Terminator hand will destroy you (in Rock Paper Scissors)

Forget beating Watson on Jeopardy. I just got dusted by The Terminator in Rock Paper Scissors. Skynet is nigh!

Andrea James is a writer, director, producer and activist based in Los Angeles. Her work often focuses on consumer activism, the free culture movement, exogenous mysticism, humor, and LGBT rights.

MORE: Games • guestblog

More at Boing Boing

Eurovision 2013: An American in London

The technology that links taxonomy and Star Trek

planettom

Only by switching to Rock Paper Scissors Spock Lizard can we keep ahead of the robots! http://www.samkass.com/theories/RPSSL.html
mraverage

i used the (pseudo) random number generator on my hp-15c and got 36/26/38
tuckertuck

There is a slight fault in the Veteran mode. Since it always draws on experience from over 200 000 games you can keep playing the same set of four hands over and over and keep beating it. You can’t do that on novice mode because it would just learn from you and beat you 100%. I just played 100 games on Veteran and have beat the computer 64-32-4. It only won the first four games and then it got caught in a loop. I assume it would learn if a lot of people started playing the same four hands over and over. I Played Rock, Rock, Rock, Scissors, Repeat…

Interesting I just started the game again with the exact same strategy and it appears to have already learned.
Monkeybaister

http://www.youtube.com/watch?v=NMxzU6hxrNA
spocko

Of course I expect that people who win will come here to brag. Why tell everyone you aren’t smarter than an algorithm? But I will.It beat me.
On one hand I’m embarrassed on the other hand, good for the developers! Let’s acknowledge their brains.

Then I used a “random” number generator on the computer and used it’s results and I beat it. (I used one computer algorithm to beat another, I’m still not smarter than the algorithm.)

What I found most interesting is what the computer was “thinking” based on looking at past throws. That was really interesting. Once you understand how it works it’s not so mysterious. In general I’m in awe of people who combined the various elements of math, stats and computer programing to make this. Well done!
tuckertuck

Just tried it 3 more times and it still works. Just can’t get it down to less than 4 victories…
“Good ol’ ‘rock’. Nuthin’ beats that!”
tuckertuck

68-32-0 just had to get by the first 5 games using Rock and Scissors and then after that repeat, Rock, Rock, Rock, Scissors,
Sparky

Won-15 Tie-16 Lost-18 Playing against veteran.
Avram / Moderator

I’m actually pretty decent at Rock, Paper Scissors; when I played Cosmic Coasters (which uses RPS as a resolution mechanism in its combat system), I was able to frustrate the friend I was playing against by getting inside his head and semi-consciously predicting his moves.

Against the robot player just now? In 7 rounds (in Veteran mode), I won 3, lost 1, and we tied 3 times, but then I had to stop because the lack of a human player to second-guess was freaking me out. I was surprised at how stressful it was.
huntsu

Out of 33 games, I won 13, tied 9 and lost 10 in the Veteran mode. Not bad!
Kaiguy

One strategy that seems to mess up it’s data crunching is to pick one of the three options (say, rock) and never throw it. Ever. Limit yourself to Paper and Scissors. At round 50 I have a score of 22-17-11. While a human would realize, and adapt to never throw paper, forcing me to change strategies, the way the program focuses on it’s entire database means it keeps throwing paper, and getting beat or tied.

Update – Round 100: 41-30-29.
Anonymous

now that’s a funny place to put the third digit.

.~.
nox

11/4/5

This is less AI and more a measure of how humans tend to play this game. Apparently I’m not very human. :(
Anonymous

Not so impressed. I tried a fancy theory and got 7-7-7. The wife (the buddy gamer) used her usual be-obstreperous-try-to-do-what-no-one expects like she uses in games, and she also got 7-7-7.

Both on veteran.
galois

When I played naturally–just clicking quickly on whichever play I wanted, the computer was able to predict my moves decently. When I consciously made an effort to play “strangely” and change my strategy after a few moves, I was able to keep ahead of the computer.
GIFtheory

If the interactive graphic is right, it would appear to be using an n-gram model. NLP 101.
thelonmon

I, for one, welcome our new rocks-paper-scissors overlords.
Anonymous

I played 100 rounds obeying the python random number generator, and lost: 28-36-36
ill lich

That’s nothing, I repeatedly got beat at tic-tac-toe by a chicken.
Anonymous

I got 67% by playing predictively, no system. “What would it expect based on the last two results?” and playing what would beat it, the way you might against a human. Going an extra move ahead backfires, so I guess most people don’t go deeper than one counter-move.
Anonymous

I was able to beat it the first 100 rolls: 38, 34, 28. It started to figure out my zen strategy at around 170. Around 220, it passed me and got up to 10 up on me. Then I switched to a d6 + intuition. I beat it by only a couple around 500 throws.
Enoch_Root

Winning strategy = have your wife pick naturally then play based on the assumption that the computer plays based on that
Anonymous

Hmm, I played fifty round using a die sitting on my desk as a random number generator and won, 13-17-10.
- mraverage
  
  come on anon; what happened to the other 10. you’re holding out on us.