If you are too successful at something then you are probably too local in your steps/bets - you should make greater steps/bets by risking more in order to incre

If you are too successful at something then you are probably too local in your steps/bets - you should make greater steps/bets by risking more in order to increase your overall progress. It is an optimal rule which was mathematically derived for convex functions, see the one-fifth rule at http://ls11-www.cs.tu-dortmund.de/people/beyer/EA-glossary/node97.html http://ls11-www.cs.tu-dortmund.de/people/beyer/EA-glossary/n.... The ratio itself is problem-dependent but the core idea is beautiful.