Police Predicting Criminality in London.
Reminds me very much of Clay Shirky pointing out the destruction of free will. The same principle applies to terrorism, but this is just because crime and terror are the forefront of "applicability".
Even if you don't believe in free will, there are problems with this approach. Feature creep is the main one - if society is, by its cultural (rather than absolute) nature, a problem-solving entity (a fixation on science would suggest so), then all problems, big or small, should eventually lead to such predictive technologies being implemented in all areas.
This leads to the other problems. Firstly, how do we decide what is a "problem" - who sets out the norms that the "rest of us" are to adhere too, and how do they get overseen? (This is, of course, a current issue, only there seems to be less trust in the personal decision-making route than in "scientific" processes.) Secondly, to what extent can we trust the science to make accurate predictions, given that reality may (or, indeed, may not) differ wildly from the models used to predict?
To solve problems with 100% efficiency, all one needs to do is to lock everyone (including the guards) up at birth. This is the epitome of technological problem-solving.