Quote:
Originally Posted by africabrass
Yeah, an effective set of trees needs to have as many sizes as is computationally feasible. The full set in 6max already costs potentially 5 figures in server time to compute, so there are practical tradeoffs. I'm certain there are some sharp guys out there who can engineer a way to mimic more simplified strategies. Kruse seems to be more of a heavy-handed cheater with off-the-shelf type software. Not sure if anyone's looked into his multiway play, but usually that's a super good indicator if there's a large differential in skill compared to his heads-up pots, as building a full set of multiway spots for every texture would be an astronomically large computational task (if it's even possible? I haven't really looked into it).
Can't he just create a solution which incentivizes heads-up play even thought it might not be optimal he would cut those branches of of the game tree and use simplified solution like you said, which he is doing anyway since he is binning etc. And the remaining branches in multiway spots are such bad positions for other players that he is just printing.
The problem with costs of this caliber is that the person creating it is not the one profiting of of it from playing poker in my opinion, he is selling it to players maybe even leasing it for cut of the profits since hosting is expensive for such database, and the knowledge of creating/hosting/maintaining it is more profitable.