Poker bots and HU Limit

[grjr]

Vexbot switched up on me a few sessions ago and started betting out every flop and 3-betting any raise when in the BB. It beat me that session but I've beat it every time since then once I figured out what was going on.

I wonder if online bots do any kind of opponent modeling like that.

[CrayZee]

[ QUOTE ]
Vexbot switched up on me a few sessions ago and started betting out every flop and 3-betting any raise when in the BB. It beat me that session but I've beat it every time since then once I figured out what was going on.

I wonder if online bots do any kind of opponent modeling like that.

[/ QUOTE ]

In my limited experience of simulations, bots are not very good at changing gears. There could be some fairly complex bots out there that do a better job at this, though.

The one thing that's somewhat amusing to me in retrospect is that I got interested in poker because a couple of my old comp. sci. buddies were interested in creating a bot in the summer of 2004. So I looked into the game and how to play.

These guys are really smart (but somewhat misguided, I think), and they have been developing this stuff for awhile getting virtually no where. I haven't talked to them in 3 months, but I doubt they're going anywhere w/ their FR LHE attempt esp. now that the landscape has changed to SH quite a bit. Most of the time they spent procrastinating and working on a generic poker room interface. A couple of months was spent waiting on me to take the lead, but I was busy reseaching the matter to see if it was worth my time. They eventually found a co-worker from Microshaft to take my role.

Instead I started learning poker and dumped the whole idea and decided not to join their "team". They wanted me to be the "project manager". I came to the conclusion that it was way too much work...not to mention that I now somewhat have a bias against real money bots now that I at least a little skilled. It's funny how life works.

The only thing that I've been hearing from this group is that it's always "almost ready to go live." Sounds like any other software project I've worked on that gets moved back. Eventually, they'll have to move on or get somewhere.

[TheScientist]

My (former) bot;



I'm a bad programmer and was an even worse poker player at the time. It took me an awfully long time to make it profitable though, and if I went back to it I'm sure I might be able to make it smart enough to play .5/1 or 1/2. Not worth my time though, I decided to actually learn the game and find playing easier than programming AI.

This bot had no opponent modeling and played an ABC super tight full ring style. It's scary to think what a minimalist opponent modeling style could do.

[CrayZee]

Yeah, you can write a minimalist bot and play micros, I guess, and be easily profitable. Originally, these guys wanted to do this for $3-$6 FR w/ some opponent modeling.

Even good programmers would have to go to great efforts. You have to manage all of the accounts, etc. How about start a normal business or pursue other opportunities as opposed to writing a somewhat complex bot? These are my thoughts now.

You also have to play the whole cat and mouse game once bot detection gets better and/or more serious, updating interface code for software changes, etc.

[TheScientist]

[ QUOTE ]
Yeah, you can write a minimalist bot and play micros, I guess, and be easily profitable. Originally, these guys wanted to do this for $3-$6 FR w/ some opponent modeling.

Even good programmers would have to go to great efforts. You have to manage all of the accounts, etc. How about start a normal business or pursue other opportunities as opposed to writing a somewhat complex bot? These are my thoughts now.

You also have to play the whole cat and mouse game once bot detection gets better and/or more serious, updating interface code for software changes, etc.

[/ QUOTE ]


Yeah, that was originally my idea too, play medium stakes games on many different sites, with many different computers, for many hours. A goldmine, if you think about it.

In practice, it's not so easy, but it can be done. (If the observations in this thread are correct) Luckily, the more easily exploitable full ring games are down in popularity and 6-max and NL are, in my opinion, a much more difficult environment to place a bot in.

I am wondering what methods the poker sites will have to use to protect people in the future though, it seems as if things will probably have to start getting fairly invasive (ex; taking periodic screen shots, disabling IM programs, complex analysis of play patterns, who knows.)

[CrayZee]

[ QUOTE ]
I am wondering what methods the poker sites will have to use to protect people in the future though, it seems as if things will probably have to start getting fairly invasive (ex; taking periodic screen shots, disabling IM programs, complex analysis of play patterns, who knows.)

[/ QUOTE ]

I know that I thought it would be fairly easy to do...esp. given our comp. sci. and programming backgrounds.

I remember the first things we discussed were complex detection prevention strategies. One of the most simple problems was the duration a bot would play. That's pretty easy to solve and randomize.

In the end, I think there will have to be fairly complex play pattern detection algorthms/applications written along w/ some sort of human verification system.

Studies on hand histories by Steven Levitt (of Freakonomics fame) hopefully will reveal something interesting.

[MicroBob]

another idea is to ask players certain semi-simple questions that they should be able to answer but a bot would be less likely to.

Unfortunately I think this could still be a problem for many of the non-english speaking players. If party (or rest of online-poker) ever penetrates the asian market that it would just make this 'conversation' method even more difficult.


But there is absolutely no reason they can't be more pro-active about this stuff.
If there is serious suspicion that this could be happening on the heads-up tables then why not just monitor those tables more closely.

Seriously, even 2 or 3 employees who kind of know what they're doing could put forth minimal effort just checking the tables (and perhaps chatting with suspicious individuals) could actually accomplish something.


Sitting back and doing nothing while the games are right in front of their faces is disappointing.
It shouldn't be too much to ask that they just show up at the table and check.

[blackize]

When I played on Prima this time last year you could call support staff to actually come chat at your table. Why not try doing this when playing with a suspected bot? If you don't want to tip of the bot operator sit down at another table, call over a member of support and when they ask what the problem is tell them to check out your heads up table.

[Glenn]

Like I said to the other guy, this obvious solution is very ineffective because all you need is someone loosely monitoring the games while the bot plays. This person doesnt even have to know how to play poker, so someone could have 10 friends run 8 bots for 5 hrs/day, etc.

[GrannyMae]

[ QUOTE ]
Seriously, even 2 or 3 employees who kind of know what they're doing could put forth minimal effort just checking the tables (and perhaps chatting with suspicious individuals) could actually accomplish something.

[/ QUOTE ]

i don't think the sites want to monitor then close a bunch of HU bots, because it would have to admit there is a bot problem. also, HU rake from bots playing each other 24/7 is probably what CEO's have sticky wet dreams about.

however, if the bots become proficient in multi-player games, then the hunt must begin.