[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Is Snowie still relevant?

Posted By: Casper Van der Tak
Date: Saturday, 11 March 2017, at 11:20 a.m.

In Response To: Is Snowie still relevant? (Jack Mack)

1. This is what Xavier has said about GNU (using your links): "In 2001 when the first version of XG was put on GammonSite, I got a jump start on how to program a Neural Network from Gary Wong's gnubg 0.02."... This does not mean that Xavier was using the code, just that being to able to look at the code was helpful and provided some ideas.

2. You think the fact that both GNU and XG fail to resign gin positions making the same mistake. Personally I would not program a feature to consistently figure out whether a position is gin or not. Waste of resources and slows the program down. If you do not program this, there is a look-ahead horizon effect at work. Both programs make the same sensible choice.

3. Meaningless plays is gin positions - yes, logically both programs may do that, because for both there is no equity difference between the moves (or in terms of the neural net, evaluation differences). It is a feature that arises because of the similarity between the programs, not because one stole from the other.

4. The 11 play XG and GNU make is correct, because it avoids blotting in X2 followed by X2 sequences, and because getting checkers off is valuable in the cases you get hit (especially against no containing board).

5. I thought Timothy's summary very helpful.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.