Fire-brigade policing

I really like Rob Reiner's ideas on how we should conceptualise policing. Seems to me this is both an accurate reflection of what the police actually do (i.e. almost anything, when called on) and a useful way of calibrating our expectations about what is possible in terms of real achievements.

That is, while police can and do solve local problems, and may, in certain situations and in certain ways, influence crime rates and even 'put a lid' on crime, there must almost by definition be little they can do about the underlying causes of crime. These are likely to be so deeply embedded in diverse social, economic and cultural structures that it seems almost naive to expect that one single state agency could, on its own, have much effect on them in the long run.

Most people would, presumably, agree that police should work toward this larger goal in conjunction with other state agencies, non-state actors, and the public at large. So why, then, should the success of the police be measured only by it's success in  'fighting crime' (which is what the current government seems to be suggesting). Indeed, if what Reiner says is true, and given the well-known difficulties involved in tracking and explaining changes in the rate of crime, what what be a suitable metric for assessing police performance purely against a crime-change criteria?