[Slynky]

I think I am in favor of Geo's formula with a twist. Let me see if I can explain it easily.

His original formula (a few pages down) was (to summarize) to average the score of the people who beat the first person out (and so on) and apply points. This resulted, as some people disagreed with, a result that simulated a single game.

So, my twist would be to adjust this "straight" score to vary the spread so that losses and wins would spread out more. And this is what people have been trying to do with other suggestions.

So, the best I can figure and the easiest on me would be to figure it the way Geo said and then multiply the points gained (or lost) by the square root of the number of players in the game being rated. So, (duh!), if 4 players were in the game, the square root modifier would be 2. This is probably a common number of people for a multiplayer game. But, if 9 (to make it easy) were in a game, the multiplier would be 3. One has to see 16 people in a game before the multiplier became 4. This, in my opinion, would lesson the effect (both on winners and losers) in a big game...while giving a pretty fair representation for the most common game. In other words, the bigger the game, the less chance that a loser would suffer a huge loss from a bad position (and less the chance a winner would gain a huge amount of points based upon his lucky starting position).

I think I will go with this, which I think is a fair representation of the suggestions we have read and attribute any disparity in such a rating to a "true chess" rating formula to luck, and a lack of true comparison to any chess tourney/game.