Checkers
A team at the University of Alberta has solved the game of checkers. The way they did it was ingenious; they solved all possible combinations of 10 pieces on the board, using 18 years of CPU time on up to 200 computers! The typical approach in Artificial Intelligence is to start at the beginning of a game and work forward, evaluating all possible moves as far into the future as possible, so people evaulate as much as they can in a certain amount of time and hope that that is enough to beat their opponent. The clever thing is to not waste time evaluating impossible moves, and there are all sorts of techniques known as pruning to do that wisely. For games like checkers or chess, there are simply too many possible configurations of the board to evaluate all of them using today’s computer technology. The team at Alberta took an opposite approach and worked backwards from possible end games, even if it meant evaluating possible combinations that could never be gotten to in a real checkers game. They then have a large database of “perfect” play from each 10 piece starting point, and from that the computer can be guaranteed to never lose a game of checkers. What is really cool about it is that the computer could play completely randomly up until the point when there are 10 pieces left. In fact, it could purposely try to lose as many pieces as possible until it got to the point where there were just 10 pieces left, and it would still win after that.
 This has all sorts of relevance for the Trust Manifesto. In situations where you have a way to guarantee success (or lack of failure) in the end game, then you don’t have to worry about the beginning and middle of the game. For instance, in libraries with RFID tags in the books and sensors at the doors, there is no need to hire guards to watch for people hiding library books in their backpacks. They will be caught at the end of the game (assuming they don’t rip the tags out of the books!). In typical purchasing transactions, I am not so worried about what the other party does, because I know that I can put stop payments on checks and create disputes with my credit card companies.
This notion of not worrying now because you have a fallback later is a very, very powerful idea, and I am convinced it is one of the most important principles for peace of mind in countless situations. The Chinook chess playing program is a very clever embodiment of that idea.
Â