DNSBL cache modelling

In response to the question I sent out a few weeks ago, I've gotten
offers from several people to provide mail server traces of 
[IP, timestamp] that should be enough to at least start on some cache
models.

I am planning to hack stuff up myself, but it would be nice to have
another person or two to work with, both to compare notes, and because
my statistics are a bit rusty, what with my most recent course on the
topic having been in about 1974.  I'd like to make some estimates
like, how long a run do you need to have an N% confidence level that
you've found the average cache hit rate.

The MAAWG meeting later this month has invited a bunch of academic
researchers, and I'll be pitching this to them, too, but you all get
first dibs.

If you want to do some of the simulation programming, that would
be nice, too.

R's,
John