>>Tedd said: > >What I should have said was number(s) generated by a computer. The main >point of random numbers is that they are supposed to be random. And, it's >not just one random number, but a series, or set, of random numbers. Any >period in the set causes errors, or bias, in the computation. Now, for most >of us the use RND for variables in games and the like works well. But, for >others of us, the use RND functions for generating solutions to things like >sparse arrays, hash tables, binary tree searches, and other such geek stuff >may not work as well. In those types of applications, the better the random >numbers generated, the better the programs' solution. > >Maybe the built-in microphone on the Mac might be a source of a random >number generator? When somebody has some time, it might be an interesting >pursuit. You are right, my suggestion about time since startup is only for a seed. After that you have to accept the numbers generated by the "Pseudo Random Number Generator". Then if your PRNG meets all the tests of randomness required by your application, the results SHOULD be indistinguishable from those generated using some "True Random Number Generator", perhaps your Gieger counter or whatever. The trouble with a TRNG that is based on some physical observation typically is quite expensive and takes a long time. With a nice new G4 (drool, drool) I could generate a lot of PRGN samples in a very short time. A TRNG using radiation would need the amount of flux generated by those guys in Japan a few days ago to come anywhere close to that speed. I don't want that stuff anywhere close to me! Other physical phenomena, perhaps monitored by a microphone may be influenced by external events (periodic or not) such as a screaming baby, yap dog vs. Irish Wolfhound barking, or the heater/airconditioner mounted on the wall about 3 feet from my computer. Also these are modified by external events (I could chase the dog away) so they are not random. It's a nasty problem, I just go with a good PRNG and live with it. Bryan Bremner bremner@...