On Tue, 8 May 2001 14:10:38 +0900
Mathieu Bouchard <matju / sympatico.ca> wrote:

> I'm not really good at probabilities, but it seems to me that the latter
> makes a perfectly fair shuffler (given a perfectly fair rand()  function),
> and that your "j = rand len" cannot make a perfectly fair shuffler,
> because N**N is usually not divisible by factorial(N). It's nonetheless
> possible that yours is very close to being a fair one; can you show me
> that it is?

You might want to look at the discussion here:
http://c2.com/cgi/wiki?LinearShuffle