I know that Kasparov played all those highly publicized matches against computers but what would be really interesting to see how young Karpov or Capablanca would do against AI. Kasparov game is based on deep calculations and he is obviously losing to massive processing power of computers but Karpov was a brilliant strategist and it would be interesting to see how computer would play against best of human strategy.
This is a quote from Karpov

Also i bet Karpov would rock Kasparov HU in any poker game.

am I reading this wrong? why would he want to make his endgame have little chance of victory

No. A GTO player is a perfect player in the sense that you can't have any edge vs. him. At 5bb deep NLHE HU poker a GTO definitely exists. The same is not necessarily true for 300bb.

Overbetting is not necessarily more involved in psychology. A lot of the time the best play is to overbet and have the ratio of bluffs to valuebets GTO. For example if your range is polarized to bluff or nuts or villains range is pretty much face up as a bluffcatcher only/mostly (and this is the instance even when both players are playing GTO some of the time) an overbet is the best play for the player who is polarized to 0 and 1 since he can bluff more often that way.
A bot can also own a human in pshycology / metagame, at least in the future.

----

But is it possible that there does not exist a GTO strategy for Xbb deep NLHE HU / PLO HU ?

because his strategy has already guaranteed at least a draw

see 'freerolling in a split pot'

yes it is.

Plz make this public. And I`m sure you will get in a year action.

Is this offer valid for a heads-up limit hold 'em match?

even rock paper scissors has a gto solution. you can't outplay perfectly balanced random clicking.

A huge part of poker is emotional control... brain power isn't sufficient.

Boom!

Having played both chess and poker competitively at fairly high levels I have to say that Howard Lederer is full of **** on this one. Chess is a more complex game and requires more intelligence than poker. Just think about it...In poker any 2 bit amateur could beat Phil Ivey heads up over a small sample. Throw some amateur chess player against Kasparov and see how they fare.

all that proves is chess has less variance than poker?

If some amateur can beat phil ivey over a large sample size then you'd have a point.

Chess is a "solvable" game, yet being far from being solved nowadays afaik. Also, it's a "complete information" game.
Poker OTOH is not a "solvable" game. U can say a ****ton about odds and EV, but that alone doesn't solve poker. Obv, poker is an "incomplete information" game.

Not sure it's really wise to go on a huge discussion trying to compare both games to one another.

There should really be a FAQ somewhere about game theory as applied to poker because the same fundamental errors crop up again and again on 2+2.

There *is* a Nash equilibrium for any form of HU poker. The complete versus incomplete information distinction does not have any bearing on this. It just makes the game tree vastly bushier than it would have been. Same deal with variable bet sizing.

Most game theory whizzes speculate that there can't be a Nash equilibrium for multiway-poker because of the possibility of non-cheating collusive behaviors.

(The simple illustration is a 3-handed LHE game where SB just 3-bets 100% when his friend opens OTB, but both play optimally in other positions. This will cost the third player $ when he is in the BB unless he adjusts, and if he adjusts he is now exploitable and no longer GTO.)

kasparov is the man if people weren't yet aware

there are no bad beats in chess

read those words slowly.......think about them

apply to life

he knows nothing about poker. and a note : limit poker = chess.

I don't understand everyone's fixation with how bots wouldn't be able to handle overbets, like overbets are some magic key which asplodes the computers CPU.

I think the bots biggest hurdle is its ability to adjust to an opponent effectively and accurately within x amount of hands. I'd assume complex algorithms would be needed in order to adjust players ranges at notable points in a match, i.e has this players tendancies and ranges changed after a suck-out/getting stuck/doubling up etc. I suppose this is moot if it's possible for a bot to play an unexploitable 300bb strategy?

Garry is one of the smartest people in the world, but he is just straight up wrong when he says dismissively:

Obviously it is easy to calculate preflop equities and equities on the flop, even for a range versus another range. But when you start to include betting, where equities and ranges can change (eg. on a 238 board, 43 performs much worse than 46 with extremely heavy betting.), the whole problem becomes intractable quickly. That is why there are LHE bots that are really really good, and the comparable NLHE bots suck.

The US gov't?

Would reap the benefits of advances in AI. Would allow them to ruin an industry that they want to kill. Would allow Senators from Conservative or protectionist states to do a favor for FoF types, all the while wrapping it up in a pro-technology, pro-military game theory argument.

They'll cost the banking industry far more on the balance sheet trying to conform to regs, so they'd make the banking sector happy. If you think it out, developing AI to combat skill based online gaming would be cheaper than the way they are presently trying, and have benefits. It would be a smart choice, hence a very unlikely move from the gov't as efficiency isn't big on their agenda.

you believe the stuff you type?

mental midget

How about one who moves plastic cards?