Can You Write a Sure-fire Algorithm to Stop Bots?

[Freakin]

[ QUOTE ]
Its naive to assume that script kiddies will always have bad bots. At some point code for a bot AI will be released that is better than the vast majority of humans. At this point online poker dies.

[/ QUOTE ]

and so does the person who stole the code from its creator =)

The better the bot, the fewer of them there are going to be, IMO.

[jjshabado]

Someone at a University will publicly release their code. Look at checkers, you could never play online checkers for money because the computer is unbeatable. People don't need to develop complicated AI to play, because all of the research and results are in the public domain after it was solved in Calgary.

Now Poker is a crapload harder then Checkers to write AI for, but the reality is the hard part (the learning/decision algorithms) will be public domain knowledge. It won't be hard to turn those algorithms into something that anybody can write a bot for.

Even if this wasn't true a small group of people with amazing bots would kill online poker. They're not going to be happy just playing one bot at one table. They'll flood the sites with them because as long as they're good enough that they can beat the rake they make more money by playing more bots.

[Nick Rivers]

[ QUOTE ]
The reality is in time AI will be advanced enough that bots will be able to beat most human players. Thats all it will take to kill online poker.

[/ QUOTE ]
If that were true, online backgammon wouldn't exist, because the AI at backgammon can defeat any human player. But, there are still plenty of sites offering real money backgammon games, and plenty of people playing them.

[indianaV8]

[ QUOTE ]
Someone at a University will publicly release their code. Look at checkers, you could never play online checkers for money because the computer is unbeatable. People don't need to develop complicated AI to play, because all of the research and results are in the public domain after it was solved in Calgary.

Now Poker is a crapload harder then Checkers to write AI for, but the reality is the hard part (the learning/decision algorithms) will be public domain knowledge. It won't be hard to turn those algorithms into something that anybody can write a bot for.

Even if this wasn't true a small group of people with amazing bots would kill online poker. They're not going to be happy just playing one bot at one table. They'll flood the sites with them because as long as they're good enough that they can beat the rake they make more money by playing more bots.

[/ QUOTE ]

I don't say computer play will never be a problem. If poker algorithms advance, if you have good equlibrium approximations and so on - the game will be endangered. This is not the case now, and I don't see it happening in the next years. Look at when the first pokerbot was done (by mike caro far back in 1980. Poker research is actively going on since 5-6 years but the topc of the icecream now from the scientific community is only the HU limit bot polaris that lost to humans). And if research advance to the point to endanger the game - if this happens - people will simply sell hardware calculators that play perfect poker ... , so this is completely independant from any bots and their detection.

So look at checkers, your example. If pokerstars or whomever - having the greatest anti bot detection techniques - offers checkers. Will people play it for real money? I will get simply a second PC and play it, if the game get popular I will manifacture hardware calculator for checkers and sell it to all fishes in mediamarkt (your compusa or whatever).

Solving the game is independant from any bot prevention and detection. If a game gets solved - it's dead. But in the meantime, while it's not the case - why bother to detect losing bots of people that do this as a hobby or as some fishy dream.

[jjshabado]

[ QUOTE ]
[ QUOTE ]
Someone at a University will publicly release their code. Look at checkers, you could never play online checkers for money because the computer is unbeatable. People don't need to develop complicated AI to play, because all of the research and results are in the public domain after it was solved in Calgary.

Now Poker is a crapload harder then Checkers to write AI for, but the reality is the hard part (the learning/decision algorithms) will be public domain knowledge. It won't be hard to turn those algorithms into something that anybody can write a bot for.

Even if this wasn't true a small group of people with amazing bots would kill online poker. They're not going to be happy just playing one bot at one table. They'll flood the sites with them because as long as they're good enough that they can beat the rake they make more money by playing more bots.

[/ QUOTE ]

I don't say computer play will never be a problem. If poker algorithms advance, if you have good equlibrium approximations and so on - the game will be endangered. This is not the case now, and I don't see it happening in the next years. Look at when the first pokerbot was done (by mike caro far back in 1980. Poker research is actively going on since 5-6 years but the topc of the icecream now from the scientific community is only the HU limit bot polaris that lost to humans). And if research advance to the point to endanger the game - if this happens - people will simply sell hardware calculators that play perfect poker ... , so this is completely independant from any bots and their detection.

So look at checkers, your example. If pokerstars or whomever - having the greatest anti bot detection techniques - offers checkers. Will people play it for real money? I will get simply a second PC and play it, if the game get popular I will manifacture hardware calculator for checkers and sell it to all fishes in mediamarkt (your compusa or whatever).

Solving the game is independant from any bot prevention and detection. If a game gets solved - it's dead. But in the meantime, while it's not the case - why bother to detect losing bots of people that do this as a hobby or as some fishy dream.

[/ QUOTE ]

It seems we don't really have any disagreements. I agree the game is not immediately threatened by bots. I'm talking about long term (5 - 15 years out) for online poker.

[jjshabado]

[ QUOTE ]
[ QUOTE ]
The reality is in time AI will be advanced enough that bots will be able to beat most human players. Thats all it will take to kill online poker.

[/ QUOTE ]
If that were true, online backgammon wouldn't exist, because the AI at backgammon can defeat any human player. But, there are still plenty of sites offering real money backgammon games, and plenty of people playing them.

[/ QUOTE ]

- I don't remember/know a lot about Backgammon AI, but I think I remember the search tree for it is really shallow because of the probability element (ie. each possible turn has many possibilities since the moves allowed are determined by rolling the dice). Is it possible that the backgammon algorithms can't be run fast enough on typical machines to play in real time against another player?

- Its possible Bot detection algorithms are good enough to find these bots. I doubt this will be true in 5 - 10 years. Similar to my argument above, the interface is just too narrow between hosting site and the player to make bot detection possible in the long term.

- There isn't the money in Backgammon that there is in Poker, and thats a big factor when writing these bots.

- Backgammon hasn't received the attention Poker has, so potentially thats a factor in the short term.

If none of those are true there will be some other short term reason I can't think of that they aren't overwhelmed with bots. I think we can safely assume that people aren't leaving these sites alone out of the goodness of their hearts, and when someone figures out the technical issues they'll jump at the chance to take advantage of the free money.

[Shoe Lace]

I don't think we have the technology yet to 100% fully prevent them without making the normal user's experience extremely annoying -- to the point of it being unplayable.

One idea might be to have poker sites send out audio based basic math problems randomly. Then, you would have to answer through a mic. Using voice recognition this is possible today, but it would be horribly inconvenient.

Imagine playing 12 tables and having to solve 12 answers while trying to figure out odds/etc..

I wouldn't like it. I'd rather my opponents can't solve basic math problems. [img]/images/graemlins/laugh.gif[/img]

[jjshabado]

Not sure if I understand your point. With current voice recognition software and voice emulation you could have software answer these questions correctly.

[StevieG]

[ QUOTE ]
Not sure if I understand your point. With current voice recognition software and voice emulation you could have software answer these questions correctly.

[/ QUOTE ]

Yes. In fact the smaller the set of expected inputs, the better voice recognition software is. In this case, spoke math problems would have a very small set of inputs.

[Shoe Lace]

[ QUOTE ]
Not sure if I understand your point. With current voice recognition software and voice emulation you could have software answer these questions correctly.

[/ QUOTE ]

Maybe when you create an account you have to activate it by phone. During this phone conversation you'll be asked to speak a few random sentences by the rep.

They'll log your voice pattern, and if the random math answers aren't the same pattern as your original activation pattern then your account gets disabled.

Seems like a bit much, but until we get some type of keyboard that can verify finger prints on the fly and send them over a network I really can't see how they can stop bots.

I'm sure when this technology comes there will be a way around it though.