Beau Brummels (9/23/04)

____________

A note on the direction of time [ignoring global considerations]: you think of the reason it goes one way and not the other [properly: not both ways at once; after that the distinction between past and future is just a matter of labels] is just the reason that, once Humpty Dumpty’s had a great fall, you can’t put him back together again. But isn’t this just a question of computational difficulty? It’s easy to break Humpty into pieces, but it’s a difficult combinatorial puzzle to figure out how to fit all the chunks of eggshell back into one smooth surface. In fact you could look at the decomposition as the verification of a proof that Humpty consists of that set of pieces, and the reconstruction as the process of finding that proof. The latter, obviously, should be harder. So this is the connection of the direction of time to P/NP.

____________

Golden protest (7/2/04))