| |
BGonline.org Forums
Last bot-vs-bot, cubeful-vs-cubeless experiment results.
Posted By: MK
Date: Saturday, 24 August 2024, at 5:20 a.m.
EXPERIMENT 29: see https://montanaonline.net/backgammon/py3.php
These experiments, reported together, are different ply versions of Experiments 27 and 28.
In the first one, GnuBG 3-ply checker + 2-ply cube, (strong bot), played 20,000 cubeful money games against GnuBG 0-ply checker + 2-ply cube skill (weak bot).
Strong bot won 24,141 (52.03%) weak bot won 22,256 (47.97%) of points, with a total of 46,397 points and an average of 2.3199 points per game.
In the second one, GnuBG 3-ply checker, (strong bot), played 20,000 cubeless money games against GnuBG 0-ply checker (weak bot).
Strong bot won 13,536 (52.76%) weak bot won 12,120 (47.24%) of points, with a total of 25,657 points and an average of 1.2828 points per game.
These results are incredibly close to the ones in Experiments 27 and 28. This time around, weak bot won 0.73% more points in cubeful games, even though the checker skill difference between the sides was even greater.
These will most likely be my last experiments in the cubeful-vs-cubeless series done to demonstrate that, contrary to the general "belief" in the gamblegammon world, the doubling cube does not add any more skill to the game but instead "magnifies luck", just as I argued since 1998 but previously could only try to demonstrate through my subjective human-vs-bot experiments. Now you no longer have to take my word for it. You can run your own sessions to prove it to yourselves. I will write a summarizing essay about this subject when I can make time for it.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.