I agree. I'd put it a little more like:

s => {
  next => 't'
  nearPrev => 'qr'
  nearNext => 't'
}
[download]

With the size of the nearPrev and nearNext dependant on the letters. M would be in the whole "lmnop" cluster.

For numbers, I would also have a "prev" which is undefined for letters.

Now, I've found all of this facinating (really!), but how can use something like this? We've talked about associations for numbers, letters, and words.

One interesting project would be a Perl program that plays word association, and learns from playing with others. Instead of just a "next", it would be most interesting to me if it had a "context" it tried to stay in (usually) also.

Well, I've got a weekend in front of me...

=Blue
...you might be eaten by a grue...