Full Tilt PokerBot Scandal Post-Mortem (Part I)

Now that the roar is dying down (Full Tilt Pokerbots(?) at Two Plus Two Forums), I thought I'd wrap it all up with what I, as a bot operator, have found interesting in this monstrous debate.

#1. "...As for 1forthethumb and mariojr, I've not only played with both of them but have talked to both of them..."

Most players still believe that a few replies in the chat window "prove" your bot is a human. You should already know how easy it is to inject a few semi-random chat messages now and then. Don't get carried away, however. Lack of response is less suspicious than a nonsensical or repetitive response.

#2. "...I mostly ignored bot issues because I agreed with the popular sentiment at the time that it was just too complicated to program a bot to play well at an kind of meaningful stakes...."

The players WANT to be deceived. They have self-inflated egos that have convinced their Id that they're some sort of super-human poker calculator...pulling information from the ether, evidently. All we have to do is lay low and allow them to believe this. Just as a fish can't believe he isn't a winner, these winning players can't believe their winning isn't the result of their own extraordinary talents.
Are these players making decisions outside the realm of what most bots do? Of course, but there is no guarantee they are making the correct decision. For many low stake players, they are talking themselves into exploitable bad plays much more often than they think...they seem to believe that the amount of thought put into a decision is a direct indicator of the QUALITY of the final decision. With the human mind, this is not the case.

#3. If you get mentioned on a forum. SHUT UP.

Don't be BrandonJoseph from the main thread, trying to spin the conversation to avert suspicion. It doesn't work and it will never work. Just keep your mouth closed. You will only give them more information by which to convict you. The forum lynch mob can't clear your name...but they can sure convict you and force the site into action. Let it die!

#4. "...30,000 hands, same site, three different players logging on at the same time, logging off at the same time, and not playing at the same table for any one of those hands?..."

If you do something so obvious, you deserve to get caught. It may take awhile for the players to catch on, but the sites are well aware. Don't rely on the site operators lax attitude. It can change on a whim and I will tell you from experience that some sites will catch you and then not take action until you've either reloaded the account for a bonus offer, or you've attempted a cash out.

#5. Players don't understand the economy of large numbers.

It's clear from this thread that many, if not most, players don't understand the scale of how a bot farm works. These players are driven to move up stakes as quickly as they comfortably can, and they assume your bot is doing the same. They simply do not understand that you can make a six figure salary mining micro stakes where each table is earning $.25/hour.
Could I move up limits with my bots? More than likely, but should I spend 100 man hours coding and testing this move, or should I spend 100 hours dropping 10 more computers (80+ tables) into my war room? The choice is easy at even the smallest of limits until you run out of IP's or funding accounts.

I will break here before this post gets out of hand. So far we've only scratched the surface of this thread. More analysis to follow.

More Pokerbots Found at Fulltilt? *whistles innocently*

Alan Bostick, an interesting blogger currently writing at As I Please gives us a bit more information on the Full Tilt bot scandal(s).

2+2's favorite tinfoil hat wearer, MrGatorade, is on the witch hunt. He doesn't give criteria, just rattles off a bunch of names of heads-up poker bots at Full Tilt.

Full details at As I Please.

And the Free WinHoldem Winscrape Site Is...

*Drum Roll Please*

NOTHING.

Why? I'm not sure.

127 unique visitors came to this site since the time I posted my offer to do a free WinHoldem screen scrape by either direct reference from the WinHoldem forums, or by searching for a phrase containing the term WinHoldem. Not a single person emailed me to suggest a site for scraping.

Is WinHoldem losing ground to the newer, more user friendly poker bot platforms?

Full Tilt Rep Responds to Poker Bot Accusations

Full Tilt Poker representative "FTPSean" responds to the angry mob at Two Plus Two. As you've come to expect from Full Tilt poker, there isn't much substance.

I'll spare you FTPSean's corporate-customer-service-speak and jump right to his bullet points.

• During the investigation we found the evidence to be inconclusive in supporting either determination (human or bot).
• After careful consideration, the evidence did not warrant the seizure of funds and permanent account closure.
• We stand by our decision. Having said that, re-opening an account after an investigation such as this one does not mean we have made an irreversible decision. We will continue to reevaluate this situation.

Not too exciting, but Full Tilt is in damage control mode...I'm surprised they replied at all.

FTPSean's full post can be found at the Two Plus Two Forums .

Full details in my earlier coverage here: Full Tilt Poker Bots

Full Tilt Poker Bots (?) at Two Plus Two Forums

A massive thread at Two Plus Two regarding poker bots at Full Tilt Poker has exploded. You can read the 1,500+ replies, or see my summation below.

An observant human player (SukitTrebek) noticed some statistical similarities in the play of 1forthethumb, full_tilting, mariojr, and 0_Drunkenboxer. When I say similarities, I mean SIMILARITIES. Their stats on all streets are nearly identical over 100k hand samples each. This is a link to SukitTrebek's collected stats...decide for yourself.

SukitTrebek reports his findings to Full Tilt. The suspected bot players disappear, and Trebek assumes they are banned. 1 month later, the entire crew of accounts re-appears at the same time and returns to grinding more hands of $200NL than anyone else for 1.5bb/100.

Suspicious? Oh yes. You really shouldn't continue reading until you take a look at the stats

When the bots return, Trebek breaks his silence and the NL Bots on Full Tilt thread is born. Read Trebek's original post if for no other reason than to know what a player will go through to catch you. Full_tilter and a buddy chime in to profess their innocence, backed by Nation, a Two Plus Two forum moderator that claims to know them in person and "knows" they do not run bots.

What do I think? I can tell you right now that something fishy is going on with their stats. These guys have posted stats over 100k samples that are more similiar than any I've been able to find on ANY of my bots, and I'm admitting they are un-tilting, automatons. I've run bots for 1M hands with untouched logic code without 100k runs this similiar. If these guys are bots (or a sweatshop...there's no difference) then they are following a script, but one that is horribly static and has almost no regard for the texture of the board or styles of the opponents around them.

The only way I can see a bot with such similiar statistics is if they are using only the number of actions in the hand and their hand strength to make decisions. They obviously don't use the bet amount as a decision making tool as hand histories have been posted showing them folding to minimum bets and raises in a way that no sane player would.

In other words, folks, this is a horrible bot/script/sweatshop. There are so many factors of the game that they are obviously ignoring and they are still turning a considerable profit at $200NL, a game and limit that most ignorant human players thought was untouchable by bots. full_tilter and his crew have done it successfully and as you've seen from their stats, they are making some downright horrible plays. They are continuation betting 97% of the time at $200NL for example.

Update: A Full Tilt representative has responded. Read my article Full Tilt Rep Responds to Poker Bot Allegations.

Reason Your Pokerbot Will Be Banned #1: Session Length

There are many ways to flush the hard work of coding a bot down the toilet. Most seem obvious, but greed clouds judgement. The most important rule is to play reasonable session lengths!

Chances are that you will have an account or two banned as you go through your career as a botter. Don't panic. Before you begin, just assume it is going to happen and plan accordingly.

The most common reason for being banned (and your bankroll siezed more than likely) is the simple act of playing an unreasonable amount of hands per day/week/month or playing sessions of inordinate length.

Need an example? Datamine a day of Ultimate Bet micros. The bots there are so obvious it is comical. You'll find quite a few players there that multi-table all day, everyday, to suck up the UB "Points for Cash" free money. They aren't the smartest bots in the world, most either being a very predictable nutpeddler or a typical TAG that takes his hands too far. The point, however, is that they sit there all day, every day, in a way that any human regular at the tables will detect.

Most of your problems will stem from human players reporting you.

Again, using UB as our example, it is obvious that even a rudimentary security check by the site would reveal these bots to the site operator. UB has a record of not finding bots until they cashout, so I think these players will be in for a rude awakening at some point. Regardless of the site's vigilance in detecting a bot on their own, they WILL react when the customer complaints come rolling in...and if you play 80 hours a week, 6-tabling $10NL, the complaints WILL come.

Whether you are playing for testing, or playing for profit, you MUST resist the temptation to play more than is reasonable for a player. Also, take note that "reasonable" differs between levels. A reasonable $1/$2 semi-pro might run 5 or 6 tables for 8 hours in a day. A reasonable $.01/$.02 player will most certainly not.

Slow down. Use your head. This is a long ride.

7 Things You Must Know Before Using Online Poker Robots

Online poker robots are not easy cash machines. Before you begin writing your own code (or tweaking code you've purchased) you should consider that this will not be NEARLY as easy as you expect. Some points to consider...

1. You don't know as much about poker as you think you do. I can't stress this enough. I was a winning player before I turned loose my war room on the online poker world. I played at mid-limits up to $5/$10 and $10/$20 online, and higher at brick and mortar poker rooms. However, the first thing I realized was that I take some of my play for granted. I did the right things, but I didn't know why. Imagine trying to teach another person to speak your native tongue. Sure, you speak it very well yourself, but can you explain past-perfect tense? Subjective pronouns? Probably not if you've been out of school as long as I have, but this is a perfect analogy for how you teach your bot to speak the language of poker.
   


2. Nothing with this much earning potential is easy. Expect to spend hundreds of hours creating a bot that makes any sort of profit above the lowest of micro stakes. Read that again...HUNDREDS OF HOURS.
   


3. Don't build a rule set, build a brain. You will never be happy with a rules based robot. Trust me, I've had several. They are a quick way to get up and running but you'll find a nearly limitless amount of situations to be faced in this wonderful game. It is inevitable that you'll miss some and make EV- plays at the table. Instead, concentrate on creating a brain for your bot that uses more information than your hand strength and number of villains in the hand. Any good player will take money from an opponent that can't think above the first level of poker (seeing his own hand only). Your bot is no exception.
   


4. You don't understand variance. I thought I did when I began. I had Poker Tracker. I had tens of thousands of hands recorded of my own play. However, it is a whole new ballgame when you start recording tens of thousands of hands per day of the identical playing style and logic. I've had runs of 100,000 hands at break even, EV-, and EV+ all from the same bot.
   


5. Security is paramount. Don't screw around and get your accounts banned. You are thinking, "Oh, it's just $50 I put in to toy around with penny tables. It won't kill me" $50 loss won't hurt you but take into consideration what that account may be worth in the future. 6 months from now after you've built a killer bot, that account may have a very large earning potential. What I wouldn't give to have all the accounts back that I foolishly squandered away.
   


6. Don't be greedy. So you've built a winner...don't go and ruin it all with 12 hour sessions of 12-tabling. Both the site and the players will notice this. Also, don't think you can farm the micros mercilessly forever. If a human wins at the micros over a long stretch of hands, he moves up limits. Your bot should do the same. Do you expect a site to not notice that you've played 50k hands of $.25/$.50 poker every month for a year?
   


7. Get a second opinion. Post hands to forums for discussion. Get a session review buddy. Do whatever you have to do to get an outside perspective on your bot's poker play. It's very easy to let all of your exposure to the bot influence your own game and theory. Resist the temptation to defend your bot's play automatically. You might be talking yourself into reinforcing a bad habit that it taught you.

A New Era? The PokerBot World Championship

Ray Bornet, the creator of WinHoldem, has always been the most outspoken proponent of poker bots. Since the days when usenet newsgroups were the heart of online poker discussion, he's been challenging anti-bot discrimination. In recent months, Mr. Bornet has begun to shift his attention to a completely new form of poker: "Live-Digital".



Live-Digital is the term given to an event where the participants are physically present in the same location, but playing from their computer. As I understand it, any computer that can connect to Ray's proprietary PPRO server can participate and use any type of poker aid they'd like short of collusion. There have been pokerbot competitions before, most notably the University of Alberta's work, but Mr. Bornet has mixed it up in several distinct ways.

Live humans are allowed, even encouraged to play.
There are entry fees up to $10,000.

Now think of the ramifications. For the first time ever, we could see the poker equivalent of Kasparov Vs. Deep Blue. Exciting. I personally won't be attending. I do hold a Winholdem subscription, but I haven't used it for play in a long time. For $100/year, though, it's a cheap way to see what my pro bot competitors/peers are working with. Winholdem is easily the most powerful, publicly available pokerbot if you can get past the steep learning curve, but I've moved to my own proprietary bot long ago. I do credit Winholdem with pointing me toward the path of knowing what it is I don't know, as well as what it is I need to know, in order to properly operate a poker bot war room. Unfortunately there are a few things that are scary above and beyond the work needed to port my bot to PPRO server format.

• Anonymity: Mr. Bornet is intent on pushing poker botting out of the shadows and is looking to publicize the names of winners. I cannot risk that. My income is at stake. I disagree with Mr. Bornet in the value of being named the winner, I guess. I can't imagine it being anywhere close to the loss of income I'd feel from having all the poker room accounts under my real name suspended and bankrolls seized. Plus, the net is full of tinfoil hat wearing wackos looking to blame anything for their losses other than themselves. I don't want that guy tracking me down, convinced I'm the reason he's bankrupt himself at Party Poker.


• EV-: The event is taking place on a cruise ship for jurisdiction purposes. This adds expense and opportunity cost. There is a very small list of confirmed participants thus far and I doubt the larger purse events will even run. This only adds to the relative cost of travel and the cruise itself compared to buy-in.

It's a great idea, but perhaps a bit ahead of its time. Will it be the next WPT or even WSOP? Probably not anytime soon, but with Bluff magazine expected to cover the event and interest from the academic world rumored, the only thing keeping the PBWC from exploding is a celebrity face.

For more information on the PokerBot World Championship, see the official PBWC Forum.