Official Full Tilt Poker Response to Bot Thread

[Balder Dash]

[ QUOTE ]
Anyone who says Full Tilt wants bots is talking out of their ass. It's so ignorant to make statements like:

[/ QUOTE ]

They're given a pile of convincing evidence that a group of players are bots, and they choose to do nothing. If they didn't want bots, they'd get rid of them.

I suppose it's possible they're really so insanely stupid that they can look at this evidence and make the post that started this thread, but I'm not willing to credit someone with such stupidity when they make a profit from it.

[ QUOTE ]
Get real.

[/ QUOTE ]

You first.

[Our House]

So you honestly believe that the whole one month investigation and account freezing was just a facade?

[NajdorfDefense]

[ QUOTE ]
[ QUOTE ]
Serious post time:

Anyone who thinks that nlnut isn't using bots is either stupid, ignorant, or doesn't understand poker very well. It's obvious he's using a bot...or his team is using bots. Now, while I concede there may not be enough evidence to warrant seizure of funds, it certianly seems that at least closing their accounts and mailing them a check for the funds is at LEAST the very minimum that should be done.

How can someone seriously argue that those post flop stats are "based off of a secret strategy"...give me a break. Either Full Tilt doesn't care and just wants the rake money, or their fraud team is headed by a bunch of morons. (I highly doubt they'd risk everything by allowing bots, so I'm assuming that their security team is just retarded)

Step it up guys, this level of incompetence is highly unacceptable.

[/ QUOTE ]

I agree with all of this. These players should not longer be welcome at the site.

[/ QUOTE ]

Agreed. Anyone else getting the sense that FTP is Moe in the Flaming Moe's episode, and we're Homer, trying to yell "You just lost a customer!' at an increasingly uncaring and unhearing Moe?

[SweetPea]

[ QUOTE ]
If the evidence was found to be inconclusive, why were the account not closed? If there is even a small chance the accounts could be bots at all the accounts should all be closed and the funds seized.

[/ QUOTE ]

Sorry, mate. I'm sure you are upset by all this as most people are, but this sentence is ridiculously dumb.

[pyedog]

[ QUOTE ]

If these statements were true, why on earth would FT have run the one month investigation to begin with? To make it appear like they're trying (because they KNEW someone was going to make a 2+2 zoo post a month later and they wanted to cover themselves)??

[/ QUOTE ]
I believe that the big poker sites like to provide the illusion that they are implementing security measures to stop botting and collusion without actually doing much at all. It is all about making the customers feel secure and comfortable without having to actually spend the time, money and resources on doing the security work.

As for the one month "investigation" who knows what that actually entailed. Some poker sites take a week just to reply to a simple email. So they probably sent an initial stock email in the first week, spent two weeks pretending to investigate (consisting of maybe one hour of actual research) and then another week preparing the reply email.

I do acknowledge that detecting bots and collusion is a very difficult problem. I have no idea how I would go about doing it if I were them. But it seems pretty obvious to me that they are terrible at detecting suspicious behaviour if these accounts were never flagged previous to Trebek pointing them out. And even after that it still wasn't enough for them to find their behaviour suspicious after a full month investigation.

What do you guys think that FTP would have had to find before they could make a conclusive determination of bots? I'm not even sure how much more evidence there could be. Basically bots are indictable according to FTP. Maybe we should all just get into botting then. The chances of getting caught and punished are even lower than the chance that these accounts are not cheating.

[Sniper]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
FTPSean, if you know anything about stats, I can show you tests proving (to the furthest extent I can) that there is no way these users cannot be bots.

[/ QUOTE ]

Can your stats differentiate between BOTs and Humans w/ Decision Support Software?

[/ QUOTE ]

The results I'm getting make it very unlikely humans had more than a tiny impact on the hands these users played (in other words bots played almost every single hand).


For example, let's say I test Ho: p1 = p2 vs. Ha: p1 != p2, and I want to compare vpip of 1forthethumb to full_tilting.

Using a two sample test for proportions, we get a test statistic of -.00005! This means it's pretty much impossible to reject Ho, that the vpip for the two players is the same. I think you will find similar results comparing pfr, betting, raising, cbets, etc.

[/ QUOTE ]

You didn't answer my question...

Players using the same Decision Support Software Profile would be expected to have converging stats over a large enough sample size. How can the stats alone determine if a human or a bot was actually pushing the buttons?

[neverforgetlol]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
FTPSean, if you know anything about stats, I can show you tests proving (to the furthest extent I can) that there is no way these users cannot be bots.

[/ QUOTE ]

Can your stats differentiate between BOTs and Humans w/ Decision Support Software?

[/ QUOTE ]

The results I'm getting make it very unlikely humans had more than a tiny impact on the hands these users played (in other words bots played almost every single hand).


For example, let's say I test Ho: p1 = p2 vs. Ha: p1 != p2, and I want to compare vpip of 1forthethumb to full_tilting.

Using a two sample test for proportions, we get a test statistic of -.00005! This means it's pretty much impossible to reject Ho, that the vpip for the two players is the same. I think you will find similar results comparing pfr, betting, raising, cbets, etc.

[/ QUOTE ]

You didn't answer my question...

Players using the same Decision Support Software Profile would be expected to have converging stats over a large enough sample size. How can the stats alone determine if a human or a bot was actually pushing the buttons?

[/ QUOTE ]

That's more of a practical concern, I find it unlikely they all sat in a room pressing the same buttons for every type of situation.

[Vavavoom]

[ QUOTE ]
Full Tilt Poker: Don't Learn, Sometimes Chat, and Play with My Bots.

[/ QUOTE ]

FYP nlnut/nation et al ...

[KurtSF]

[ QUOTE ]
[ QUOTE ]
TRANSLATION: They sent us their utility bills and credit card information. There is no credit card fraud going on, so we no longer care.

[/ QUOTE ]

hahaha

I don't think these guys are bots and even I think thats exactly what happened on FullTilts end

[/ QUOTE ]

I do think these guys are bots and even I think that's exactly what happened on FillTilt's end.



[ QUOTE ]
[ QUOTE ]
• During the investigation we found the evidence to be inconclusive in supporting either determination (human or bot).

[/ QUOTE ]

So your investigating team could not reach a conclusion that multiple accounts playing 100K hands/month with the same stats from the same IP address were humans or bots yet they never asked the players for any more information to help them reach this conclusion (according to nlnut -except for maybe IDs and utility bills)? [img]/images/graemlins/confused.gif[/img]

[/ QUOTE ]

I think its pretty clear they either didn't do their own statistical investigation/analysis, or that they completely failed to grasp the significance of the stats. Or both.



[ QUOTE ]
Anyone who says Full Tilt wants bots is talking out of their ass.

[/ QUOTE ]

Actions speak louder than words. FTP, and other major poker sites, are acting in such a way as to conform to this ideal (tolerate bots for profits) even if their words say otherwise.


Edited to add:

[ QUOTE ]
<font color="green">

If I may make one request, it's the following: you (at FTP) have access to every single HH played on your site. The most damning piece of evidence in this thread was the similarity of the many "sweatshopper's" stats which, although they may not be out-and-out bots, clearly (to EVERYONE) indicated some type of automation. Throughout the 1,000+ posts in the other thread, that was the one piece of evidence that no one -- and i mean NO ONE, not even NLnut's supporters -- could explain satisfactorily. We even had stat geeks post their detailed calculations showing how incredibly improbable attaining such uniformity, ESPECIALLY in post-flop actions, would be. I, for one, would be very grateful if you could tell me whether, looking at the unlimited HHs that you have access to and talking it over with the stat geeks on your end, could confirm for us -- and give reasons how -- individual players could possibly achieve that kind of uniformity. You don't have to give us any 'evidence' from THIS case in particular...I just want you, given your access to millions upon millions of HHs, to help this forum understand how those post-flop statistics could come about without some kind of automation. I think this is a fair and reasonable request, and one eminently within your powers.</font>

[/ QUOTE ]

[Sniper]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
FTPSean, if you know anything about stats, I can show you tests proving (to the furthest extent I can) that there is no way these users cannot be bots.

[/ QUOTE ]

Can your stats differentiate between BOTs and Humans w/ Decision Support Software?

[/ QUOTE ]

The results I'm getting make it very unlikely humans had more than a tiny impact on the hands these users played (in other words bots played almost every single hand).


For example, let's say I test Ho: p1 = p2 vs. Ha: p1 != p2, and I want to compare vpip of 1forthethumb to full_tilting.

Using a two sample test for proportions, we get a test statistic of -.00005! This means it's pretty much impossible to reject Ho, that the vpip for the two players is the same. I think you will find similar results comparing pfr, betting, raising, cbets, etc.

[/ QUOTE ]

You didn't answer my question...

Players using the same Decision Support Software Profile would be expected to have converging stats over a large enough sample size. How can the stats alone determine if a human or a bot was actually pushing the buttons?

[/ QUOTE ]

That's more of a practical concern, I find it unlikely they all sat in a room pressing the same buttons for every type of situation.

[/ QUOTE ]

And if they are pressing the same buttons in the same situations, because the same decision support software profile (that seems to be allowed by the sites) tells them to push the same buttons?...