XKCD

[kingnat]

Notice where poker sits. I <3 this comic, but this doesn't seem correct.

[JKDS]

Well, computers can beat some humans considering all the fish that complain about the bots.

In terms of actual competition, the University of Alberta has a bot that has been able to win against stoxpoker pros for some games including HU limit texas holdem. Theyve had moderate success against players in other games, but fall short.

Maybe it should be further down, just because they dont specify which poker. Certainly if it was ALL poker...computers still lose to most humans just because the UofA's bot Polaris i dont think can play things like Omaha or stud yet. If it was just limit...its probably fine where it is.

[d0zer]

yeah it's tough to clump all forms of poker together. FR limit is closer to scrabble AFAIK, where HU omaha would be closer to GO (or maybe calvinball).

[oskar]

You can teach your grandmother to beat four computer enemies cranked up to hardest difficulty at starcraft, so that should be way up the list from poker. I'm not sure if you think poker should be higher or lower. I think since it's a game that can be easily broken down mathematically it will be solved eventually, it's just really tedious. I would imagine that top computers can compete with top players heads-up. FR probably not so much because of a higher theoretical complexity.
I would not be surprised to learn that bots already compete at midstakes. People smart enough to program these are likely smart enough to hide from detection.

[bode]

there is already a great deal of proof that mid stakes games up to 10/20 have been beaten by bots over the last year or two to the tune of millions of dollars. Look at the Iipoker bot threads on 2+2.

[jyms]

there is already a great deal of proof that mid stakes games up to 10/20 have been beaten by bots over the last year or two to the tune of millions of dollars. Look at the Iipoker bot threads on 2+2.

This but once the bots are getting figured out they are getting beat by some regs too. That said, when a bot is good enough to be a winner at 10/20 on iPoker and Ongame and gets figured out, it's still better than 3/6 and 5/10 regs. Poker has a lot of lines to draw.


For those that don't know, there are different types of bots. These are not the same bots that were playing 01c/02c on UB.

[eugmac]

Random musings:

Seeing that bots seem to be taking over, maybe it's simply the future of poker - in order to win you have to program the best bot which involves having the most optimal strategy, adjusting ranges to different players based on HEM databases, basically do all the stuff that we do anyway as poker players but automated and tilt-free. The only thing the human player does is continue to gather more data, program in read-based optimizations vs opponent bots (this bot 3bets in this position X%, but WHICH X%?) etc, figure out which enemy bots have made adjustments to your bot, which then you have to write more code to counter-adjust vs those specific bots...

In a theoretical situation where the above became the norm, could one still consider it to be poker?

[Imthenewfish]

i might be thinking of a different botting scandal, but the last time i heard about bots winning at midstakes they were colluding

[jyms]

Possible collusion/softplaying/botting on 1knl on iPoker. - Medium Stakes Pot Limit and No Limit - Medium Stakes Poker Forum

It becomes more clear in the later pages

[surviva316]

i don't want to state the obvious, but having a really good winrate at 1knl != beating top pros, nor does it even mean that they're anywhere near the possibility of beating top pros. There is a ton of money to be won off of fish, and getting software to not spew mad money to regs doesn't exactly seem like science fiction.

Having an edge over the best players in the world is a COMPLETELY different issue. I don't mean that it is to a completely different proportion, so the programming has a long way to go down the same road they're already on; I'm saying beating top players is not in any way at all the same as beating the royal hell out of fish at a higher rate than they spew to regs.

[bode]

i don't want to state the obvious, but having a really good winrate at 1knl != beating top pros, nor does it even mean that they're anywhere near the possibility of beating top pros. There is a ton of money to be won off of fish, and getting software to not spew mad money to regs doesn't exactly seem like science fiction.

Having an edge over the best players in the world is a COMPLETELY different issue. I don't mean that it is to a completely different proportion, so the programming has a long way to go down the same road they're already on; I'm saying beating top players is not in any way at all the same as beating the royal hell out of fish at a higher rate than they spew to regs.

i could give a fuck if a bot can beat durrrr or phil G or anyone even close to their skill. What matters to me, and i assume anyone who has an interest in internet poker sticking around for years to come, is the fact that last year alone multiple millions of dollars were taken out of the poker economy by bots.

[surviva316]

i could give a fuck if a bot can beat durrrr or phil G or anyone even close to their skill. What matters to me, and i assume anyone who has an interest in internet poker sticking around for years to come, is the fact that last year alone multiple millions of dollars were taken out of the poker economy by bots.

Well yeah, I mean +1,000,000, but that's a completely different discussion from what's going on in OP.

[surviva316]

As for the topic specifically at hand, I have no opinion on what games it should be moved in front of and which it should be moved behind, but I do think that it's probably graphically shown too close to the line for "Computers can beat top humans." Unless a study of the iPoker bots came out to show that they were winning players against regs, I'd have to guess it's (at worst) closer to the comfortable middle of "further R&D needed."

As it's drawn it's like they're saying that they've got a lot of tweaks and bugs to fix, but it's not too far away. Beating fish does not make me feel it's all that imminent.

[JKDS]

Bet on the bot | The Economist

^^Polaris information against pros like Phil Laak in HU limit matches in case anyone was interested

[surviva316]

Bet on the bot | The Economist

^^Polaris information against pros like Phil Laak in HU limit matches in case anyone was interested

Playing break-even-ish HU-limit poker against Phil Laak puts zero fear in me.

EDIT: This was more meant to be a joke about Phil Laak and limit poker than it was meant to be me shitting on your article that I found interesting.

[oskar]

and regarding jeopardy's place on the list:
IBM&#39;s Watson supercomputer destroys all humans in Jeopardy - YouTube

[Imthenewfish]

and regarding jeopardy's place on the list:
IBM's Watson supercomputer destroys all humans in Jeopardy - YouTube

ehh in the video it looks like the computer's advantage is being capable of buzzing in the fastest, but thats about it

[oskar]

ehh in the video it looks like the computer's advantage is being capable of buzzing in the fastest, but thats about it

That's part of the game, isn't it
You can get the whole thing on Youtube, it's pretty damn impressive. I think it buzzes faster the more sure it is that it got it right. Even with a delay I think it would still hold its own. Can't wait until there's a publicly available version *_*

[andy-akb]

ehh in the video it looks like the computer's advantage is being capable of buzzing in the fastest, but thats about it

this completely misses the point of the accomplishment of watson. yea, of course a machine will buzz in faster than a human opponent. the point is that a fucking machine actually knows the answer often enough that it can buzz in before everybody else and get the question right.

[BennyLaRue]

There are top humans at Snakes and Ladders?

How is there any skill to that game at all?

[eugmac]

There are top humans at Snakes and Ladders?

How is there any skill to that game at all?

There is no skill, so everybody is a "top" human, because everybody is as good as each other at that game. Which means a computer is also just as good. At least I think that's what the author is trying to say.

[NightGizmo]

This describes how Watson buzzes in: IBM Research: How Watson “sees,” “hears,” and “speaks” to play Jeopardy!

Watson’s buzzing is not instantaneous. For some clues he may not complete the question answering computation in time to make the decision to buzz in. For all clues, even if he does have an answer and confidence ready in time, he still has to respond to the signal and physically depress the button.