to apply complex statistical analysis and game theory to Wikipedia arbitration is fatally flawed.