I would very much love to hear on HN about any results that you or anybody else come up with. My suspicion is that if this works out as you intend the distribution won't even require any fancy math to show people use it differently, it'll be blindingly obvious in a histogram. It's brilliant.