I rewrote my @1010Game AI to use Monte Carlo Tree Search, and it's performing much better -- just passed 1M points!pic.twitter.com/Ni7WTnbrX8
Show this thread
-
It still plays strangely though -- pieces all over the board, not grouped together. I've seen it in positions where an unlucky piece could mean game over. A good player should be able to about those. I guess next step is to add an expert policy.
Show this thread
Finally lost at 2,364,370 points. I was worried that it would never lose, haha. It's pretty obvious that if pieces are truly random, you're guaranteed to lose eventually (e.g. 100 3x3 pieces in a row)... but how long, on average, should that take?
-
-
I guess this guy still in the game:https://twitter.com/TenPiecesOBread/status/931296197293117440 …
0:25@1010Game@RacerZach 6 months later and he's reached 3 million! Still wondering if he has the world record!
pic.twitter.com/BAF6EXe5zm -
Wow, that's incredible. No matter what algorithm you use, 1010 requires a degree of luck. I think my AI could get to 3M if it got lucky. It's also possible that the real game generates pieces differently than my impl (simple unbiased RNG)
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.