How Online Poker is Like Quantum Mechanics

Date October 14, 2007

I was listening to this week’s Ante Up podcast, and the guys had a long argument about something that has always bugged people who play poker online: how do we know these sites aren’t screwing us over? It would be facile for a poker site to rig the deal in various ways, such that the cards a player is dealt aren’t truly random.

The guys on the show were arguing about two things: 1) continuous shuffle, and 2) weighting toward action hands. Unfortunately, I don’t think any of them really knew at any point which, if any, of these things they were actually talking about. I’ll address them one at a time, and give my opinion about both.

  • Continuous Shuffle

    A standard randomizing shuffle algorithm (since I don’t know what real nerds call it, I’ll refer to it as a “static shuffle algorithm”) would do one random shuffle before beginning dealing. It’s the digital equivalent of what they do at a casino—before each hand is dealt, the dealer shuffles the full deck. Once this shuffle is completed, the order of the deal is set in stone. If someone had x-ray vision, he could look at the deck at any point and predict with 100% accuracy what card would be dealt next at any point in the hand.

    So online, if you’re using a single random shuffle, the order of the cards being dealt is set from the start of each hand. Theoretically, if someone could crack this algorithm it might be possible to predict what cards would come out (or, perhaps more likely, to predict with a bit more accuracy the chances of a given card coming out at a given time). Either way, the hypothetical hacker would have a significant edge over the honest poker player.

    Obviously, sites don’t want people cracking their randomizing algorithm—it doesn’t get them anything to allow certain players to cheat in this way (in fact, it seems pretty clear to me that it would hurt a poker site to be seen as vulnerable to this kind of cheating, since players would probably just take their business elsewhere). So to increase the difficulty of figuring out the next card, poker sites use an additional level of complexity: the continuous shuffle algorithm. With this in place, the poker site is constantly reshuffling any cards that haven’t already been dealt out. The next card to be dealt is perpetually changing, so that even if you could figure out in one moment what the deck’s order is, it would be randomly changed by the time you tried to use that information.

    Now what does this mean for you as a player? Assuming that the randomizing algorithm is, in fact, truly random, it means absolutely nothing is different. Your odds are identical, with or without the extra step of the continuous shuffle. This is Schroedinger’s Cat—until you see your next card, it is all possible cards and none of them. The only way this should make a difference is if you believe in the supernatural (in which case, maybe we should be talking about what gods to make sacrifices to before tournaments, instead of implied odds). Casinos preserve “your” cards in the case of a misdeal as a nod to the credulous people who believe the random card they “should be” dealt occupies some special place in the cosmos. It doesn’t.

    Okay, so to summarize: as long as the randomizing algorithms are legit, it makes no difference whether a poker site uses a continuous or static shuffle algorithm. The only distinction is that if you’re worried about someone cracking the shuffle order, a continuous shuffle algorithm is probably a little bit safer—it adds another level of complexity, increasing the difficulty of cracking it.

  • Weighting Toward Action Hands

    Okay, this is the other topic. The idea here is that although the algorithms are random (in other words, the order of the cards is not pre-determined or predictable), the odds of the randomizing algorithm are weighted toward results that lead to more action. In other words, two good hands go up against each other more often than they would in a real life game played with a physical shuffled deck. This seems like a very real possibility, if you play a lot online—high value hands go up against each other all the time. I’ve lost with full houses to better full houses many times, often in situations where neither full house was obvious. Why would a poker site do this? Well, it increases the excitement—some people play for the rush of action, and will play more at a site that has more action (even if they keep losing). It also increases the perceived bad beats—if I lose a lot of money on a good hand, I might want to keep playing to try to get that money back. After all, I “deserved” to win that big hand. Action hands also lead to bigger pots, which means a bigger rake for the poker site. The more action hands, the more money the site takes in (at least in the short term).

    But here’s the thing: there are many, many, many differences between online and live play. For example, at least at the low stakes most people play online, you’re going to get a lot more callers. More callers means more ugly hands catching on ugly flops (and more medium hands catching on medium flops). More hands catching means more action. Another example, which has been discussed plenty of times: you play more hands per hour online. More hands means more action hands (just in actual number, not in terms of action hands per total hands). Another example: seeing the river (and the result of hands). I think you see way more folding before the end of a hand live. This is related to what I alluded to above—I think you see more conservative play in live games, meaning more folding before the end of the hand. This gives players fewer opportunities to catch a one-outer or runner runner straights, etc.

    Anyway, I don’t intend to go into too much depth here—there are plenty of other differences between online and live play. I just want to point out that you can’t just compare the number of action hands you see online to the number you see live and say that the disparity means poker sites are rigging their algorithms toward more action.

    And (as was mentioned on the show), I think it would be short-sighted and stupid for a site to risk getting caught (or even just being suspected) of rigging its shuffling algorithm. As soon as that perception is out there, a site’s players will flee for other sites (which sites will be sure to advertise about their legitimate algorithms). Action works great for games that play to people’s pure gambling rush—slots, roulette, lotteries—but for players who consider themselves to be rational and calculating, a poker site that messes with the odds is a joke (and a good place to get ripped off). They won’t put up with it. So the risk of being unmasked as a flawed poker site is too great for anybody with anything to lose to actually do it. At least, I think so.

My suspicion is that any legitimate poker site is extremely concerned with the integrity of its shuffling algorithm (whether it’s continuous or not)—and that any appearance of a higher rate of action hands online is due to the myriad differences between the two media, not due to anything fishy in the software.

The question of how difficult it might be to crack the system, either through the shuffling algorithm or another way, is completely separate (and, to me, much more concerning). However honest and diligent these sites may be, no system is perfect. If the stakes are high enough, sooner or later someone will probably figure out a way to gain an edge. Poker bots, collusion, algorithm-cracking: these are all serious problems with online play, and I am sure that they are all out there, to some degree or another.

So although I trust the good intentions and fair design of these poker sites, I still think the rule for online play should be: play at your own risk, and never put money on the table you can’t afford to lose.*

* Actually, this is probably a good rule for poker anywhere.