Here's the most anarchy-positive thing I think I can sincerely argue: I find it to be probable in the abstract that any particular rule X has been extended beyond the situations in which it is more useful than any alternative. Or to shave some words and some subtlety, if you want to tell me that some rule doesn't make sense then I'm likely to agree with you.
Think about the Peter Principle. People get promoted beyond their level of usefulness. But people leave positions and change ensues. Rules, by contrast, never expire and can be adapted to changing situations by being entrenched in layer after layer of compensatory sub-rules, exegesis, or selective enforcement.
The most bizarre thing to consider is that rules themselves take on certain inhuman alliances with other rules. A rule that promotes ABC is compatible with rules that promote AB, BC, & so on. They will Velcro onto each other without anyone's intent to do so.
I think this is what Deleuze was getting at with his work on machines. I think this kind of memeplex is also what many postratd describe as an egregore, but I could be mistaken on that point because I've never been comfortable with the term.
So it seems to me that rules are more likely to be promoted and extended and applied based on their topicality, abstraction, and synergy -- but that these qualities are independent of human judgment or the informed deliberation of the greater good. Judgment and deliberation are supposed to be holding the reigns, but you know how these things go.
Incidentally, this reminds me of the emphasis that early modern republican-liberals placed on the "virtu" of a people (Machiavelli) or concerns like character, judgement, and temper (Washington). I think they were responding to the claim made in favor of autocratic rulers that autocrats (queens, kings, Medicis, et al) had the human sensibility to deliver more humane judgments -- as opposed to the flat, committee-drafted policies pronounced by impersonal deliberative processes.
Furthermore, every extant rule can become extended to new situations for the simple reason that it does not require such an explicit change as the creation of a new concept or the deliberation of one approach over the other. If there's a rule that's already on the books and it's close enough to the situation at-hand, then you can save everyone a bunch of arguments and make an existing rule more generalized. This is a choice justified by ease of process but not quality.