Paul Hoppe offers his thoughts on the divisive topic of datamining in online poker.
Editorial/Opinion
January 20, 2012
Share:

Datamining is bad for pokeralarch, Creative Commons Attribution 2.0 License

If you’ve spent any time playing poker in casinos or card rooms, you’ve undoubtedly heard endless complaints about online poker. “It’s rigged” is probably the most common, but a close second is concern over collusion and other ways players can easily cheat online. While the day-to-day reality of online poker is nothing like what these live players imagine, it does seem like a new cheating scandal pops up every week. With sites apparently ill-equipped to catch all of these cheaters, players have long had to police themselves. But for the online poker industry to survive and flourish, this must change.

In poker, datamining is the practice of recording hand histories in which the observer did not play. “If the goal of datamining were only to catch cheaters, then the information wouldn’t be sold.”Analysis of these hand histories has allowed players to catch some notorious cheaters. While use of mined hands can help to spot anomalies which indicate collusion or other unfair advantages, they can also be imported into a personal database and displayed on a HUD during play. Anyone can use their own played hands to power their HUD, but a larger database which includes mined hands can allow the user to make better-informed decisions at the table. This provides a competitive edge over players who do not have datamined hands in their databases.

There are many companies specializing in the acquisition and sale of mined hand histories. With the rise of datamining sites, the use of such information has become almost standard practice among winning players. It’s a bit like the steroid era in Major League Baseball, where players who declined to bend the rules were at a disadvantage to those who used every edge available. While many in the poker industry condemn datamining in theory, it’s hard to ignore the fact that notorious cheaters have been uncovered thanks to analysis of mined hand histories.

My view is that if a site explicitly disallows the use of mined hands, then using those hands in one’s database is cheating to gain an “It’s a bit like the steroid era in Major League Baseball, where players who declined to bend the rules were at a disadvantage to those who used every edge available.”advantage. Cheating for an advantage in a game played for money is stealing. It’s not on the level of aggravated multi-accounting or collusion, but let’s not kid ourselves. People datamine to gain an advantage and then rationalize why; “it’s not that bad.”

I have little to no problem with the act of harvesting data. That’s not the real issue. The problem arises when enterprising individuals or companies utilize the data for personal gain at the tables or commercial gain through the sale of hand histories. If the goal of datamining were only to catch cheaters, then the information wouldn’t be sold.

There are a number of reasonable-sounding arguments in support of datamining. I will address these one by one.

“Datamining does not provide an unfair advantage.”

When you join a poker site, you enter into an agreement to abide by the site’s rules. If the site allows datamining, as many sites once did, then datamining is fine. But if the site declares datamining to be against its rules, then datamining is cheating. Those are the house rules of the game. Breaking the rules of a game is cheating and provides those that do it with an unfair advantage.

“The edge datamining provides is not that large.”

The edge gained from datamining is not overwhelming. If you don’t know what to do with the data you obtain, you’re still going to lose at poker. Mined hands are not a magic formula for victory. A great player without mined data might still beat a good player who has mined data. But if it’s “not really that useful,” then why do mining companies charge a premium for their hand histories? This isn’t some snake oil sold by charlatans. It’s real information that provides a real advantage. Better information will lead a good player to better decisions.

“Anti-datamining rules are unenforceable, therefore they’re bad rules.”

There is merit to the argument that sites should be held accountable for prohibiting datamining and then doing next to nothing to prevent it. That much is on them. Sites need to do a better job of preventing datamining if they’re going to outlaw it. But how enforceable are anti-collusion and multi-accounting rules? No one suggests that collusion should be allowed since it’s difficult to prevent. Just because a rule is difficult to enforce doesn’t mean it shouldn’t exist. It just means a better effort needs to be made to prevent and prosecute.

“Hand histories are not private property.”

This is another half-truth. Hands are observable on most sites. As a result, anyone with an account (and sometimes anyone without one) can observe what’s going on in a given hand. Sometimes the actual hand histories are available to the observer, and sometimes they have to be reconstructed artificially. But there is usually a limit placed on how many tables you may observe at once, and there are limits to human memory. If a site says that you can’t record the data from every table and store it for later use, that’s a perfectly reasonable stance. It’s like if I say that it’s okay for you to use my lawn furniture but not okay to take it. The pro-mining argument here is like saying that it’s okay to take my lawn chair because it’s not inside my house. Looking at a few hands here and there is one thing, but compiling and distributing the entire win/loss history of every player is a mass invasion of privacy.
“The fact that an activity can yield ancillary benefits does not justify that activity. Here’s a metaphor: someone comes home every afternoon to have an affair. One day they catch the gardener trying to steal some jewelry. Should the spouse not be angry about the cheating because it led to the discovery of theft?”
“Dataminers have helped catch cheaters.”

This is true. If not for dataminers, the superusers may never have been revealed, and Nick Grudzien and his collusion buddy might not have been caught. But so what? This is more an indictment of poker site security than it is an argument for allowing datamining. The fact that an activity can yield ancillary benefits does not justify that activity. Here’s a metaphor: someone comes home every afternoon to have an affair. One day they catch the gardener trying to steal some jewelry. Should the spouse not be angry about the cheating because it led to the discovery of theft?

If we seek to eliminate datamining, then we do need a better way to catch cheaters. I have a suggestion:

An independent security and auditing firm

What if every regulated poker site sent every hand history and all player data to an independent security firm? This company would need to be comprised of extremely intelligent and trustworthy former poker players capable of analyzing reams of data to ferret out cheaters of all types. Information would be kept private and secure, but there would be a level of transparency in investigations. The accused would know what they stood trial for.

With player data in addition to hand histories, investigators could compile larger and more relevant samples of data for each player by combining hands from multiple sites. When it was proven a player had cheated, they could be banned from all sites, not just one. In a regulated environment, the firm could potentially work with local law enforcement in an effort to recover funds that had already been withdrawn from the poker sites.

The firm’s investigators would be proactive and analyze the data for irregularities in both player behavior and in the RNGs of each site. There could be multiple firms which are rotated through different sites, but with a centralized mechanism for players to report suspicious play.

Having effective third-party security like this would free sites to combat unauthorized datamining by allowing screen name changes or by making tables unobservable. This would work much better in a regulated environment, since someone needs to audit the auditors. This oversight would likely be done by a gaming commission.

Poker players are right not to trust the ability or inclination of sites to effectively combat cheating. But the answer is not to turn to a new way of cheating to stop the old ways. The answer is an independent security and auditing firm.

Paul Hoppe is a professional poker player, coach and author. His books and blogs can be found at zenmadman.com and giantbuddhapoker.com, and he can be followed on twitter @ZenMadman.