They are vicious because they are incentivized to be so, their success is determined by their prosecution rate. If you incentivize good people to do bad things they will do bad things. The question is, what other metric can you measure them by to determine how "successful" they are?
Is it possible that metrics are a bad way to determine success relative to actually looking into and understanding the situation? The latter is certainly more work. The former is certainly a lower resolution understanding.
Are we certain that any metrics communicate anything meaningful about "successful" performance in the realm of prosecution?
There are obviously better ways to judge a public servant's performance than a handful of decontextualized numbers — but unless you're going to force people to take a test before they can vote (which, umm…), what criteria are best doesn't matter as much as what criteria are most easily communicated.
No one ever got fired for buying IBM, nor for prosecuting someone. They have been fired though, for buying a cheaper solution, or for exercising discretion. Look at what happened to Dukakis...
