| |
BGonline.org Forums
Does XG's definition of PR annoy you?
Posted By: Bob Koca In Response To: Does XG's definition of PR annoy you? (Timothy Chow)
Date: Tuesday, 9 July 2024, at 4:16 p.m.
What annoys me is the movement towards PRgammon instead of Backgammon in the past 20 years. If one wants to play a variance reduced match there is a perfectly fine way to do so (use variance reduced result) without the downsides being present (for those who really would rather play Backgammon) of the PR approach. In the following list I consider 3 and 4 to be the most important.
i) As being discussed in this thread there are arbitrary definitions of what counts as a decision.
ii) The bot isn't perfect. In some deep deep backgames with many checkers back it might not be good at all. One game like that can be played reasonably well but one's PR is destroyed from bot errors. That effect is compounded by those games tending to be long games. Also, for positions where the bot is off those who happen to play in the bot's style are unfairly rewarded
iii) There are many artificial ways to decrease one's PR which has nothing to do with actually being good at backgammon. Here I will mention just two. One example is suppose you have a cube in a close long race that you are sure is a drop but that it is very close. To improve PR would want to take to increase the decision count while having an easy game to play out. Another is making a slow down play in a bearoff. Suppose you have one checker each on the 1,2, and 3 points and roll a 21 vs opponent with 5 on the ace point. In backgammon the plays 3/off and 2/off 1/off are exactly equally good. In PRgammon though 3/off is better since it adds a decision if your next roll has an 1 other than 11.
iv) Backgammon is played against an actual opponent and not against a machine giving a rating. Against a particular opponent the best practical play may go against the best theoretical PR play. This aspect particularly punishes players who tend to play a complicated game. For backgammon it can help a stronger player win since it would probably cause both players to have a higher PR but likely that would happen more the weaker player. It does raise the PR of course though.
v) Clock usage and relative importance of a game is not correctly captured. Suppose the first game of a 31 point match that the cube gets to 8. The next game the cube stays on 1. Clearly, one would want to play more carefully when the cube is on 8 and would be willing to use a larger allotment of clock time. For PR gammon though errors in both games count equally.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.