This is interesting. I've wondered about the possibility of a site being rigged under the threshold of statistical detection, not really from a practical perspective because I don't see how it would be possible to capitalize on such small margins, but knowing that there's a limit to the precision of RNGs makes me think that it's not entirely implausible.

I don't have a proposed mechanism, I'm not a mathematician, nor a computer scientist, but I did some coursework in the natural sciences and one of the most basic principles of lab work is that there's a limit to the accuracy that something can be measured based on the precision of the instruments employed, i.e. significant figures. So, it was this notion of significant figures that sparked this line of inquiry. I never bothered to follow it up, mostly because I wasn't sure if the precision of an RNG is at all analogous to an analytical instrument, and honestly because it's only marginally interesting from a theoretical perspective.

Don't get me wrong, I'm not in the 'online poker is rigged' camp at all. I've always argued that there's just too much hard data available for a site to pull it off in any meaningful way and that doing so would only jeopardize whatever limited credibility the poker site has managed to accrue over the years for marginal profitability at best.

I know this isn't at all what the good professor is talking about, but it reminded me of this idea and I wanted to see if there are any eggheads out there that would care to chime in.