OK, I'll concede that point -- the initial focus on him was probably not random selection. But they had to have evidence beyond simply the prior conviction (which, as you probably know, would not even be admissable at his trial). That brings me back to my point. Is it so easy to build a case against an innocent man that the cops could simply pluck off the street the first convicted rapist they could find and, BINGO, get a conviction? That, to me, is a tough proposition to sell.
I am far more ready to believe that the science of DNA typing and matching may not be as advanced or as foolproof as we have been led to believe. I think that if you have evidence against a suspect and then are able to match his DNA, that probably is a clincher. However, I am not persuaded the process works as well in reverse.