I think that it is indeed important, as suggested by the starting post, to define what is the free will we want to discuss about, before discussing it. Among many possible definitions of being free, I think the most significant is feeling free. But since we want to get (at least partially) out of solipsism when discussing in a forum, I would like to go for:

Is it in principle possible to use science to predict choices (or the probability of choices when quantum mechanics is involved) with arbitrary low uncertainty?

The question is of course ambiguous in many ways, but it's a starting point.