This is a lot like the "efficient cross-domain optimization" definition from Eliezer Yudkowsky. It looks like they even use the same information-theoretic measu

This is a lot like the "efficient cross-domain optimization" definition from Eliezer Yudkowsky. It looks like they even use the same information-theoretic measure of optimization power across the distribution of possible futures. http://lesswrong.com/lw/vb/efficient_crossdomain_optimization/ http://lesswrong.com/lw/vb/efficient_crossdomain_optimizatio...