MIT Poker Attack?

Factual Questions
1

I was watching a fascinating program last night about the MIT boys and girls who fleeced Vegas blackjack tables during the eighties and nineties. Essentially they took card counting to another level, and were the first to introduce teams. The teams would appear to the outsider not to know each other. In reality they comprised card counters who would play all night at the table minimum and rich kid types who would turn up and stake the table maximum, apparently on a whim, but in reality only when they had been signalled that the table odds were favourable. This technique allowed them to avoid being detected as card counters by the pit bosses. They managed to win millions before the casinos wised up. Very clever kids.

Anyway, I have heard a bit about the twin dangers faced by online poker rooms: bots and team players. Which got me wondering how real those threats were. Anyone have the Straight Dope? Are a new generation of MIT kids developing bot teams to fleece online casinos, or at least their customers? Is there any legislation that, in theory at least, would stop them trying; or is it a free for all if you can avoid detection?

2

There’s a thread here with some more info in it.

Quick answer: It’s difficult to write a good stand-alone poker bot, and even if someone did it’s not much more of a threat to the other players that a decent human player. Sure it can play all night but you probably don’t so that’s not a big deal.

Teams and bot teams on the other hand would violently skew a game away from you. Even two players knowing each others cards is a huge advantage, more if they’re betting from the same funds. If they’re strategically placed at a table then even better.

Whatever rules you agree to in the casino (real or online) will include ‘Not Cheating’ so you’ll get turfed out if caught. You may be in line for fraud charges.

As for finding people out, team betting would produce a particular pattern of two players doing things that are ‘odd’ for their cards amongst other things. Of course they could just be playing strangely so you need a pattern of interaction between several players to prove that and in the time taken to establish that they may have made money then cut and run.

So the scammers need various accounts and to keep getting more accounts, the online casinos make getting duplicate accounts difficult and keep track of how people are betting to look for team play. The scammers vary the play to keep the stats less obvious, the casinos improve their tracking. It becomes a game of cat and mouse.

Real world casinos have it a bit easier as they just need to look for communication between the team. They’ll also have scanners and jammers for electronic communication. They may also use stats.

In addition I’d suspect that online casinos look at IP Addresses and network routes for all players at a table to try to tell if they’re physically close. There are ways round that too.

SD

3

I don’t think detection regarding shared card information is that big a deal. If, say, two people sit at a table together and share their card information (trivially easy to do online), they do have an advantage, but that itself is not a threat. Where it becomes a problem is when they share that information and when one has a good hand, they work together to jam the pot, raising each other constantly and getting those who are still in the hand to pay along (they can fold, but they don’t necessarily expect the pots to be raised like that and keep paying). Then, after the river card is flipped and they both reraise each other, one of them ‘suddenly folds’. THIS kind of cheating is generally tracked by the more reputable sites (Party, Stars, etc), and if a player reports this kind of behavior, the management will look into it and can ban the cheaters, confiscating their accounts (whether or not those who reported the cheating get reimbursed is tricky; sometimes yes, sometimes no).

Regarding bots, people are supposedly working on them and I am of the belief that if a bot could be created such that in not only made the right plays based on position, the cards they have (of course), AND the statistics/behaviors of the other players at the table, there would be a potential for a problem. What little I’ve heard of bots at this point, however, is that they are being designed with just the first two factors in mind. I think that unless the bot is designed to play as a solid rock, it will end up as a losing player as one of the important skills in poker is knowing who at your table is loose, who is weak, who is aggressive, etc. and adjusting your play accordingly. Right now, we human players can do this by taking physical or mental notes (for those players you are playing with at the moment) or via tracking software (Poker Tracker keeps a history and stats of every player I have ever played since I started using it). Based on this, I know who I can try to raise against to induce a fold versus who I need to have a solid hand with. The bots that are in design now (I do not really think there are any bots in use at this point in time, but if they are, they are losing money), from what little I’ve read, are not taking this into account.

If that hurdle is overcome, bots may become a serious issue, but the online casinos have way too much financial incentive to not allow them to succeed. This can be done via having management pose random questions to players after they have been on a certain amount of time, and if they don’t respond, kick them off (if it happens repeatedly, perhaps banning their account and confiscating their moneys).

I am by no means an expert on this so take it with a grain of salt; I’m just a guy who’s been playing/reading/learning for the past year. Perhaps Senorbeef will have something more to add.

4

I don’t see bots as cheating though casinos do. They’ll seize your account if they catch you using a bot. I think bots would make money because of tight aggressive play, not because of superior computing power. Collusion is another thing entirely. That’s outright cheating and illegal. You don’t need to be an MIT type to create either a bot or collude.

5

There is lots of work making good computer poker players, it’s an interesting problem from the Artificial Intelligence side of it. One of the universities has a poker bot play-off thing going.

Also there are poker bots out there playing now. I know of one person who wrote and ran a poker bot. It did some reacting to other players but mostly played the odds, if I recall the playing engine was based on some code from one of the university projects I mentioned. It won on smaller stake tables and tended to get stomped by higher stake tables. This is presumably because they had better players who worked out it’s reasonably predictable stratagy and took advantage. No idea if he’s still working on it or using it.

SD