|
BGonline.org Forums
New bot-vs-bot cubeful-vs-cubeless experiment
Posted By: MK
Date: Tuesday, 20 August 2024, at 3:49 a.m.
No mutant strategies in this one. Seed is set to 130,000 + (1000, 2000, 3000, etc.) for each batch of 1,000 games to duplicate the exact dice rolls in both versions.
In the cubeless version, GnuBG 3-ply checker level plays 20,000 money games against GnuBG 1-ply checker level with bots playing normally.
In the cubeful version, GnuBG 3-ply checker and 3-ply cube level plays 20,000 money games against GnuBG 1-ply checker and 3-ply cube level with bots making checker and cube moves normally.
Obviously 1-ply checker will do badly in both versions. The questions are:
1- How badly 1-ply checker will do against 3-ply checker in cubeless session?
2- How badly 1-ply checker will do against 3-ply checker in cubelful session "in proportion to" in cubeless session? (with 3-ply cube skill added equally to both sides)
The question is to see if the results will be as expected; and if not, why not?
Please give your predictions in terms of winning percentages (or at least Elo's) that can be meaningfully useful in comparing results in different types of backgammon games, (i.e. no basketball, hopscotch, etc.) and feel free to explain the reasons for your predictions.
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.