I Hope Phil Gordon Wins the World Series of Poker

The main event of the World Series of Poker gets underway today at the Rio in Las Vegas. Why do I want Phil Gordon to win?

It’s not just because he’s such a nice guy, or because he’s so smart, or because of his philanthropic endeavors, or even because he’s so tall.

It has to do with the game of Rock, Paper, Scissors, a.k.a. Rochambeau. As Levitt blogged a few days ago, he was in Vegas this past week to do research with a bunch of world-class poker players. Part of that research included a 64-player Rochambeau tournament that Phil Gordon organized, and which Annie Duke won.

I flew into Vegas just for a couple of days, mostly to see the Rochambeau tournament and to see what might be worth writing about in SuperFreakonomics. I also got to spend time with Gordon and a few of his Full Tilt Poker pals, including Rafe Furst, another terrific and very bright guy who just won his first WSOP bracelet.

One night, Full Tilt threw a big party at Pure, the sleek nightclub at Caesars Palace. It was big and noisy and really fun, and I had a long and interesting conversation with Phil Gordon about a number of things. In the end, talk turned to Rochambeau. Words were exchanged and suddenly there was a challenge — me against Gordon, head-to-head in Rochambeau, best of nine throws for $100. I think I was the one who challenged him to play; he was the one who suggested we bet the $100.

Levitt held the money. Then Gordon, who is about 8 inches taller than anyone I know, leans over into my face and says, “I’m starting with Rock.”

And he did. I threw Scissors, so he beat me. Score: 1-0.

But I had something up my sleeve. I started the match throwing a Seamstress — i.e., a three-throw gambit of Scissors, then Scissors, then another Scissors. Gordon, after his initial Rock, threw a Paper, then another Paper. I was up 2-1.

Finally, on the fourth throw, Gordon threw a Scissors. But I had thrown my fourth Scissors in a row, which meant we tied on that throw, leaving the score at 2-1. That’s when Gordon leaned into my face again and said, “You do know that you can throw something besides Scissors, right?”

But my four consecutive Scissors throws (let’s call it a SuperSeamstress) had plainly shaken him. He recovered to tie it up at 2-2, and then took the lead briefly at 3-2, but I tied him, then went up 4-3. He managed to tie me at 4-4 but, never in doubt, I threw one more Scissors and beat him, 5-4. He looked pretty stunned. Poor guy. It turned out that he really hate to throw Scissors.

So why do I want him to win the WSOP? Not because I feel sorry for beating him. Now more than ever, I believe that Rochambeau is a game of luck, and I happened to get lucky against a guy who is a really good poker player.

No, the reason I want Gordon to win is simply so I can tell my grandchildren someday that I beat the WSOP champ at something, even if it was something as meaningless as Rock, Paper, Scissors.


leoboiko

I had no idea janken-pon had a French name.

gawks

"Now more than ever, I believe that Rochambeau is a game of luck, and I happened to get lucky against a guy who is a really good poker player."

My roshambo strategy has always been to take out a random dollar bill, and use the serial number to decide what I throw. If the first number is between 1-3 inclusive throw a rock; 4-6, throw paper; 7-9 throw paper, and ignore zeros. If it's best of 5, use the first 5 eligible digits.

I can prove that no other roshambo player can adopt a strategy that would beat me more than 50% of the time in the long run, so I guess that makes me the best roshambo player around...

autiger

I bet I could be you 100% of the time. If you used that exact formula to play me I could win best out of five every time. All that I would have to do is use paper everytime. Unless you were being careless and meant to say 7-9 throw scissors, because then I'd be back at square one.

byomtov

I guess that makes me the best roshambo player around…

Or tied for it.

RC of strangeculture

is it wierd or make any sense that to me, I've always known this rock-paper-scissors game as being called "Farkling?"

--RC of strangeculture.blogspot.com

ggrot

Rock, Paper, Scissors is only random if at least one player tries to use a semi-random strategy like gawks mentions. If both players try to outpredict their opponent, then it is no longer a random game.

An interesting example is the RoShamBo Programming Contest: http://www.cs.ualberta.ca/~darse/rsbpc2.html which pits computer algorithms against each other playing rock,paper,scissors for thousands of rounds. It is tournament style and randomly seeded into the tournament are algorithms guaranteed to use some type of non-random strategy. If your algorithm uses a random strategy, you will fare no better against those seeded algorithms than 50/50 odds, but algorithms that do try to predict the next move can indeed beat the seeded algorithms with better than random chance and hence will win the tournament. This slight twist makes a random strategy a poor one.

Mango

"I can prove that no other roshambo player can adopt a strategy that would beat me more than 50% of the time in the long run"

What about the strategy where I steal your wallet and figure out which dollar bill you used?

Mango

In seriousness, I bet I can beat the dollar bill strategy. If you throw rock on 1-3, paper on 4-6, and scissors on 7-9, then I'll just throw paper every single time. I think this wins more than 50% of games.

Since currency serial numbers are printed sequentially starting from 00000001 up to 99999999, then it makes sense that at any given time there will be more bills starting with 1, 2 or 3 than 7, 8 or 9. Ditto with the other numbers, though the difference shrinks considerably with each successive digit.

711buddha

You are missing the point.

Its not an accident that this occurs at the WSOP. If you play poker long enough everyone gets exactly the same cards, yet some players consistently win.

The reason is that they play the other person as well as the cards.

A note on randomizing actions. While Gawks is right that he never gets beat more than 50% of the time, he also never really wins more that either. I would say he's the best at not losing, but that's not winning.

I poker we call this type of player a "rock." They are very conservative, play few hands and seek to win their fair share of pots while avoiding donating to everyone else's. It's far from optimal strategy, as these players are easy to identify, and don't get much action when they do get a hand. (usual caveats about the game apply here...)

Back to roshambot, if your strategy is only to randomize, then we may split every time we play, but in the long run with many opponents, I win because everytime we both play a three year old who only throws rock, I'm going to win and you'll split.

Read more...

Papagorgio

I think most of the above misses the point of the RSP tournament. If every player plays the randomisation strategy every players expected win ratio is 1/2- they'll all fare the same (expectation wise anyway).

However, in a 64 player tournament, if even one other player is playing anything other than a pure randomisation strategy, then if you can beat it you can have the best chance of winning the tournament.

He has a 50/50 shot against everyone but you (if you have a superior tailored strategy to beat his) and so will come last. You have a 50/50 shot against everyone but him, against who you are expected to win! You're expected to be the champ.

This is why there still exist annual RSP tournaments for programmers- who have to build/code programs which can beat opponent programs over tens of thousands of games.

Interesting.

Ken D.

It sounds to me like being good at competitive r/p/s forces you to be a hustler. You have to be smarter than the other guy, but you have to try to make him think you might be dumber. If your smarts show, the opponent can neutralize you with a competent randomization, rather than making guesses that you may be able to outguess. This is in some ways quite similar to poker. A key difference is that the knowledge and technique to be a good "rock" in poker takes a fair amount of learning, but it r/p/s it is trivial.

autiger

I understand that a computer may be able to figure out a pattern, and the algorithm that is being used by the opposing computer, but is a person capable of figuring out an algorithm of one player who the only play best out of five times, of course assuming that a player isn't playing randomly, but subconciously does have some sort of formula that they use to play. Using a dollar bill is random, and the way most people is simply to try and guess what the other person is thinking, when they really probably have no idea.

Matti Leshem

Sir,

I know Phil Gordon and you are no Phil Gordon. While you may have beat him (I have as well) at the end of the day Phi Grdon is a superlative Rock Paper Scissors players. While he is an excellent poker player (and I too hope he wins)he is an even better RPS player. As the commissioner of the USA Rock Paper Scissors League (www.usarps.com) I have witnessed and played all manner of players, both amateurs and pros. His expertise and acumen is unparalleled in the sport. He has a zen quality to his game and over the course of several games you never would have beat him. Still it is a victory to be savoured.

Matti Leshem
Commissioner USARPS League

cp

Wow...

Please tell us how you can beat an opponent who acts randomly.

leoboiko

I had no idea janken-pon had a French name.

gawks

"Now more than ever, I believe that Rochambeau is a game of luck, and I happened to get lucky against a guy who is a really good poker player."

My roshambo strategy has always been to take out a random dollar bill, and use the serial number to decide what I throw. If the first number is between 1-3 inclusive throw a rock; 4-6, throw paper; 7-9 throw paper, and ignore zeros. If it's best of 5, use the first 5 eligible digits.

I can prove that no other roshambo player can adopt a strategy that would beat me more than 50% of the time in the long run, so I guess that makes me the best roshambo player around...

autiger

I bet I could be you 100% of the time. If you used that exact formula to play me I could win best out of five every time. All that I would have to do is use paper everytime. Unless you were being careless and meant to say 7-9 throw scissors, because then I'd be back at square one.

byomtov

I guess that makes me the best roshambo player around...

Or tied for it.

RC of strangeculture

is it wierd or make any sense that to me, I've always known this rock-paper-scissors game as being called "Farkling?"

--RC of strangeculture.blogspot.com

ggrot

Rock, Paper, Scissors is only random if at least one player tries to use a semi-random strategy like gawks mentions. If both players try to outpredict their opponent, then it is no longer a random game.

An interesting example is the RoShamBo Programming Contest: http://www.cs.ualberta.ca/~darse/rsbpc2.html which pits computer algorithms against each other playing rock,paper,scissors for thousands of rounds. It is tournament style and randomly seeded into the tournament are algorithms guaranteed to use some type of non-random strategy. If your algorithm uses a random strategy, you will fare no better against those seeded algorithms than 50/50 odds, but algorithms that do try to predict the next move can indeed beat the seeded algorithms with better than random chance and hence will win the tournament. This slight twist makes a random strategy a poor one.