Quote:
Originally Posted by kruidenbuiltje
Thanks Mr. Robertie (or can i say Bill?),
this brings me to some other questions:
Is the same true for software like Jellyfish or Snowie, and are there other types of positions/situations where the decision of the software is bad.
For instance how are they on backgames? Are they good in timing issues involved in back games? And things like dump play? I presume they are good in holding games, races and bearoffs. I have ordered your book Modern Backgammon, does this book also discuss these issues?
greeting kruidenbuiltje
Bill is fine.
There are several available BG neural nets now, developed by different teams using different algorithms and assumptions. People have their own favorites. I like Snowie, but others on this forum prefer GNU or XG.
Snowie has a special neural sub-net optimized for backgames. It plays them well, and its 3-ply rollouts for backgames are very good.
Modern BG is a book about the underlying principles of checker play. It doesn't discuss bot issues at all.