My recent post Manipulating The Deck In Online Poker seemed to “stir up the pot” a bit as Bill Rini put it. I thought a lot of the feedback was interesting, and I responded to some of it in the comments section of the original post.
But I thought Bill’s feedback needed a new post to properly address. If you haven’t been following the discussion, definitely check out Bill’s post Online Poker Isn’t Rigged . . . Again!.
Bill makes essentially two points. First, he doesn’t think the risk/reward ratio is there to make it worth a big site’s while to try to rig their deck, and second he thinks actually rigging the deck is a lot harder than I made it sound.
Why Rig A Deal?
So, if we’re talking about a top ten room then I would say that the motivation for rigging the game favors not rigging it. As I articulate in a previous post there are so many other ways for a poker room to increase its profits from you that are entirely legitimate. And if you took the time to brainstorm a bit I’m sure you could come up with ten or fifteen more suggestions of minor tweaks the room could make to the game that would generate more hands per hour and/or more profits for the room. Until someone can answer for me why a room would go to all of the trouble to rig the game before having exhausted these other much more simple methods then I simply cannot buy this argument as being logical. Granted, a poker room might act in an illogical manner but if we’re to assume that all actors act in a logical manner then this doesn’t hold up.
I do sort of dismiss this argument, and I do so for two reasons. First, Bill poses a false either/or here. He suggests the cardroom could increase profits either in legitimate ways or by rigging the deck, and he prefers the legitimate ways. But obviously cardrooms could try both at the same time. There’s no reason for me to think that a cardroom would order these options in a “try all legitimate options then try rigging” manner.
Second, Bill says, “if we’re to assume that all actors act in a logical manner…” which I always think is a poor assumption. Most people aren’t strictly logical thinkers, and online cardrooms have, in my opinion, demonstrated many times over that they are run by people who often don’t make entirely logical decisions. And, beyond that, even “logical” people can be prone to making rash decisions when doom and gloom appears to be forthcoming. I’m reiterate the scenario I posed in the comments of the previous post. A major cardroom does an analysis of its traffic and concludes that even though it appears healthy to the casual observer, it’s actually in danger of losing a critical mass of customers within two or three years time. I could easily see a manager panicking and making an “illogical” decision under these circumstances.
Before I move on to the meat of this post, I want to reiterate that I’m not accusing any site of being rigged, and the vast majority of claims of rigging can be dismissed nearly on face. My whole point in this series of posts is to encourage people not to stick their heads in the sand. A site’s deal could be altered now and not yet have been discovered, or at some point in the future an honest site could turn dishonest. It’s possible, and the player community should be as vigilant as reasonably possible. That’s my argument.
How To Rig A Deal
Moving on. Bill argues that rigging the game to the benefit of a cardroom is neither easy to accomplish nor easy to conceal:
One of the other factors such an argument fails to properly consider is that over 90% of poker players are break-even or losing poker players. So what exactly is a fish? How are you going to rig the game in favor of the fish when there are so many fish and so few sharks? How would you determine which players to rig the action for and which one’s to shaft?
In my previous post, I mentioned that cardrooms could alter the game to improve their profitability. It’s been suggested by others that by rigging hands to favor “the fish” one could do so. That’s certainly a reasonable place to start, but I’m not arguing that helping out the fish means more money for a cardroom. That argument comes from other people, not from me.
My only point is that there likely exists some change one could make to the deal that would improve cardroom profitability. If you don’t grant that point, then you’re claiming that no-limit hold’em as it stands is essentially the perfect game for a cardroom to spread.
In any event, I think Bill is chasing his tail a bit here when he talks about how it’s hard to identify the “fish” from the “sharks”. I don’t think that it’s necessary to solve that problem at all. The theoretically dishonest cardroom cares about altering the game to make more profit, not necessarily helping one set of players against another.
Having said that, I do concede that if a cardroom were to explore these options, they might start by considering ways to blunt slightly the advantage a skilled player has. Here are possible alterations to the deal that might accomplish this goal without being too complicated:
- The cardroom could skew all-in confrontations to slightly boost the chances that the underdog hand wins. For instance, once two players are all-in, it could quickly calculate the equities of each hand and then deal cards such that the underdog hand consistently wins slightly more often than it “should”. Perhaps it could turn all 80-20 confrontations into 78-22 affairs. I mention this change first because it’s a specific one I’ve seen posited before.
- If the proposal above is too blunt an instrument (and possibly too detectable), then the cardroom could slightly shade the odds in favor of the player who called all-in at the expense of the player who bet all-in. So if I semi-bluff all-in and you call me, I would have a slightly worse chance than I “should” to win the hand.
- If you’re worried about detection, all-in hands are possibly not the best ones to alter since hand histories report the results of them with nearly complete information. So perhaps you alter the deal in non-all-in situations to slightly favor the player who called a bet over the player who made the bet. This would have the effect of making plays like firing turn barrels less effective since the calling player would “get there” more often than they should between the flop and turn.
Again, I’m not sure what game alterations would benefit the cardroom and what alterations wouldn’t. But I think it’s clear that at least some alterations exist that would benefit the cardroom, and I think a dishonest cardroom could use trial and error to try to find some that work to their benefit.
Also, note that these are all simple, systematic alterations to the game. They don’t require identifying “fish” and “sharks”, and they don’t require rigging hands individually according to some complex algorithm.
How To Hide The Evidence
The obvious criticism of this sort of alteration is that it would be detectable. Build a database of a few gazillion hands, run the right tests on it, and viola, you’re a newly minted whistleblower. Unfortunately, I don’t think it’s that simple, and part of the reason is that the burden of proof currently lies almost entirely on the accuser and not at all on the cardroom.
Plausible deniability. It’s a powerful weapon. Someone posts an article saying, “I think the deal at cardroom XYZ is rigged subtly in this peculiar way that benefits them. Here’s my data.”
Cardroom has the following response:
- We emphatically deny it. Why would we rig a deal when we already make so much money?
- That effect you observed is so small it’s probably just luck.
- If you go digging through the data enough, you’ll always find something that looks out of the ordinary.
- Prominent players A, B, and C have won a ton of money on our site. Clearly that makes the deal fair.
- Our code is audited by 3rd, 4th, and 5th parties. By the way, you’re just some jerk on 2+2.
Yada yada. Standard stonewalling tactics. In the meantime, the cardroom just backs out the offending code and the effect disappears. Just a weird run of cards, guys… nothing to see here.
In fact, the cardroom would likely get early warning that someone was on to them by reading 2+2. It’s not like the AP/UB thing were there were a few offending screen names that could be tracked. We’re talking about a systematic effect that would be reasonably subtle and where no one could easily point to a timestamp on, “See, as soon as we started talking about it the offending parties fled to a desert island.”
In other words, as long as they were sufficiently subtle about it, it would be very hard to detect in the first place (especially because no one that I know that has any skill at this sort of thing is really looking for it), and if detected, there’s simply no smoking gun. Just hire a PR firm and blast away until the controversy goes away.
Hiding The Evidence In An Even Sneakier Way
If you wanted to alter the game more than a few percentage points and still not get caught, you could probably do so nearly undetectably by piggybacking on a well-known effect that’s not well-studied. For instance, card removal.
I saw some post where someone noted that small cards were more likely to come on the river than big cards. A deuce is the most likely river card, an ace is the least likely, and the other cards are all distributed uniformly in between. This effect makes some sense from a card removal perspective. Two big hands might tend to see more rivers than one big hand and one small hand or two small hands. Big hands are more often made of big cards, and therefore the deck should be slightly rich in small cards by the river.
But how big should this effect be? Who knows? Is a 0.01% difference in frequency between a deuce and an ace reasonable? How about a 0.1% or a 1% or even a 5% difference? Sure, you could take a whack at it with simulation, but it’s hard to simulate the way people actually play… and how people actually play is important when we’re talking about card removal effects.
So say this difference “should” be 1%. The cardroom decides they want to systematically shift results to favor one hand type over another… and they look for opportunities to replace big cards on the river with small cards to make that happen. They replace the cards until the small card big card difference is 3% or 4%.
Anyone who went looking would notice that the all-in results didn’t line up with the theoretical results. But then when they attempted to account for card removal, they’d note that small cards came more frequently than expected on the river… an effect that piggybacks on card removal. Time for alarm bells? Or maybe card removal is just a little stronger an effect than we previously thought. Who knows?
None of this is too complicated to pull off. I’m going to alter my deal to favor underdogs in all-in situations. I’m going to carry that out by occasionally substituting a high card that would have come on the river with a small card that helps the underdog. I’m going to do it with a frequency such that it’s hard to detect unless:
- You’re looking specifically for it
- You have a huge sample of data
- You know exactly how much of an effect card removal should have and you can back that out of the data and still detect the anomaly
And then, after you detect it, you have to win the PR war to convince the world that this obscure effect you’ve found is real and you’re not just another “OMG IT’S RIGGED!!!!” kook. And you have to keep people convinced even if the effect subsequently seems to disappear or reverse.
If I’m a cardroom owner with plenty of greed, not a lot of scruples, and a huge ego, I might give it a shot…. even if I ran the biggest, most profitable cardroom in the world. I might be especially inclined to try it if I thought for some reason that my cardroom was on the way out and I needed to act to save it.
Finally, I feel compelled to reiterate some main points from both articles. I have no evidence or reason to believe that any current site has an altered deal. I don’t know how to rig a site to improve its profitability; I merely posit that it’s possible to do so. If a site were rigged in the ways I’m proposing, it would be entirely undetectable to the naked eye. In other words, “I’ve been noticing my good hands getting beaten a lot recently,” is not an observation that would lead me in any way to conclude that anything was wrong with the deal.
But I do think it’s possible. It’s possible an honest site will turn dishonest in the future, and it’s possible that a site we all think is honest today isn’t. It doesn’t mean a good player couldn’t still beat the game consistently or that any one player would get singled out for the doomswitch. But it’s possible that all is not quite as it seems in online poker.
Tags: altered deals, bill rini, card removal, manipulating the deck, online poker rigged, online-poker, poker, rng