Official Full Tilt Poker Response to Bot Thread

[Skallagrim]

OK, I trust the regular posters here on statistics (my previous posts all recognize that these are EXTRAORDINARY numbers).

But now listen to an informed perspective on the "criminal" mind: suppose I develop a supertight system that, as long as its undetected, can acheive a marginal win rate at low stakes NLHE (hardy that difficult). I program a script with that system (again easy). I set up a group of computers AND PLAYERS to sit and execute this script over set periods of time. I am present at these times to oversee compliance and occasionally play. The players really make no decisions, they just make the clicks instead of the script. I do this, even though a real bot would be easier and cheaper, precisely because I see it as a loophole in the rules.

The above scenario is what I believe happened. It technically is not a bot nor a violation of the TOC. But it sucks none the less.

Can you statistics guys assure me that the above scenario is not consistent with the stats?

Skallagrim

[Adde]

[ QUOTE ]
[ QUOTE ]


If playing the same system, in the long run they all would get hungry and sleepy and make mistakes at an even rate, making their numbers look similar.

[/ QUOTE ]
Only if they were written by the same programmer.

gogogogo

[/ QUOTE ]

I take it you ran out of arguments and played the joke card.

[cardcounter0]

[ QUOTE ]
If playing the same system, in the long run they all would get hungry and sleepy and make mistakes at an even rate, making their numbers look similar.


[/ QUOTE ]
Random humans executing a complex task which required taking many notes and having in-depth discussions on actions under a limted time frame will exhibit the same error rate?

Sorry, I thought you were the one that was joking.

[Sniper]

[ QUOTE ]
OK, I trust the regular posters here on statistics (my previous posts all recognize that these are EXTRAORDINARY numbers).

But now listen to an informed perspective on the "criminal" mind: suppose I develop a supertight system that, as long as its undetected, can acheive a marginal win rate at low stakes NLHE (hardy that difficult). I program a script with that system (again easy). I set up a group of computers AND PLAYERS to sit and execute this script over set periods of time. I am present at these times to oversee compliance and occasionally play. The players really make no decisions, they just make the clicks instead of the script. I do this, even though a real bot would be easier and cheaper, precisely because I see it as a loophole in the rules.

The above scenario is what I believe happened. It technically is not a bot nor a violation of the TOC. But it sucks none the less.

Can you statistics guys assure me that the above scenario is not consistent with the stats?

Skallagrim

[/ QUOTE ]

Of course they can't...

But you forgot 2 things in your "hypothetical scenario"...

1. After the script read card info, it would cover them on sceen, so a human couldn't think about the situation, just execute the proper scripted click on command.

2. A central command "station" that would evaluate PT tracking data on the fly, to assure that everything was within standard variance, and specifically that any particular opponent wasn't exploiting the system.

[Adde]

[ QUOTE ]
[ QUOTE ]
If playing the same system, in the long run they all would get hungry and sleepy and make mistakes at an even rate, making their numbers look similar.


[/ QUOTE ]
Random humans executing a complex task which required taking many notes and having in-depth discussions on actions under a limted time frame will exhibit the same error rate?

Sorry, I thought you were the one that was joking.

[/ QUOTE ]

They aren't random. The task isn't that complex.

[cardcounter0]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
If playing the same system, in the long run they all would get hungry and sleepy and make mistakes at an even rate, making their numbers look similar.


[/ QUOTE ]
Random humans executing a complex task which required taking many notes and having in-depth discussions on actions under a limted time frame will exhibit the same error rate?

Sorry, I thought you were the one that was joking.

[/ QUOTE ]

They aren't random. The task isn't that complex.

[/ QUOTE ]
They aren't random?
At last we agree. They were written by the same programmer.

gogogogo

[Adde]

Three friends using the same system can't hardly be called random.

[cardcounter0]

[ QUOTE ]
Three friends using the same system can't hardly be called random.

[/ QUOTE ]
"Friends" is a poor criteria to use to determine error rate of performing a task. So yes, they are random. Unless some testing criteria was given, like a timed test putting pegs in holes or something, before they became a "friend".

[Skallagrim]

Hey Sniper, I consider myself an expert in certain fields, but computer assisted poker is not one of them.

I appreciate your reply, but quite honestly I, and I suspect many other non-experts, want to know if the "hypothetical scenario" with your improvements is reasonable? Or possible? Or even most or least likely?

It seems some still insist it must have been bots ... So I want someone to address my guess as to the truth specifcally. It is, essentially, my "reasonable doubt" in this situation.

Skallagrim

[Brewmeister B]

[ QUOTE ]
Hey Sniper, I consider myself an expert in certain fields, but computer assisted poker is not one of them.

I appreciate your reply, but quite honestly I, and I suspect many other non-experts, want to know if the "hypothetical scenario" with your improvements is reasonable? Or possible? Or even most or least likely?

It seems some still insist it must have been bots ... So I want someone to address my guess as to the truth specifcally. It is, essentially, my "reasonable doubt" in this situation.

Skallagrim

[/ QUOTE ]

I believe your scenario is very likely given the facts:

-Long time to make trivial actions - a bot would most likely take less time to act unless specifically programmed to "think" for a while.

-The players never improved their strategy over time - nlnut admitted that they didn't want to change their system as it was making them enough money as is. Sounds like BS since if they are making some money with this system with known flaws, then more money by improving it would be better. That is unless they have a simple system programmed into a script and modifying it to increase the win rate would cause too much complexity or they do not have the technical knowhow to program more than just a basic system.

-The win rate is fairly low - some people have argued that NL is too complex to program because of the almost infinite possiblities of situations. This is true if they wanted to win big, but if they are willing to live with a low win rate, programming a tight playing system that generalizes many similar post flop situations will do exactly that. It will clobber the fish and get exploited by the sharks that figure it out, providing a small +EV.

-Low variance in pre AND post flop stats - some have argued that this is a case of bots preflop with human interaction post flop, again because NL is too complex to program. If this were the case, then the postflop play would have much more variance than the preflop play.