The Problem With Turing Machines

The way I try and explain it in a much less formal way than Jan would in his paper is if we have three functions you could think of it as a function. One is the mind that function two is like the body and function three is the environment that's sort of determining what intelligent behavior looks like. We can find the optimal function one for the environment. But then if we change if we compose these functions you've got function one inside function two inside function three. If you change function two it doesn't matter how good your choice of function one is you can always change function two to sort of ruin the day of function one so that the performance looks terrible with respect to the environment you