The game of go has such a large board that it swamps traditional approaches to solve it via computer. To get around this, I have a simple hack: initially, the computer moves in exact symmetry to the player. Then, when the board has filled up and the search space has reduced in size, use a more traditional approach to go on the offense. This should work as moving symmetrically maintains a completely neutral board state. I've tested this by hand against computer go programs to verify it works, and I can tie them every time without even looking at the board. The tedium such an approach would create for a human opponent would also give the computer a possible advantage, assuming they don't quit in disgust first.
There is no room for what we wish to be true in science; that belongs in engineering.
Science is essentially reverse engineering nature; ultimately there is only one correct answer. In engineering, there are theoretically (if not practically) unlimited ways to approach the same problem. This is the biggest reason I am drawn to engineering over science, important as it may be.
Some believe that engineering is the intersection of art and science. I agree: it destroys one by turning it into the other.
Markets are a means, not an end.
The aim of the American Dream is the freedom to be fat, dumb, and happy. In practice, we've got two out of three.
Science finds the needle of truth in the haystack of delusion.
No comments:
Post a Comment