Right. Studies on deterrents or the effect of sentencing are some sort of wasteland. They tend to cancel each other out. You have one that says it's going this way, and another one for that.
The point I want to make is that if that is the situation, on the basis of research results that are uncertain and cancelling themselves out, you do not increase the potential harm that will be done to the lives of people through incarcerating them. I think that's a very shaky basis. From uncertainty, nothing follows, and certainly not punishment.