CmdrTaco's Journal: Research Into Unique User Counts On The Internet

So I'm curious if anyone out there has seen any real research (not just anecdotal crap) in trying to figure out accurate user counts... like, the average number of IPs used by an individual online over say, a week.

I know lots of strange stats for Slashdot: 321,000 IPs viewed the index yesterday. I also know that some IPs are shared by hundreds of users (like the time we accidentally banned microsoft's proxy server ;) and I know that some people hop IPs like crazy. But it seems like all of those numbers could be melted together to get a real true 'Average'. If the "Average" user uses 3.5 IPs/week, then we could take a rough stab at translating unique IPs into actual numbers of people viewed.

So what i'm looking for is scholarly research on that sort of thing. Anyone see anything? I know these numbers will vary tremendously from place to place... like I imagine Slashdot is doubly skewed: I bet we have a high density of users with static IPs... a static IP is sort of a thing of the past... something generally now more rare since most consumer level networks have gone to DHCP... but Slashdot readers would like to have DNS and services, so they'd be more likely to sit on the same IP over a long stretch of time. On the other hand, they also would be likely to have multiple devices, so I could see a single user checking Slashdot from their handheld, their laptop, desktop and office.

Anyway, I'm going to try to figure out some numbers here for us, but I'm curious what others have bumped into. You can email me if you can't post.