I've long felt that the question of whether we have free will or not is impossible to prove, but that we probably do not. However, I've long felt that as a practical matter, an assumption that we have zero free will leads to a complete breakdown of society. Not having free will means you aren't responsible for your actions, and makes a murderer little different than the tool he used to commit the crime. That's a profoundly dehumanising sort of idea, and leads, I think, to two main problems.
First, there's no point in "punishment". Someone who has committed a crime is no different than someone who will commit a crime in the future - both have, we assume, the same brain structure (or the same fate, or whatever you think drives people). What sort of justice system would that logically lead us to? What sort of society? (Something like that depicted in Minority Report.)
That leads me to the second point. Such an emphasis on crimes being, ipso facto, proof of a sickness which needs to be cured has been tried before. In Soviet Russia, dissent was seen as something not to be listened to, or ignored, but to be cured. It led to some...pretty nasty things.
Dershowitz has written some interesting stuff about balancing preventive imprisonment (which we already have, of course - plenty of people are in jail awaiting a trial, for example) and the crime reduction it would give versus the civil rights violations it implies. He is against the idea of an expansion, and I agree with him. The largest problem (seen also in the USSR) is how one goes about proving that one will not commit a crime. It's hard enough to prove that you have not - but there's no way to really prove that you will not.
Once you accept the concept that people are mindless, clockwork automatons, wound up by their genes, their environment, their neurobiology (or whatever else tickles your fancy), and destined to do...well, whatever it is they're destined to do...how can you not then accept that it doesn't really matter whether they've actually done it yet or not? There's a character in some of Pratchett's novels who sometimes comments that everyone is guilty of something - sooner or later. He uses this, on occasion, to justify lax police investigation procedures, since after all, whomever the police arrest for a given crime will be guilty of something eventually, even if at the moment they're innocent. Sound like a good idea?
Few people explicitly argue that we should try and predict crimes, but that's an inherent part of what many are suggesting. When someone says that we should treat the commission of a crime as evidence of someone's deviant mind structure, this is implicitly arguing that once we have determined (or think we have, anyhow) that someone has bad wiring that we should then act on this. That's heading really quickly down a very nasty slippery slope, I think.
As an engineer (more or less) by training, not to mention by temperament, I'm more inclined to judge by results first, actions second, and intents a distant third. My reasoning could be summed up, at its most basic, by the argument that this is how the universe does things. A bridge either falls down, or it does not. If it falls down, someone has screwed up. The bridge should be standing, now it is not - this is a result, and it's pretty clear. Next comes actions. Why did the bridge fall down? Maybe the engineer forgot to take some force into consideration. Maybe a construction worker used the wrong proportions when mixing some concrete. Maybe an inspector didn't notice some substandard materials. Maybe some combination of factors - but these are a lot harder to understand. We can all agree that the bridge has fallen down (things like that tend to be noticeable).
We can't, however, always agree on what actions led to it. Sometimes it's clear - maybe we can double check the load calculations, and see the engineer dropped a decimal place. Sometimes it's not clear at all - some bit of concrete crumbled when it shouldn't have, the remains look right, and we never work out exactly why the bridge fell down. That's why we shouldn't weigh actions as heavily as results - they're a lot less clear. Even when we can nail down an improper action, trying to work out if it actually led to the bridge collapsing can be hard - yet if it didn't lead to a bad result, how important was it? How can we weigh actions without looking at the results?
Lastly, the worst way of judging things is intent, because it's usually impossible to determine, and even when it isn't, it has almost no bearing on results. Plenty of bridges have collapsed, yet you'd be very hard pressed to find anyone who intended that! Even if, say, the engineer purposefully fudged the numbers in an attempt to make the bridge collapse, in the final analysis, either it fell down, or it didn't. If it did, then it's all rather simple. The bridge fell down; the engineer messed up the numbers; he is at fault. Where does intent come into it? Worse, how would we ever determine his intent? And if the bridge does not, in fact, fall down, it becomes trickier. Did he fail to make it fail, or did he not actually intend to cause it to collapse after all? Has a crime actually been committed? Singer might argue yes, but I'm not so sure. The bridge didn't fall down, ergo no successful actions were taken to make it fall down. Isn't that the important thing?
More importantly, how can we determine the intent of the engineer? I don't think we can - but lets assume we could. I don't think I'd like living in a society that did! And if we in fact cannot, looking back at history, I don't think I'd much like to live in a society that just pretended to measure intentions.
In summary - I think we probably will never know if free will exists. I think free will almost certainly does not exist. However, I think that assuming it does exist is the only way to build a functioning society. If we assume it does not exist, there are several ways things can fail - medicalizing the justice system is one of the more obvious ways. I don't really see a way for that to work.
(Note, by the way, that at no point did I use the word fair. I don't really think the universe is fair - or, rather, that it is infinitely fair. Pi is, as far as we know, the same throughout the universe. The gravitational constant is doing an excellent job at being constant. The weak nuclear force doesn't seem inclined to fail any time soon. This is, I think, about all one can expect of the universe - that the laws of physics are applied universally, and without favour - but I will note that it fulfils these expectations very well. That is, in my view, a very fair sort of arrangement, but it doesn't have a great deal of relevance to what most people consider "fairness".)