Online Poker and Data Mining

Paul Nobles from Bluff takes a rather interesting look at the whole data mining brouhaha that’s been going on of late. I started to leave a comment on the Pokerati site but my response started to get a little unruly in terms of length so I thought I would just direct you there to read the original and then give some of my thoughts on the topic here.

First off, I think Paul hits the nail on the head. His analysis of the situation is very good. The reason I chose to respond is that I think there are some additional lines of thought that are worth exploring as well.

Okay, you read my blog, which means you’re probably too lazy to read the entire post over at Pokerati so I’ll summarize Paul’s post by saying:

TableRatings and similar companies are in violation of the T&C’s at many poker rooms by collecting table data and sometimes even hand histories. And although the poker sites say they are against it they haven’t done much to stop it (with the exception of Cake).

First off, having seen this issue via the operator’s perspective, it’s really a matter of priorities and whether or not you want to draw attention to the problem. In many ways it is similar to the problem rooms have with poker bots.

All rooms agree that bots are bad for business but only insomuch as they are bad if people are aware of them. If your average players aren’t aware of poker bots you don’t want to be doing too much to draw attention to the fact that there was a problem to begin with.

This is also one of the reasons so few poker rooms took to sticking the knife into UltimateBet or Absolute during their cheating scandals. The vast, vast majority of players don’t read 2+2, PokerNews, or much other poker media other than poker television programming and such. Crowing that your site can be trusted while UltimateBet and Absolute cannot makes people aware of a problem that they might never had even suspected existed.

So, the average player out there doesn’t even know these data mining companies exist. Should you make too big of a deal about it and bring attention to something that you might not be able to prevent anyway? That’s the million dollar question.

Stuff like this will always be a game of cat and mouse as long as there is money to be made. As soon as any of the rooms makes a change the data collectors will counter. The only thing you can really do is make it too expensive for them to keep up and thus remove the financial motivation to offer such a service.

But, how big of a priority is this? How many customers does TableRatings actually have? How much is it really impacting the game? How noticeable is it to the average player? Will addressing the problem bring unwanted publicity that is more harmful than the actual problem?

These are the kinds of questions card rooms will ask themselves before invest resources in fixing the problem. Believe it or not, despite the massive amounts of money being made by online poker sites, most are woefully understaffed. They only have the ability to focus on a certain number of priorities and where an issue like this ranks for them will be a function of a combination of the personal philosophy of the decision maker and business realities.

Nobles does mention a possible solution of allowing players to change their screen names which is an idea that I know has been kicked around inside the online poker industry for years. The obvious argument against is that your high-raking players will move onto a site where they can maintain hand histories on other players.

Personally, I support frequent screen name changes. Lee Jones, formerly of PokerStars fame and now card room manager at Cake Poker (who, does allow screen name changes) came out in favor of this some time ago, if I’m remembering correctly.

The major problem with this solution though is that it would be very hard for most existing rooms to shoehorn into their existing software. First they have to build the ability in and then they have to rewrite portions of all of the backend tools. That’s actually not as simple as it sounds (if it did sound simple).

Many rooms don’t officially allow name changes as part of their policy but under extreme circumstances will permit a name change. I would guess that at a minimum they have a manual process in place to do it but they either don’t want to open it up to the public for fear of the repercussions or it’s not automated enough or reliable enough to do on any sort of massive scale (perhaps even as crude as opening up a second account and importing all the data and balances over).

There’s also the issue that this does make certain types of fraud easier to commit. Nobles downplays this risk but I think he underestimates the potential based on the argument others are making. Yes, the problems at Ultimate Bet and Absolute Poker were internal issues but that does not mean that every poker room has state of the art collusion or fraud detection. UB and Absolute might have been the absolute (no pun intended) worst but look at what it took to catch those players who were OBVIOUSLY cheating.

Additionally, now you’re going to have players who say that they are absolutely positive that Player A is Player B because of some convoluted association they’ve connected together and thus based on their combined stats are cheaters. Poker rooms will end up facing regular accusations of fraud going on on their sites and since poker rooms tend to not release much in the way of details into their investigations it gives the tin-foil hat wearing crowd more ammunition to proclaim that online poker is rigged.

So again, you end up right back at the question of the priority of addressing the issue. Is it worth the costs? Does it create more problems than it solves? Is the company willing to skip features in the next release or two in order to implement a solution?

Each room will make their decision on a case by case basis but up until the last few weeks I’ve only ever heard it talked about as a philosophical debate within the industry. It’ll take a lot more pressure from players before this works its way into the spotlight.