On 4/27/2024 2:49 PM, Philippe Michel wrote:
On Fri, Apr 26, 2024 at 06:04:50PM -0600, Murat K wrote:
I posted an article including pretty images of some positions on: https://www.bgonline.org/forums/webbbs_config.pl?read=213189
There have been a few more posts about this in that forum also.
I have seen somewhat similar positions in the training database. Maybe not as extreme but ludicrous stacks in the outfield....
I really couldn't understand what you wrote in order to respond meaningfully and I hope that others may/will. I want to add a few more comments though. When I discovered this, one of the things that immediately crossed my mind was whether it may have happened other times during the 500-600K games that I ran during my various mutant cube experiments. If it did, that would surely invalidate all the results. Luckily, my scripts kept track of number of rolls in each game and I could easily check if there were any games that lasted hundreds of moves. In the about 260K games that I had shared on my web site, only 23 games had lasted more than 150 moves and only one more than 200 moves. (These numbers are a useful bit of info in themselves.) However, this is by no means a conclusive evidence that such weird games hadn't happened in the rest of the unrecorded games nor that they can never happen again in the future. Another more general concern I expressed in my post at bgonline.org https://www.bgonline.org/forums/webbbs_config.pl?read=213249 is that since my experiments are essentially rollouts using "mutant settings", what can happen during my experiments can happen during rollouts and not just in GnuBG but in XG and all other in-bred bots. Most of my experiments were run unattended, often overnight. It's by luck that I happened to look at the command window at the right time to catch this. Rollout are run "in the dark" without even any chance of catching a glipse of positions scrolling on the screen. Could this kind of bizarre games have been happening in rollouts for all those years without anyone noticing? And if so, what would the implication of that would be?? MK
