But there’s also a third approach: instead of turning to full-bore randomness when you’re stuck, use a little bit of randomness every time you make a decision. This technique, developed by the same Los Alamos team that came up with the Monte Carlo Method, is called the Metropolis Algorithm.
