To be honest, I fully agree with you. There are too many editors, not just sysops, who feel they are better than the rest. There are too many sysops, then, taking the wrong decisions and not worrying about it. And there are too many people following along, taking their word for it and assuming nothing has gone wrong.
Wikipedia is, as you say, broken.
However, I too am leaving as of today, for a different reason. Process is broken. Notably, AfD, RfA and DRV. These were excellent procedures, but after some close analysis on IRC with fellow editors, I can conclude that they have degraded over time. The importance of AfD and RfA is such that we cannot afford to allow them to degrade, yet we have.
For example, in the current RfA of Rory096, we have a clear case of where the process has fallen apart. I have no doubt that the closing bureaucrat could easily justify granting Rory096 sysop. In fact, I think he should. But with the way RfA is going, it is no longer the 'discuss and reach consensus' system that it was intended to be. It is a raw vote that can lose all meaning. We have editors visiting once a week and voting on every RfA. But what if the situation changes? What if an interesting diff warranting their attention is brought to light?
In this case, their vote would remain the same, as they would not revisit the RfA till it was over. Which presents an interesting point of view. The one week length of an RfA should be a time for an editor to come under the scrutiny of the community, giving fellow editors a chance to determine the worth of the candidate as a sysop. But no, RfA is a straw poll. Based on pure numbers. Numbers which could, in the case of certain RfAs (not unlike Rory's for instance), mean nothing.
My point is not to discount the votes of certain users, rather to point out the worth of a vote from a user who has analysed a users contributions, checked their edits in certain namespaces, concluded their familiarity with policy; as opposed to a user who has spent a matter of seconds reviewing their edit count and the length of their time on enwp and then voting accordingly.
The problem we have is that while there can be no difference in the worth of these votes, it is obvious that one should place more weighting on the one from the user who has taken the time to think before they vote. Naturally, this is impossible for the closing bureaucrat. Therefore, the only options are to either a) keep the current broken system of straw polls, or b) fix it and move to a system that aims for community based consensus to be achieved. (Alternatively, adopt karynn's proposal at http://nonbovine-ruminations.blogspot.com/2006/10/proposal-for-adminship-on-... - however, I don't fully agree with this and it seems a little too radical for enwp)
As you can imagine, I would err towards option (b). Some fellow editors have attempted to keep me from leaving, but unless anyone has a convincing argument as to why I should remain, or something is done about these policies, I cannot give myself enough justification to stay. Goodbye, Phaedriel, Buickid, AmiDaniel.
--Draicone