Thanks for that explanation! That's pretty interesting! I'm curious about what the input and output variables are. Was the impact of self-upvoting considered, but then rejected for the current system implementation? Or was that a surprise? Or was it seen as a possibility with low enough probability that it wasn't worth bothering with, until evidence accumulated?
So interesting! Now I really want to see a comparison of the projected and actual impacts, knowing there's game theory model behind your design decisions.
The white paper has a bizarre ref to Metcalfe's law to justify the n^2 reward curve. Metcalfe's calculation has been shown to be inappropriate as a measure of "network worth". The maths is simple: a network with N users has a max N(N+1)/2 connections, and as N increases can be approx to N^2. But this is to calc computer power needed, not social worth. More recent calcs show this to be closer to N.logN so approx proportional to N is the right order of mag.
Basically, not everyone connects to everybody else in a network, but rather superusers emerge and most users connect to one or more of these. Therefore as N increases, the size of the network only increases proportional to N.logN and not N^2.
So the new whales are not the same species as the old whales, it seems. I've seen so many recent posts arguing that a whale-sized self-upvote on a comment is simply return on investment and to be expected. And your comment about complacence and custom on Reddit and Digg still fuels my interest in that game theory model!