Figuring out what the hell the bot was fishing for is a puzzle for the ages... NONPOETIC is as good of an answer as I can find and it's surely not it, for the life of me I can't find what he was hoping to hit. This will keep me up at night.
Perhaps the bot is just going for a good leave, and hoping that you'll open a line. Depending on how it simulates, it may not assume that you are 100% guaranteed to purposely keep things closed. Not all of its opponents may play that way.
If one can win only if opponent makes a mistake, does it make more sense to assume perfect play, or imperfect play? Who knows how simulation works. If it's not finding optimal plays for itself, how could it find optimal plays for opponent in simulation? And if win probability is very low, then simulation results might end up being very close and somewhat random anyway.
@@jahonasnapkins What I've heard (now this is just from comments on these videos, so might not be correct) is that the bot essentially assumes its opponent will play like HastyBot (maximizing equity, not necessarily using strategy).
Maybe. Doesn't seem likely but admittedly I'm yet to come across an explanation that makes more sense. FWIW (and partly in response to the below), I don't believe BestBot assumes that it's opponent will play like HastyBot -- this is the biggest difference between BestBot and, say, Quackle sims. Quackle sims aren't recursive, which means they assume that any future turns in the sim will be the top equity play regardless of strategy. But BestBot sims (to the best of my knowledge) are recursive to at least some extent, which means they should assume I'm going to make a play that actually sims well instead of just has good equity (which, presumably, means I'm not going to blow the board open while up 150 lol).
I have no idea what the bot was doing, but ... At the end of the game, the bot had at least one out in 2: FISC (with FITS) leaving multiple outs with DN. Blocking H(OK)E was better for spread, though.
![ARGENTINA vs. BOLIVIA [6-0] | RESUMEN | ELIMINATORIAS SUDAMERICANAS | FECHA 10](http://i.ytimg.com/vi/C3wPtKNbAOA/mqdefault.jpg)
spoiler blocker comment !!!
god i need to know what the hell the bot was doing!
Don't we all!
Well done! I would like to think my calling out the bot for staking the deck in the last video had something to do with it ;o)
Figuring out what the hell the bot was fishing for is a puzzle for the ages... NONPOETIC is as good of an answer as I can find and it's surely not it, for the life of me I can't find what he was hoping to hit. This will keep me up at night.
Hahaha... has BestBot come to you in your dreams yet?
@@mackmeller Not quite but I've definitely been coming back to this video and just kind of staring at it a few minutes every day hahah. Mindboggling
Perhaps the bot is just going for a good leave, and hoping that you'll open a line. Depending on how it simulates, it may not assume that you are 100% guaranteed to purposely keep things closed. Not all of its opponents may play that way.
i thought it assumed perfect play? or no?
That leave was apparently not good enough
If one can win only if opponent makes a mistake, does it make more sense to assume perfect play, or imperfect play? Who knows how simulation works. If it's not finding optimal plays for itself, how could it find optimal plays for opponent in simulation? And if win probability is very low, then simulation results might end up being very close and somewhat random anyway.
@@jahonasnapkins What I've heard (now this is just from comments on these videos, so might not be correct) is that the bot essentially assumes its opponent will play like HastyBot (maximizing equity, not necessarily using strategy).
Maybe. Doesn't seem likely but admittedly I'm yet to come across an explanation that makes more sense. FWIW (and partly in response to the below), I don't believe BestBot assumes that it's opponent will play like HastyBot -- this is the biggest difference between BestBot and, say, Quackle sims. Quackle sims aren't recursive, which means they assume that any future turns in the sim will be the top equity play regardless of strategy. But BestBot sims (to the best of my knowledge) are recursive to at least some extent, which means they should assume I'm going to make a play that actually sims well instead of just has good equity (which, presumably, means I'm not going to blow the board open while up 150 lol).
PRIMATE is what the bot calls MacMel
Was hoping for ARMPITS on the first play lol
I have no idea what the bot was doing, but ...
At the end of the game, the bot had at least one out in 2: FISC (with FITS) leaving multiple outs with DN. Blocking H(OK)E was better for spread, though.
Ah, good call, yep
1000th times proving HastyBot is better than FraudBot apart from end game strategy
spill the hot pocket