Date: Wed, 3 Oct 2001 21:01:18 -0400 From: Jamie Bowden Newsgroups: rec.arts.sf.written.robert-jordan Subject: Re: I calculate a great disturbance in the force... On 3 Oct 2001, John S. Novak, III wrote: > In article , > Brian Luethke wrote: > > >> Yes : the number of disintegrations per time unit should give you a > >> Poisson distribution, that is, "random" numbers. > > > I like the quotes on random. That is part of the point, they are not > > neccassarily random, just very close to random. Or at least appear random. > > Of course there are also people who argue that some very fancy things like > > that are random also, I tend to side with the people who have trouble with > > true random. > > I would think there is both a Nobel Physics Prize and a Turing Award > to anyone who can find a deterministic algorithm for nuclear decay. > > I don't think anyone has ever won both a Nobel and a Turing. It's hard to do random with systems designed specificly to be anything but. It is a systemic problem. Until we have systems that can actually deal with truely random events, you aren't going to be modelling true randomness. This cuts to the very core of why we don't hav AI that's worth a damn, and why discrimination and modelling rapidly break down when you start introducing real world data for comparison. We don't do random. When we encounter random, we cry in a corner and issue a segmentation fault, or a bus error, or silently corrupt data causing even more spectacular failures later on. When we can deal with random at the systemic level, we'll be able to generate random. Jamie Bowden -- "It was half way to Rivendell when the drugs began to take hold" Hunter S Tolkien "Fear and Loathing in Barad Dur" Iain Bowen