> Nothing particular about that, what I was asking was how Microchip generates
> *true* random numbers. Do they (you) use some kind of electronic phenomena
> that has been proven "random"? I believe there is no *true* random numbers
> but one can get very close ... well that depends on how you define *true* of
> course.

Because Microchip does not specify anything about how their "random"
numbers are generated, I would be strongly disinclined to make any
assumptions about randomness though I would expect that it's unlikely
that anyone would go through the trouble of exploiting any non-randomness
for anything except highly-sensitive cryptographic systems, and for those
I would strongly recommend doing the programming in-house anyway.  Other-
wise, my guess would be that Microchip is using something reasonably ran-
domish, but nothing that can be relied on as such.

PS--In the original specification for closed-captioning, it says that
    field 2 is supposed to contain "random" data.  Can anyone think of
    any reason why something would be spec'ed to be "random" with no
    indication of required spectral properties, non-correlation, etc.?