A few months ago, I came across a peculiar post in an online poker forum. At first, it looked like your typical “ZOMG poker is so rigged” rant, but something about this one jumped out at me; take a look and see if you can spot it. By the way, I’ve heavily edited it for grammar, spelling, and general readability.
“Recently, I uncovered some irregularities in my hand histories. In this sample of 250 to 300 hands, a total of eight hands won when having less than a 5% probability of doing so! Now, I want to believe that this is just variance and not indicative of results in online poker in general. As such, I have begun to compile a database of unusual hands. Since I don’t play nearly enough hands to produce a worthwhile sample, I ask that you do the same. It is quite simple, just play your hands and if during the session any really bad beats occur, jot down the hand number and the details. I will create an e-mail address where you can send the data and we can put it all together and see what the results say.”
This guy wants to verify whether or not there’s any funny business going on at an online site either because there are cheaters or there is a glitch in the random number generator. Hats off to him for being proactive, I suppose, and for recognizing that he needs a large sample size to prevent skewed results due to short-term variance. But even if he gets a million e-mails from other players, the results of his experiment are going to be about as useful as chocolate pants. Why? Because of a little thing called Confirmation Bias.
According to Wikipedia, Confirmation Bias is the tendency for people to favor information that confirms their preconceptions or hypotheses regardless of whether the information is true. In other words, we “cherry pick” evidence that supports what we already believe rather than gather all available data and then draw objective conclusions based on the results. The guy who wrote this post had a hypothesis: “ZOMG poker is so rigged.” To prove or disprove it, he wanted people to send him hand histories of bad beats. Logically, if he gets a million responses, then he’ll have a million bad beats. Statistically, this will prove nothing, but it will provide him with a million reasons to believe that poker is rigged. “I knew it! Look at this avalanche of bad beats! Wow, they’re occurring 100% of the time!”
Even the “irregularities” that he supposedly uncovered are tainted by Confirmation Bias. So what if eight long shot hands won? We have no idea how many long shot hands from this same sample didn’t win. If it were eight out of eight, then maybe we have a bonafide irregularity. But it’s also possible that it was eight out of 200 long shots, which is right in line with what we might expect. A number is just a number without context. But this poster is cherry-picking only the data that supports his hypothesis and it proves nothing except that he’s paranoid.
If he really wanted to prove or disprove his hypothesis, he would need virtually every hand history from every contributor so he could compare the bad beats to the times when the player in the lead actually won the hand. Only then could he figure out whether the actual number of bad beats matched up with the expected number of bad beats. But wow, that would require a lot of work and computing power to process and I suspect this guy wouldn’t want to go to the trouble. He claims to want to know whether or not the site is on the up-and-up, but reading between the lines, I think what he really wants is just to be proven right.
Human beings want to be right. In fact, we want to be right so badly that our brains actually distort reality when it doesn’t conform to our beliefs. You’d think that this would require a lot of mental energy, especially when the evidence is piled high against us. But here’s the kicker: distorting reality almost never requires more effort than actually changing your mind. That’s what’s really hard. Who wants to go to the trouble of changing our ingrained belief system when it is much easier to twist the facts to fit what we think we know?
This is one of the reasons why so many contentious political issues are perpetually deadlocked. One side looks at the evidence, cherry-picks the data that supports their view, and proclaims victory. Meanwhile, the other side looks at the exact same evidence, cherry-picks the data that supports their view, and likewise proclaims victory. And each side can’t figure out how the other can be so blind!
Confirmation Bias is so strong that even professional researchers often have to trick themselves into ignoring it. That’s why the most strict medical trials use a “double-blind” approach where neither the subjects nor the experimenters actually know who is getting the real medication and who is getting a placebo. It’s all too easy to be fooled into cherry-picking the results and disastrously skewing the results of the experiment.
It’s as though our brains are a kind of Google Search. We type in a hypothesis like “ZOMG poker is so rigged.” And then even if there is a mountain of contradictory evidence out there, our brains’ search parameters filter the evidence and spit out only the data that supports our case. You find exactly what you look for and nothing more.
So what are your preconceptions about poker? Do you believe you’re unluckier than the average? Do you believe your flush draws never hit? Or that your aces always get cracked? Maybe it’s that ten-six of hearts is your lucky hand. Or that you play better after a few beers. Whatever your beliefs, take some time to reexamine them and use hard evidence – all of the evidence – to confirm or disprove them. Once you stop cherry-picking the results, you might be surprised to find that ZOMG, poker’s not quite so rigged after all.