watch out for this hustler

Water Cooler
#1

http://gothamist.com/2014/09/23/rock_paper_scissors_hustler.php

rock paper scissors…lose

#2

That is awesome

#3

I think the AF’ers in NYC should seek him out and see if he can overcome the superior performance guaranteed by the charter.

#4

There used to be a professional rock, paper, scissors league. www.usarps.com. I know they used to have sanctioned events in Vegas. Lol

#5

This guy quit a job at Apple to make shapes with his hands in dive bars.

#6

Yeah, but he was probably a “genius” working in an Apple Store. Rock, Paper, Scissors is more prestigious.

#7

He brings in as much as $300/night. Only Uber drivers do better than that.

#8

$300/night, 4 days a week is a little more than $60k a year without taxes. Not shabby, but how much of it does he spend on alcohol.

#9

So more like an all time best rather than a consistent figure. More power to him, though.

#10

But why doesn’t he just go sell cars or something?

#11

The dominant strategy for each player that results in a Nash equilibrium for RPS is to randomly decide if you’ll go for a rock, paper or scissors on each turn. Analytically, two things are left to be exploited in a real-world scenario:

  1. If you follow the theoretical dominant strategy of making random selections, you will not lose in the long run, but you will also not win even if the opposing player’s strategy is completely predictable. Therefore, there is an incentive to deviate from the dominant strategy and search for patterns in the opposing player’s game. That of course leaves you exposed to a counter strategy where the opposing player is looking for patterns in your game.

To find a pattern in the opposing player’s game with some confidence level that will make it reasonable for you to exploit it, you need a rather long sequence of consecutive games with the same player. For instance, a simple pattern could be that you suspect he is more likely to play rock than any of the other two options. You can start wih a prior assumption of beta(30,60) for the probability of playing rock - peaked distribution centered around 1/3 with stadard deviation of +/- 5%. After G games, you observe he played R rocks, so your posterior becomes beta(30+R, 60+G-R). The expected value is no longer 1/3, but it is (30+R)/(90+G). Is this significantly different from my prior assumption that the guy plays rock with probability 1/3? That depends on what “significantly different” means to you, but let’s say you want to see a result of 43% (1/3+10%) to be sure. To get there, if you’ve only played 30 games (G=30), you’ll need to see at least 22 rocks (R=22) - seems pretty unreasonable. Your opponent must be drunk or have a strong bias for rocks to engage in such predictable pattern. But if you played 300 games, you only need to see at least 139 rocks - more subtle. You’d likely get bored after 50 games though.

  1. Humans are remarkably inefficient in generating truly random sequences. You’d probably be uncapable of following the dominant strategy in the game even if you tried to do so. Your selections for rock, paper, scissors will deviate from truly random ones in the long run even if you think you are totally random with your choices.

For example, how likely is it to see a consecutive stretch of “no rock” if you have played 10,000 games? The probability of no rock is 1-1/3=2/3 in a single game that is decided in a truly random fashion. The expected value for the “longest run” for an event with probability p for a sequence of n trials is LN(n*(1-p))/LN§. So the expected stretch of “no rock” in 10,000 games would be 20 (±6), between 14 and 26 with 95% confidence. Most people only have a short memory and would perceive a stretch of 20 consecutive games with no rock as non-random, so they’ll switch more often. You can exploit that by anticipating shorter stretches from a human player than what would be normal in a truly random sequence, but you need to find a really patient player for an opponent.

I call BS on this guy.

#12

^whoa

#13

^^ Search him out in a NYC bar and prove him wrong.

#14

Mob Strip, your new sig;

8=============================D

#15

thanks homie, I ain’t getting played by some halfway RPS crook… he’s just a shook one, as quantitative analysis suggests.

#16

^ Respect.

#17

Awesome