I just listened to the latest podcast, and Scott, the argument you made on the backgammon doubling was completely wrong. I've already explained why, but I guess you failed to understand my point.
It was not a matter of some players being exploited by others. The crucial factor is that the tradeoff involved is asymmetric; under the rules as they appeared at the time, the advantage of winning by a massive amount was, in most circumstances, vastly greater than the disadvantage of losing by a massive amount.
In fact, it's trivial to show that two of the people who lost to massive doubles were 100% guaranteed to have no real downside to such doubling. If you look at Bronzdragon vs 5ro4 in the 6th round, both players were on 3 wins going into the match. This means they were guaranteed not to have the same number of wins at the end, and hence it wouldn't matter to them what score their opponent had if they lost. In such a situation, it's worth it to double even if you have only a 0.001% chance of winning the match. The same argument applies to Jaytee vs Yoshokatana - both were on 2 wins going into their 6th round match.
However, even in cases where there is a downside, it's still justified to double. Here's a fairly conservative example. Let's say that by doubling insanely, you potentially gain 4 places in the standings if you win, but potentially lose 1 place if you lose. So, assuming each place gained/lost has equal value (in fact places gained tend to be worth more), the consequence of this is that as long as your chance of winning is greater than 20%, it's worth it for you to double.
As such, doubling while losing is not merely the action of a stupid player, but in fact in a great many situations absolutely the correct and optimal play (assuming a points-based tiebreaker).
If the tiebreaker would have been score difference instead of score total, people would have been much more reluctant to play the insane-doubling game. The implicit threat of being tied against the person to whom you are possibly "giving" a very good tiebreaker is much, much less than the direct threat of getting a really shitty tiebreaker immediately.
As I said earlier, though, while this would help a lot, it doesn't eliminate the problem. For one thing, if both players have relatively low values for their total score delta, they still have more to gain than to lose.
Also, the nature of the Grand Prix scoring means that places gained are worth more than places lost, so you generally gain more by gaining X places in the standings than you lose by losing X places.
Comments
It was not a matter of some players being exploited by others. The crucial factor is that the tradeoff involved is asymmetric; under the rules as they appeared at the time, the advantage of winning by a massive amount was, in most circumstances, vastly greater than the disadvantage of losing by a massive amount.
In fact, it's trivial to show that two of the people who lost to massive doubles were 100% guaranteed to have no real downside to such doubling. If you look at Bronzdragon vs 5ro4 in the 6th round, both players were on 3 wins going into the match. This means they were guaranteed not to have the same number of wins at the end, and hence it wouldn't matter to them what score their opponent had if they lost. In such a situation, it's worth it to double even if you have only a 0.001% chance of winning the match. The same argument applies to Jaytee vs Yoshokatana - both were on 2 wins going into their 6th round match.
However, even in cases where there is a downside, it's still justified to double. Here's a fairly conservative example. Let's say that by doubling insanely, you potentially gain 4 places in the standings if you win, but potentially lose 1 place if you lose. So, assuming each place gained/lost has equal value (in fact places gained tend to be worth more), the consequence of this is that as long as your chance of winning is greater than 20%, it's worth it for you to double.
As such, doubling while losing is not merely the action of a stupid player, but in fact in a great many situations absolutely the correct and optimal play (assuming a points-based tiebreaker).
TL;DR Scott sucks at practical game theory.
As I said earlier, though, while this would help a lot, it doesn't eliminate the problem. For one thing, if both players have relatively low values for their total score delta, they still have more to gain than to lose.
Also, the nature of the Grand Prix scoring means that places gained are worth more than places lost, so you generally gain more by gaining X places in the standings than you lose by losing X places.