Neverlose

[Turning Stone Pro]

"I would imagine that unleashing 100 bots on the party 15 would be far more profitable."

Me and Steve G will take care of those bastards.

TSP

[glen]

Since people seem to like poker on tv, maybe they should start televising a "battle of the bots," kind of like a poker equivalent to the show where nerds design mechcanical constructs that try to render their opponent immobile. . . .

[Evan]

Do people think Neverwin is a bot too? I heard they are related or something, although I have no idea where I heard that.

[AviD]

I haven't given poker bots much thought, despite being a programmer myself...but many of these "signs" indicating potential bot usage are not that difficult to overcome...but there are more complex issues to overcome that make an effective bot nearly impossible (or at least will take a sick amount of time) to develop IMO.

The bigger issue is spending the time to actually develop a bot to address them, which most programmers probably are not willing to do.

Furthermore, and per TSP's analogy to Deep Blue, you would need some insane hardware to run an application that could effectively analyze player statistics, in conjunction with hand values, relative board value, betting strategies, etc, etc, etc...and apply them to decision making logic ala AI. Quite complex, and it has to act FAST...i.e. can't sit and think for 30 seconds and certainly not some "set time" at a clip to make relatively easy decisions.

So in order to keep under the radar or even keep from timing out during the decision making process, you'd have to either find an alternate methodology for the decision making process, say some kind of average hand playing style matrix of a given game based on its average respective texture. Granted, this will be flawed and even a minor flaw could prove costly...but at the same time could overcome the -EV offset with +EV playing style over greater lengths of time on increased numbers of tables and therefore increased numbers of hands.

But to do all that, when I'd imagine a straight ABC poker bot might "do the trick", seems like a serious amount of time most programmers are just not willing to invest in. Not too mention (as with Neverlose) it will be entirely suspect and you will need to implement an offset losing strategy, while still maintaining winning play and positive profits.

Ultimately a truly effective poker bot cannot be exclusively a game crusher. It has to have some level of AI to adapt to both playing conditions and address the psychology of players whom become suspect. This increased complexity is not something I can see anyone investing in, given the recorded nature of playing time and results, and the overall difficulty of its implementation.

Not too mention, whomever is running this bot would not be able to have a full time job, or essentially do anything that may provide evidence that he is running a bot and not actually playing himself. The lifestyle commitment to actually do this effectively is pretty intense IMO, although being that most gambling sites (if not all) are offshore, perhaps it is a non-issue.

Alot of "holes" to developing high stakes poker bots IMO, too many that will often be short sighted. An inferior bot is the likely result, one that is too simple and loses more than it wins, or one that is too good and wins more than is credible. It sounds like some of you have already seen both of these types, developing one in the middle ground is far, far more difficult and eventually like most games, some form of authentication security will be added if the problem persists.

I agree it is equally disturbing that PokerStars "detected" this issue after it was reported by live players, yet really did little about it by the sounds of it. That's obviously not a step in the right direction in terms of handling potential botting, especially when real money is on the line.

[B Dids]

'Stars has investigated Neverlose, he's not a bot.

[turnipmonster]

avi,
what if someone had access to a good codebase for development? I don't think it would take half the amount of time if someone had the source to something like poki, or anything that implements opponent modelling.

--turnipmonster

[Evan]

[ QUOTE ]
'Stars has investigated Neverlose, he's not a bot.

[/ QUOTE ]
That must be the coolest thing ever, being so good that you get investigated for being a bot.

[Kevin J]

A bot playing a short-handed $100-$200 games on Stars, is an investment I'd want to stay away from. Would you want to put up 100k and hope it's programmed well enough to beat some of the best short-handed players in the world?

All you'd need is two people to suspect Neverlose is a bot and it would be an easy matter to wipe him out. And yes.. I think that if Neverlose can play as a bot without detection from Stars, then he could also be broke in this fashion without detection.

And forgetting all else, a bot should not be that hard to beat in a short-handed game anyway, so I don't see what the big deal is.

Lastly, I'm pretty sure I know who Neverlose is (used to play in my room), but of course, I'd never say. And even if I'm right, it doesn't mean it's not a bot doing the actual playing.

* I haven't read TSP's post about chess yet, but I know that it would be a MUCH more complicated matter to program a bot for short-handed poker than it would be for chess. It's not even close.

[Kevin J]

I believe Mason has already written about this and if I remember right, his conclusion was that poker is a much more complicated game than chess. At least from a programming standpoint.

[Kevin J]

Well I just read the chess sub thread and it appears I'm wrong. If the posters are right, Bots CAN play well short-handed. I don't understand that. Bots must resort to game theory and while they can compile lots of data about your play, I would think it would be a simple matter to start over-using and over-applying concepts to throw it off. But I know little about this and haven't really thought that hard about it. So I'm probably very wrong.