So far as I recall, Ainslie’s thesis is that the various “modules” of the brain have hyperbolic discount curves which are then composed to yield an exponential curve. Akrasia is what happens when particularly strong specific impulses spike above the exponential discount curve. Ainslie predicts what you actually see: lots of people making rational decisions punctuated by failures of “willpower” large and small.
I’m also unsure whether you’re overstating LessWrong’s obsession with akrasia. It’s never felt over-generalized. The focus on it seems reasonable enough insofar as LeWers seem to be drawn heavily from students and techies, two groups for whom akrasia can be particularly destructive. So even if hyperbolic discounting is rarer (than I’m still not sure what), the expected negative value of akrasia may be particularly high for LeWers, leading to its perennial popularity.
Well, that’s my point. There’s all these arguments hanging around here and when you take any of the general approaches, like utility theory, you tend to bump into them with nasty consequences. As I said: I don’t really have a way to “solve” this.