Re: Theoretical Questions regarding Poker Bots
 

Here's my theoretical question about bots... Everyone talks about how easy they would be to detect... but how different would a bot play than a 10-tabling low limit 2+2er???

Seems like the easiest way for sites to reduce the profitability of bot development is to do away with multi-tabling. Most fish do not multi-table. Only robots and profitable semi-robotic semi-pros... Why make it easier for them? (From the site's perspective, anyway) What percentage of PartyPoker's handle is contributed by tables 2 - N played by multi-table players?

Re: Theoretical Questions regarding Poker Bots
 

[ QUOTE ]
The problem with programming a robot is that even though you could make it beat the games , software updates ( like color change -image or different structures)makes your programmed code useless. You will have to do it from the start when an update occurs and the sites are aware of this and change apperance from time to time.

[/ QUOTE ]

There was a guy recently talking about his software for using a gamepad to play poker on.

It uses 'hotspots' on windows to click the buttons. All you would have to do is adapt something similar to a bot and away you go. Provided the buttons dont move, no problems. And if they do you can tweak it quite easily i imagine so it now works.

The playing rules arent too hard. Im no expert but i think i could write the rules to make a profitable bot at party SnGs (there is software in the classifieds that can be a basis for calculating push ranges). It wouldnt be perfect, but it could work with a small ROI at the 10s or 20s. I know there are others a lot better than me in the 1TT forum who would come up with better rules.

The input is found through the hand histories found on the harddrive so that is easy enough - no image scaping is required.

All it would take is a good enough programmer to put it all together. And i dont think finding one capable would be hard - scour a uni or two, you will find at least 3 capable i imagine. Slip the 3 of them a grand or two, and as a team they could get around any problem you may have.

As an addition, you can plug in a piece of software that automatically joins tables (this is the hard part of the running, its so easy to do it that this doesnt exist yet for human players, despite the demand being there). If not possible, just come back every 45 min and start another 20 games over 5 skins.

The only difficulty is camourflaging it enough that it wont be spotted - this is the major difficulty. This is the one saving grace for human players right now. If you could hide it well enough, the above is easy enough to implement.

The research done publicly is easy enough to find. Its the above model, or the saabpo mk2 that we need to worry about. If sites arent up to scratch with security - we all know the potential consequences.

Phill