Hi, I just recently bought "501 Essential Backgammon Problems." I am just looking through the first set of problems (opening moves and reply moves). I have noticed that some of the moves recommended by Robertie are slightly contrary to the moves recommended by GNU BG.

I have my evaluations set to "Supremo". My question is: Is GNU wrong or are there small errors in the book? The copyright date is 2000, 2004, so software analysis was available at the time of writing. I am guessing that GNU is miscalculating the equities of the resulting positions...?

I haven't done a full rollout on these, which I guess would be due diligence. But, if someone knows right off hand I can save some computer time

As many of us use this book regulairly, would you mind posting 1-2 examples, and maybe we can talk through them?

I too have had trouble somtimes getting GNU to agree with various suggested problems from different books etc; I find that GNU will sometimes like a highly counter-intuitive move (at least, counter-intuitive to me) that is marginally better/worse than the recommended move...

The advice in the book is based on bot rollouts from the late 90s. (Mostly Jellyfish, if I recall, but it's possible I also did rollouts using Snowie 1.0, which came out in the summer of 1998.)

The bots are much stronger now but the only opening move that I think is affected is a 2-1. The early bots all played 13/11, 24/23 and their rollouts also indicated that was best. After Snowie learned how to play backgames much better (release 4.0 and later) both its evaluation and rollouts switched to indicating that 13/11 6/5 was preferable. However, many players (myself included) back in the 90s believed that the slots were better than the bots said because of their obvious backgame weaknesses, so I was recommending the slot at the time.

Note that a lot of the opening rolls (43, 32, 52, 51, 64, 63) are pretty much toss-ups anyway.