I agree that the adminship process is broken and has been for quite some time. People become admins by making a sufficient number of edits over a span of some months without making a career-limiting mistake. It helps to participate in IRC and to be part of a mutually reinforcing group of people who are also seeking adminship, and there are certain purely mechanical requirements involving edit summaries, minor edits, and participation in various housekeeping tasks.

The amount of time and number of edits requirements are now high enough that they have little to do with understanding Wikipedia. They are as high as they are just to be sure that potential admins have had sufficient opportunity to make a career-limiting mistake, if they are prone to that sort of thing by their nature.

While RFA is more or less functional at a basic level of being effective in getting admins promoted that have received some sort of vetting, there are problems with the ill will it generates and the fact that the project could benefit from more admins than are being promoted presently. Though rare, there have also been some admins that have slipped through that were, in hindsight, clearly not suitable. The fact that RFA has become politicized is also a problem because it means that admins as a group are more predisposed to behave politically than was once the case.

Tyrenius is an example of the problem we face. It's clear from the RFA page (http://en.wikipedia.org/wiki/Wikipedia:Requests_for_adminship/Tyrenius) that he is unaware of the "RFA culture" which is rife with unwritten rules, such as the unwritten rule against attempted rebuttal of an oppose vote, the unwritten rule that you really ought to vote on a bunch of RFAs before nominating yourself, and the unwritten rule that self-noms in at least some peoples' minds must be well qualified.

While I don't necessarily believe that Tyrenius should be an admin at this time, I do believe that he is justified in feeling unfairly treated by the project. The lack of consensus on a minimum number of edits, and the ever-growing minimum in the minds of many, is particularly a problem. We used to require 500 edits.

I'm not sure what all the answers are but the two thoughts that come to mind are the sequenced granting of rights, which I've proposed in the past (in general, granting deletion, undeletion, and rollback first and the other privs more or less automatically after a six month or yearlong "apprenticeship" period), and a sponsorship and mentoring system where existing admins take responsibility for shepherding new editors through the process.

uc