This post is part of a (somewhat illustrated) series on poker bots and their threat to online poker. If you’d like to be notified whenever I make a new post, please subscribe to the RSS feed or follow me on Twitter. (See the sidebar on the right.) In this post, we will be looking at the different public poker botting communities that have existed over the years. 

The history and social order of our electronic overlords

Humans are social animals. Even the evil masterminds of the world have people they see as friends. (Wannabe) poker bot operators (from now on called botters) are no different. They too have gathered in places to discuss poker strategy, AI techniques, bot stealth technologies, the latest developments in the security of poker sites, and football. These forums allowed them to form groups, starting “companies” to engage in the act of botting together. In some ways, these evil people seem surprisingly human.

Many of these forums are public. We’ll take a look and see what we can find out about the average botter who visits them. Since many parts of this world exist in the shadows, this overview will never be a complete map of the botting world; we can only get a glimpse. But, in my opinion, that glimpse is interesting enough.

If you missed the first part of this series, I strongly recommend you read that first. I’ll be using the terminology introduced there in this article, and it might clear up some misconceptions as well. 

A disclaimer

While I’ve been in poker for a while, I’m still pretty young. I got into poker around 2005, so I missed much of the early developments. I didn’t experience all of the things described in this post first-hand, so it’s quite possible that some stuff will be slightly wrong. Either way, I doubt it will matter much, and I’m sure someone will correct me in the comments.

Going back to the nineties

The first games of online poker were played in the early nineties. This wasn’t done on fancy poker clients like we have now. Instead, it looked like this:

In case you don’t recognize this from the format: it’s IRC. IRC is a chat protocol. Internet users have used IRC since 1988 to chat over the internet. It consists of servers, which are subdivided into channels: (click to enlarge)

Internet users around the world connected to one or more servers of their choice, which each hosted a variety of channels. After connecting to a server, the user could join some of those channels. Inside a channel, he could chat with other people in that channel.

As more people started to use IRC, it became used for a variety of activities. It developed a warez scene (e.g. sharing of illegal material), and people started to play games on it. One of those games was poker.

To be able to play poker on a chat program, some special software was needed to deal the cards and manage the actual game transactions. They called this software a poker bot, since non-human users of IRC were called bots. One of these poker bots was PBot, the history of which is described here. Just to be clear again: the usage of the term “poker bot” back then is completely different from the way we use it these days.

Anyway, IRC (and the internet in general, in the nineties) was used mainly by tech-savvy people. So, it is no surprise that people quickly started to write their own poker-playing bots for IRC as well. The communities around these IRC poker channels were the first to discuss poker bots. Since no real money was involved back then, there wasn’t such hostility against them as there is today.

Unfortunately, not much written history is left of this time, and I was barely alive at the time. So, I do not have much more to add about this time. I did want to mention it, since it’s where everything began, both poker botting and online poker in general.

The poker boom

As the modern poker sites developed and the poker boom came along, interest in poker bots grew greatly. It’s easy to see why: players were making hundreds to thousands of dollars per day (during the poker boom) – now suppose you had a bot that could do the same. You could go to sleep while the bot made you a millionaire. And what if you let it run on many accounts at once?

So, many people started working on poker bots. Luckily for players, it’s not very easy to make a strong poker bot. Otherwise online poker might have been killed off rather quickly.

WinHoldem

In 2004, a program called WinHoldem was published, which allowed its users to easily write and use their own bots.

The main interface (click to enlarge)

You could write AIs by writing formulas, a weird programming language of sorts. You would write these in the formula editor: (click to enlarge)

A formula for determining when to raise pre-flop looked something like this:

0
 //Always Raise Pre Flop
|| [ br==1 && { $aa || $kk } ]

 //Usually Raise Pre Flop
|| [ br==1 && { $qq || $aks || $ak } && { dealposition/nplayersdealt > .499 } ]

//...

The 0 indicates the default action, fold. The || symbols denote “or”, and br==1 determines if the betround is pre-flop. So, it will raise if it has aces or kings, or if it has queens or ace-king and it’s in medium-late position.

Since you have read my first article in this series, you should recognize the type of AI you can write with this: rule-based bots. As we also noted there, these are usually not very strong. So, while the introduction of WinHoldem made botting much more accessible for the general public, it was of no big danger to online poker. A second reason for that is the fact that sites simply started looking for WinHoldem installed on computers, and banned those users. ((Of course, a cat-and-mouse game of stealth and detection followed.))

WinHoldem became quite popular, and had an active community on its forums. They discussed strategy, technical issues and many other things. At some point, various labs formed. These labs consisted of groups of forum members, who worked together to create a certain type of bot. Some were more successful than others; some of these labs died, others were supposedly successful at writing a profitable bot, and yet others had their products leaked to the public.

Some products around it were developed: WinHoldem supported plugins in the form of custom DLL files. DLLs were developed by the community for several tasks, such as querying a Poker Tracker database for opponent stats.

The great split

Some people on the forums were unhappy with WinHoldem, its creator and the community, so they moved away. They created OpenHoldem, an open-source WinHoldem clone. They also created their own community forum at maxinmontreal.com. Different levels of private forums were introduced, so only some had full access to all the content. OpenHoldem is actively developed to this date. I personally suspect that most users of it use it for the built-in scraper ((See the first post to learn about scrapers.)), and that they write their AI in other languages than the standard formula format.

Academic interest

Over the years, academic interest in poker AI also increased. At this time, many papers have been published about it. The University of Alberta was one of the first institutes to do in-depth research into poker AI with its computer poker research group (publishing about their first bot in 1997). They are also responsible for the annual poker AI competitions, where academic (and some non-academic) AIs bet and bluff for first place.

The CPRG also created the commercial Poker Academy Pro, a downloadable program using which you could play against their AIs. It was presented as a tool for players to train their poker skills against. Version two included an online component, using which players could play against one another (for play money), and against CPRG’s strongest bot: Polaris. Unfortunately, the website of this program no longer seems to be working.

The website contained a forum, with a small community, mainly consisting of people attempting to write AIs similar to the ones Alberta published about (using techniques such as counter-factual regret minimization, the basis behind Alberta’s strongest poker AIs). These forums were relatively dormant.

The state of the art: PokerAI.org

Currently, the most active community is over at PokerAI.org. Their forum consists of several parts, mainly:

• General forums, where poker AI related discussions are held.
• Poker papers, where the latest academic papers related to poker AI are posted.
• Online botting, where poker sites and their security are discussed.
• Restricted forums, which you can only access by showing to be a valuable contributor. I don’t have access to this, but I’ve heard that there’s nothing exciting in there.

Users on this forum openly discuss ways to circumvent the security of online poker sites, they help each other implement AI techniques, etcetera. I advise every poker player to browse these forums for a while, just to grasp the scope.

Others

Of course, there have also been others that I have not mentioned. For example, some people have sold ready-made bots to the general public. One of these parties is Shanky Technologies. Their product contains pre-written strategies, but like WinHoldem it can be programmed with custom rules. It has a community of users, that also shares custom strategies on their forum.

The poker bot market has also historically been full of scams. You can find tons of videos on Youtube of the newest superbots, available for a measly $39.95.

Observations

Looking at the botting communities over the years, one can make several observations. One is the fact that ninety-nine percent of people who set out to build a poker bot fail. Many fail before they even start: they might buy WinHoldem (which you need to write your own strategies for) and assume that they can click one button to make it win money. Others decide to write their own bot without having ever programmed “Hello world” before. Then there’s the people who are not good at poker, but decide to write a rule-based poker bot. A rule-based bot will never be substantially better than its creator.

The people who are most likely to succeed are people who have a lot of experience programming, and who are academically interested in poker AI, and who are able to understand the papers published on the topic. They are able to take the techniques described in those papers, implement them, and then work from there. This takes a huge amount of time, so even among these people, the success rate is low. However, if someone like this does not succeed at getting a working bot, I wouldn’t call that a failure. He has still gained useful experience with AI, mathematics and computer science (and the poker world is saved from another bot!)

What’s up next?

I still haven’t decided on the exact future of this series. However, I’m planning to cover at least the following:

• Current state of poker botting and counter-measures taken by poker sites.
• Detailed description of various types of AI.
• A detailed statistical analysis of some hand histories to find some bots.

I’m also considering writing a tutorial of sorts into writing poker AIs (not bots). It’s extremely fun, and very challenging as well. You might learn computer programming along the way! Anyway, we’ll see. If you have any ideas, or if you have any other questions, suggestions or comments leave them in the comments below or send them on Twitter.

To stay informed, please follow me on Twitter, or subscribe to my RSS feed. If you like this series, please tweet about it, send it to friends, etc. The more people read it, the stronger the incentive for me to hurry up and continue writing.