Can you beat this mind-reading machine invented in 1953?

[frauenfelder]

Originally published at: Can you beat this mind-reading machine invented in 1953? | Boing Boing

…

[generic_name]

@frauenfelder

“You bet. I’m a machine and you’re a machine, and we both think, don’t we?”

“uhhhh. . . I think so. . . ?”

[jlargentaye]

Claude Shannon isn’t appreciated enough for being the real founder of our information age.

This reminds me of a story of a professor asking his class to generate a random sequence of heads or tails, either by actually flipping a coin or just generating the sequence themselves, betting them he’d be able to tell the coin- vs human-based ones, and being right at an extremely high percentage.

Turned out humans are pretty bad at being random, tending to skew towards an overall equilibrium. In particular, the professor was looking for sequences of 5 (or something) identical characters, which would likely be produced by a truly random algorithm but humans would avoid.

Edit: indirect citation

[Wetterschneider]

Super sus.

This is using the “AI” against a random coin flip. If this is true, then this guy should be in Vegas, playing roulette, using this AI. He’d win huge.

[Tamsin_Bailey]

Pitiful machine.

[Mister44]

I was able to beat it 2 out of 3 times. I tried to turn my brain of and click as randomly as I could, being jerky with my motions so not every click even registered.

Neat, though! I imagine if i kept playing I’d get closer to 50/50.

[libelle]

It’s no match for my defective randomization.

[euansmith]

Those are rookie numbers.

[Tamsin_Bailey]

Nicely done. That’ll learn 'em.

[generic_name]

Similar experience here.

I had a hard time beating it at first, but I noticed if I just clicked L/R on my keyboard and paid no attention to the red/blue scoreboard it was easy to win.

So possibly looking at the score effects how you click and plays into the algorithm’s favor.

[mspie500]

Same score

Though I did lose round 1

[Tamsin_Bailey]

Me too, but I learned fast

[mspie500]

I went with the “move the cursor back and forth and randomly click strategy”.

[Flanker7]

Third time’s the charm, and I’m ok with that

[KathyPartdeux]

Looks like the poor machine just isn’t prepared for the kind of randos we have here.

[NukeML]

I used the digits of pi. If odd, select left; if even, select right.

[blackanvil]

Well, with the score 3 to 8 in my favor, I’m pretty sure that it could be beat – once you know it’s looking for patterns, it seems to be easily fooled.

[Tamsin_Bailey]

“Unbreakable” just sounds like a challenge to a Mutant.

[gatto]

@blackanvil

once you know it’s looking for patterns, it seems to be easily fooled.

yeah, i wonder if the original game – not having that context around it – would have been harder to beat. if you know it’s looking for a pattern, you can avoid patterns. if you don’t… well, we’re only human