Except that this has been tried in police states where they can kill just about anyone they like, and it has never worked. Part of the problem is that humans rarely make judgments of risk that are remotely reasonable, rational, or correct, a fundamental characteristic of our cognitive architecture. A solution predicated on humans routinely making good judgments and evaluating risk even vaguely correctly is ipso facto useless.
That POV seems to cast considerable doubt on the ideas of democracy and self-government.