Counterfactuals and Probability: A Plea

Dang, missed the whole month of July, sorry: busy race schedule and those
damn comps are killing me...KILLING ME I tells ya.

I *would* have been reading Pollock's _Thinking about Acting_ and Sherrilyn Roush's _Tracking Truth_ if I didn't have to study for those comps. I'll take this study break to
vent about the semantics for counterfactuals since I just uttered one in mentioning a book about them.

I see *nothing* in the use or understanding of subjunctive conditionals to
suggest an analysis in terms of similarity. Surely the best answer to “What
would have happened if (contrary to fact) E occurred?” is whatever has/had the
highest probability conditional on E.

In fact, consider the question: What *will* happen if φ? (with future assumed
open and φ contrary to fact). Surely the best answer is what has the highest
probability assuming φ. Now look to the past rather than the future.: What would
have happened if φ? Why would we change the whole basis of our answer just
because we were looking at the same contrary-to-fact scenario from a different
angle?

Near as I can tell it's just the elegance and authority of Lewis and
Stalnaker's accounts that carries it forward in spite of so many problems.
OK, that's not fair: it works (i.e. gets it extensionally correct) in a lot of
common situations. But it just falls to pieces under pressure. Since
this is a study-break-rant I'll cite as evidence all the intuitively correct,
common sense analyses which get "Shoped" due to technical problems with the
traditional semantics.

One of my greatest hopes for formal epistemology is a probabilistic
replacement for the traditional semantics for counterfactuals.