Cybersharps outplay poker websites

[DimPrawn]

Show us your bots!

[DaveB]

That really is the key point, you don't want to be pitting your bot against potentially good players in a high-stakes game.

Eventually, it will be a case of who has the better bot!

Eventually, yes, but not yet. Setting up bots takes a lot of time and effort. It's not a case of simply downloading the bot app and running it. It needs to be configured to play the way you want it to play, it needs a specific PC configuration to do it without getting caught and they need constant tinkering to keep up with changes on the poker sites. To run an effective poker bot operation that is going to make money is a full time programming job in itself.

[d000hg]

Winning at poker against good players requires a lot of complex game theory and other mathematics... not something easy and deterministic. But, as someone said, on the noob tables (the low stakes) you don't need anything as clever.

The point is, even if your bot averages only a pretty conservative £1/hour on 1p/2p table, if you run bots on 10 tables at once, 24 hours a day, then you're up to £240/day... in fact more since that's 365 days a year... it works out as more like £350/day if you average over 48 working weeks per year.

However I wouldn't personally want to use someone else's bot. I'd rather develop an algorithm, force myself to play using it, and test it works before coding it up.

[DimPrawn]

Surely a set of CAPTCHA tests throughout the game will separate the bots from the humans?

http://en.wikipedia.org/wiki/Captcha

i.e. at random points in a game present a CAPTCHA for the human players to easily pass, meanwhile the bots are generally fooked.

And alternative is MAPTCHA, where the human must answer a simple math test:

http://blogs.crammerz-inc.net/media/...random.irb.png

[sasguru]

Winning at poker against good players requires a lot of complex game theory and other mathematics...

basic probability theory and psychological nous.

HTH

[d000hg]

basic probability theory and psychological nous.

HTH

No that's what humans need. To model this in software is a complex task because poker has such a human element. That's where game theory comes in. It's a very hard game to 'solve' - your algorithm has to be able to cope with both very good and very bad players which in itself is not a simple thing.

[d000hg]

Surely a set of CAPTCHA tests throughout the game will separate the bots from the humans?

http://en.wikipedia.org/wiki/Captcha

i.e. at random points in a game present a CAPTCHA for the human players to easily pass, meanwhile the bots are generally fooked.

And alternative is MAPTCHA, where the human must answer a simple math test:

http://blogs.crammerz-inc.net/media/...random.irb.png

You can do it at login, but even that will piss people off. If you frequently get interrupted - especially by captchas which many people find hard to read - people will simply go to another site.
Besides, there's a whole industry in developing captcha-breaking software.

[Moscow Mule]

You can do it at login, but even that will piss people off. If you frequently get interrupted - especially by captchas which many people find hard to read - people will simply go to another site.
Besides, there's a whole industry in developing captcha-breaking software.

You need KittenCha, or whatever it was called. Makes you decide if a picture is a kitten or a puppy, apparently quite hard for a computer but pretty easy for a human.

And who doesn't like looking at pictures of kittens & puppies?

[d000hg]

Still problematic as this is exactly what neural networks can be trained to do. Any site with a "cha" of any sort needs to offer a "I can't tell what this is, show me another" button, which is great for obtaining training data for a NN.

It's not an easy thing to break these tools, but if money is involved you have to expect some very smart people to be motivated.

[DimPrawn]

It's not an easy thing to break these tools, but if money is involved you have to expect some very smart people to be motivated.

So Atw won't be working on this then?