Determinism in computer vs. computer matches

Discussion of chess software programming and technical issues.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
johnwbyrd
Posts: 9
Joined: Tue Jan 20, 2015 2:22 am

Determinism in computer vs. computer matches

Post by johnwbyrd » Tue Jan 28, 2020 8:18 am

Dear all,

I've written a popular article on Quora on the subject of why computer vs. computer matches have unpredictable outcomes.

Please feel free to review the article; not so much because I think you will learn much from it, but rather because I encourage those more experienced than me, to find errors in my writing.

Thank you in advance for helping me improve the quality of the article. :)

https://qr.ae/TDNx7r

bob
Posts: 20748
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Determinism in computer vs. computer matches

Post by bob » Wed Jan 29, 2020 2:39 am

A pretty good description of a problem that has been pretty well-known (at least among programmers) for many years. The main difficulty it causes me is it makes debugging a very difficult task at times, since you see a bug of some sort during a game, but when you go back to test things run perfectly as expected. You have to repeat the debug run over and over until it fails, and hope you captured enough information to find the issue. If not, add more measurement code and rerun a bunch more times and hope you hit the bug again. And infrequently, the extra code you added will completely eliminate the bug by causing its own timing perturbations...

johnwbyrd
Posts: 9
Joined: Tue Jan 20, 2015 2:22 am

Re: Determinism in computer vs. computer matches

Post by johnwbyrd » Wed Jan 29, 2020 7:18 am

Thank you Bob, I sincerely appreciate your sanity check on the article. Big fan of your work and your help to the field, by the way.

Post Reply