[Closed] man vs machine

@ skiff - Sorry, I did not mean to attack you. My "read" was indeed wrong. I apologize.

There's definitely nothing wrong with ps.com articles and strategies. I think, if someone would make a bot playing strictly according to this strategy, it would beat NL2, probably also NL5, but I doubt it would work on NL25. You can say exactly what should he do preflop and on the flop, but on turn and on river there is really a lot of different situations, that you just can't pick every one of them and tell bot what to do if this one occurs. They are just too many. You can just make few situations, and create an algorithm that would give you one of the few situations, to whick this particular one is nearest. And creating algorithm like this is imo not that easy. That's why I don't think it's possible to make a good bot this way. Even if he table selects according to your goals. And, e.g. how would you programm him to exploit the fish? This is propably the hardest thing.

HUHU LHE is definitely in trouble due to bots.

The more advanced ones out there which have been shown off can play almost perfect equilibrium (Polaris being one) so in the long run these bots can beat pretty much anyone.

6-max and FR are a little more complicated as there are a much larger degreee of possibilities to analyze. I have no doubt that you could certianly program a bot to play a 32/23 style with default settings vs player parameters (I think there was one on Party recently wasn't there?) which will crush fish and poor playing TAGs but to actually solve 6-max and FR may take a bit more time.

Again, as far as NLHE goes thats a more complex game still so, again, programming a bot to play a NIT-TAG push/fold shortstack game is not going to be that hard but to actually solve the game that's years away.

PLO...forget it, the bot would probably go mad and we'd have a "Matrix" scenario on our hands and it'd all be PLO's fault.

With the amount that you can bet varying a lot in nl, it just makes a bot building so much tougher. I just don't see how a bot could ever figure out a general rule to e.g. over betting on different textures. It's just going to be so tough to get a sample on even just a one player type with over betting that I don't see the nl bots as a huge threat. Obv. on any fl games that's a lot easier but with nl esp. it's just going to be ridiculously tough to build a bot without really glaring leaks, regardless of the method used.

Originally posted by supeyrio
@fusionpk: i think you forgot about adaptions.. a simpler game like rock scissors paper, sure a bot can estimate my frequencies etc e.g. i choose rock on higher frequency. the bot will then choose scissors on a higher frequency to win. however as human, i would have spotted that and adapt. while the bot still relies on the old data it has.

poker is a game of adaptation, whoever can adapt faster and better wins.
.

+1

I find it unlikely that a NLHE bot will be crushing the top players any time soon.

To take the analogy a little further, levelling is one area I believe a nlhe bot simply cannot compete with humans.

You could intentionally pick rock at a slightly higher frequency, and then adjust by picking scissors. When nlhe-bot adjusts to your over-frequent rocks by picking paper more frequently, you are already one level ahead.

Why? Because you understand nlhe bot's perception of your range and can outlevel the bot appropriately. To make a direct application to NLHE, this could apply when merging your value-bet range since it requires an understanding of how your opponents perceive YOU. I really think current programming does not allow bots to think on this level, the human mind is more capable than any supercomputer, even if it lacks the raw computational ability of a calculator.

Levelling is just one of many reasons why programming a decent nlhe bot is off the cards (so to speak ;)). You might be able to program one to beat a mediocre/poor player, but it will certainly not be the competitive equivalent of bots such as deep blue (chess) or polaris (lhe)

Originally posted by supeyrio
@fusionpk: i think you forgot about adaptions.. a simpler game like rock scissors paper, sure a bot can estimate my frequencies etc e.g. i choose rock on higher frequency. the bot will then choose scissors on a higher frequency to win.

Lol, that would be a pretty poo bot!

Phil Laak and Ali Eslami beat the computer opponent provided for them by the University of Alberta fair and square, the only worrying issue is though, that they too won by the narrowest of margins.

As we all know, there are few players in online poker that could ever hope to match either Laak or Eslami, so yes indeed, the technology to beat most human opponents, is a reality already.

Commercially available bots however, do not appear serious enough just yet.
YET!!

Originally posted by Hahaownedlolz

Originally posted by fusionpk
There are already NLHE bots built that would crush nearly anyone, im confident some of these bots could beat midstakes+ infact I heard of a bot that took on some HSNL players before and it was a close match iirc.

proof? I've never heard about any good no limit bot. In theory it's ofcourse possible to make a bot that could beat high stakes online.. But if your that good at poker and programming then your wasting your talent

THEN YOU MAY HAVE CHECK OUT THIS FOR INSTANCE

http://www.pokertableratings.com/blog/2010/07/bot-ring-discovered-on-poker-stars/

Originally posted by Johniiicek

THEN YOU MAY HAVE CHECK OUT THIS FOR INSTANCE

http://www.pokertableratings.com/blog/2010/07/bot-ring-discovered-on-poker-stars/

Although it is worth noting that this is a 20bb shortstack bot, not quite the same as writing a bot for 100bb+ where skill plays a much bigger factor.

Also, as soon as you realise this bot is in fact a bot, it is absolutely crushable.

Sure but this bot is one of the more basic bots, this is one of the bots that is actually writeable by a hobbyist and hence they are alot more common than the bots I was previously talking about.