Glad to see this topic being hashed out here on DW -- it's been much in my thoughts lately, especially on the commute to and from work.
Hope you don't mind if I throw in some amateurish observations.
The post earlier by Modus.Ponens struck me as the most plausible explanation. In a world governed by causality, we could -- in theory -- know the outcome of any "choice" if we knew all the conditioning factors. (Indeed, this is probably why dhamma posits the Buddha's omniscience). To dispute this would amount to setting up some sort of autonomous agent that is capable of making unconditioned choices. Such an idea presents logical problems, not to mention a dhamma problem.
Practically speaking, though, it's impossible for any of us to know all the factors and thus the illusion of choice remains in effect. From the conditioned POV it always appears that we have a choice to make, and therefore the concept of free will has functional meaning, as Geoff said.
There's another complication, however: our belief in free will is itself one of the conditioning factors. A person who rejects the idea of choice and one who accepts it may act in different ways. If you have two nearly equivalent sets of factors, but one contains "belief in free will" and the other contains "fatalism" it's likely we won't see the same outcomes.
Humans often make apparently illogical and/or unpredictable choices...often out of the sheer joy of being contrarian. Anyone who has been around small children knows this to be true! However, we would not act in this way if we didn't believe in our own willpower; instead, we'd behave like automatons (David Chalmers' zombies, perhaps).
Therefore, although determinism is true in principle, in practice it is subject to observational constraints and something akin to the observer effect.