> In a class I taught at Berkeley, I did an experiment where I wrote a simple little program that would let people type either “f” or “d” and would predict which key they were going to push next. It’s actually very easy to write a program that will make the right prediction about 70% of the time. Most people don’t really know how to type randomly. They’ll have too many alternations and so on. There will be all sorts of patterns, so you just have to build some sort of probabilistic model. Even a very crude one will do well. I couldn’t even beat my own program, knowing exactly how it worked. I challenged people to try this and the program was getting between 70% and 80% prediction rates. Then, we found one student that the program predicted exactly 50% of the time. We asked him what his secret was and he responded that he “just used his free will.”
I used the digits of pi, transposing and changing the order, and got 14% on the original game. Professional poker players will use external sources of randomness to make sure they’re perfectly balancing their positions and not exploitable.
There's another version where a stats professor divides people into two groups. One group flips a coin 50 times and writes on a whiteboard the results. The other comes up with 50 results.
They leave the room and when they come back immediately know which is which. Specifically because of the over-alternation that yours mentioned. A string of HHHHHH is exactly as likely as HTHTHT, but people don't intuit that.
The strings are equally likely if they're being purely randomly selected, but not if they're being selected by people with a poor understanding of random. In the latter case, people are more likely to select HTHTHT.
I wonder what thought process led you to believe that even though the strings are equally likely and random, people who don't choose the one you think is more random have somehow inferior understanding of randomness
The issue the anecdote is reflecting is that across the data of 50 naive made-up coin tosses there are too many alternating sequences and probably zero HHHHHH.
I understand your point, but the context is that repeatedly, experiments have shown that humans inject too much pattern to be good rngs. It turns out that is visible with training after the fact.
Yes, it is possible that any two 50-sequence patterns are actual rng. But some of them are much more likely, and we can tell.
I have managed to get 38-48% the times I tried. I believe I might have been influenced by the article I just read.
It would be interesting to see if by giving people advice on statics it would produce, on average, a measurably different outcome.
Similar to me, when I have to choose randomly I look for a number anywhere around. If it's odd, one choice, even is the other.
My strategy for rock paper scissors is to always go rock, and tell everyone that's my strategy. It's more fun for me seeing how people deal with that information than the win would be. Not very random though
I don't think Zipf's law gives you this. It mostly says that common words tend to be short, but it doesn't tell you what proportion of the text will be common words. For example, in English writing two-letter words are found much more than one-letter words.
However, I think you're correct anyway. The right distribution to look at is something like the orange series in the bottom graph on [0]. Eyeballing it, it does look like 3/5/7/9/11-letter words appear more in aggregate than those of 2/4/6/8/10 (49% Vs 44.5%), and longer words are a rounding error.
> (...) Then, we found one student that the program predicted exactly 50% of the time. We asked him what his secret was and he responded that he “just used his free will.”
- Scott Aaronson, Quantum Computing Since Democritus.
This isn't very interesting, but it scratched an itch I had to see what would happen.
The house's record
The house called 18 of your 190 scored numbers: 9%.
Blind guessing would average about 21.1 of them (11%).
Only 1 of the 206 people in a psychology study was more random than you. The house found you extremely hard to read.
I really want to know who was more random than a RNG
It might not be that someone is more random than an RNG, but that the distribution of numbers in your 190 vs. theirs appears more "random". It's a limited sample size.
If yours happened to contain consecutive digits or common ones (ending in zero, significant numbers like 911 or 42 or 69), that would imply less randomness even if it's actually the result of "real" RNG.
Getting 0% isn't the goal though. You want to sit right at 11.11%, the blind guessing rate. If the prediction algo always predicts you wrong, it actually always "predicts" a number you won't guess. That too is a feat, reducing the pool of numbers you're going to choose from, therefore making you more predictable.
It looks like it's a fair implementation of guessing. Giving it digits of pi gives a very high (low) score - most times. To get a lower score I think you would have to be adversarial and/or get a lucky run. Based on the house doing better/worse on some runs I guess it not deterministic.
The house called 8 of your 76 scored numbers: 11%.
Blind guessing would average about 8.4 of them (11%).
Only 4 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
I was. But if you're more random then you're probably not random. I was actively aiming to exploit the game - try to figure out which pattern detector it's keyed on based on its guesses, and then swap to a different pattern. Ended up getting 6% over a decent sample. You could probably get zero with a bit of luck!
The problem I had was somehow the double-pick algo randomly clicking on at seemingly illogical times. I'm guessing that maybe it's something like if you're not hitting any pattern matcher then the next guess uses a random pattern guesser, because it nailed me multiple times with the doubles even after not having one in 10-20 numbers prior. Though maybe that is itself a pattern that it checks for? It wasn't mentioned in the docs.
Does someone know whether people saw the "predictions" in the actual test? Because saying random numbers blindly is not what I was doing. After my next number was predicted correctly, I changed my "strategy" to increase randomness (be less predictable).
So I don't know if this result is comparable to the one from the study.
My results were:
The house called 10 of your 87 scored numbers: 11%.
Blind guessing would average about 9.7 of them (11%).
Only 5 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
You’re right. If you notice the instructions in the game, it encourages you to go as fast as you can. I keep the numbers up there just to make sure the voice recognition is working properly and to make the game seem more engaging.
But yeah, I agree: I think this does influence the results, since it makes it easier to be more random because you can essentially try to play in an adversarial fashion by adjusting your results according to feedback from the predictor itself.
In the original study, that wasn’t the case. They had a three-by-three grid of numbers and a metronome as well, which meant that they couldn’t slow down or speed up and had to go at a very specific rate with zero feedback until the very end.
Maybe I should add the equivalent of a "Hard" mode which makes the game more faithful to the original study.
I only have a few digits of pi memorized, so I had better luck using digits from fibonacci's sequence. I can derive them in my head, albeit slowly as the numbers get large.
> The house called 15 of your 113 scored numbers: 13%.
Blind guessing would average about 12.6 of them (11%). The house called 1.2 times as many.
Only 5 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
I did a PhD in a stochastic-related field though, so I know about some “tricks” (in that most humans are too random and eg. dont sample long running sequences)
An article was posted a few months back about a competition for the best rock-paper-scissor program, where submitted programs played each other. Point was to build the program that would be able to predict best the opponent program's moves (which itself is trying to predict you program's moves).
It was a very interesting read, and it explained a really "simple" strategy which I found very memorable: build 3 strings representing "all previous moves + one of the 3 possible next moves", run a text compression on them, pick the move that yielded the smallest compressed result. That's the one where the compression algorithm identified the strongest pattern.
I found it really smart, instead of modelling the game it went one step higher in abstraction and works directly at the level of pure pattern recognition.
That keeps coming back to mind when thinking about things like randomness, the "bitter lesson" and LLMs.
Many years ago, as a side project to help me apply to tech companies, I made this little site [0] where the premise is trying to pick the number that you think other people won't pick very often. I still think it's pretty neat.
Also, in part because of this project, I got to go from being unemployed in my hometown in the South to working in SF on some pretty neat projects. It's a good memory.
Nice~ The website made me recall an old mentalist force trick we learned as kids where you ask somebody to "quickly think of a two-digit odd number between 1 and 50 where each digit is a different number".
Most people will say: 100101 (binary to avoid spoilers)
The pattern of holes is really interesting there. Everyone avoiding nice round numbers, so a bunch of those are very low frequency - except weirdly at the very ends of the range
What about the following strategy? You first think of a random integer n1 from 1 to 9, then you think of another one n2, perhaps one that you have not used very often, and finally say (n1+n2) mod 9 + 1. I think this strategy could make your selection more random . Perhaps someone could try it and report the results.
Imagine that instead of two numbers you select a million numbers, and apply a zip algorithm to that sequence to obtain a file then you map the first byte of that file using (first byte) mod 9 + 1, intuitively there must not exist any correlation between consecutive tries of this strategy. Looking for this example I realized that instead of using n1+n2 you should use a non commutative operation to improve the strategy, but when using just two numbers the addition suffices.
The essence of von Neumann's trick is that all permutations of independent events have the same probability. So the trick generalizes to any other set of possible results.
This is pretty interesting. I've found it's fun to throw off the algorithm. Use your warmup to repeatedly guess the same number, and then don't use it for a while.
That's quite a creative game idea.
The house called 4 out of 40 number picks of mine.
I did not use the speaking mode, so I do assume my randomness was inflated by that.
The house called 9 of your 117 scored numbers: 8%.
Blind guessing would average about 13.0 of them (11%).
None of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
> The house called 1 of your 32 scored numbers: 3%. Blind guessing would average about 3.6 of them (11%). Only 1 of the 206 people in a psychology study was more random than you. The house found you extremely hard to read.
I found it interesting when I felt my focus shift from the number-generating task to noticing (and responding to) the house’s guesses.
As soon as I noticed it repeating its guess, I had a good long run of adversarially avoiding that number… it’s kind of fun being on the machine side of the slot-machine-type racket!
The first time I did badly. But the second time I got really far by deciding on a number instinctively and then picking something next to it. Maybe we instinctively pick similar numbers, but if we fight our instincts we end up with something truly random.
Alternating and advancing from both ends seems to work, it calls n-1, and n+1 correctly if the previous guess was n, but got no calls in a sequence like - 1, 8, 3, 7, 4, 7, 1, 7, 3, 8, etc
I was able to beat this quite easily by just saying the first 100 digits of pi out loud. The house guessed 1/10, no better than random. Maybe you should add a predictor for "well-known mathematical constants".
Yup! Most of the way I put the system together was by having it build several n-gram Markov models of your numerical sequences on the fly.
But we also have a couple of predictors that aren’t markov chains and are based on simpler heuristics. Without seeing your record, I'm guessing that the "numerical avoider" predictor got overweighted and just kept trying to guess the same number.
OP here - I went with Vosk because it's pretty trivial to lock input down to a set vocabulary (in this case numbers) and from my testing works relatively well on most browsers. (webkit, firefox, etc)
this is such a poor model, I've tried it before. Try yapsnap it has better word error rate and runs super fast on CPU. https://github.com/kouhxp/yapsnap/
Thanks for the suggestion - I've been looking yadsnap over but it seems like the ONNX model for english is significantly larger than the one used by vosk (30mb versus 150mb) so that would be my only hesitation to use it.
Maybe there's a way to shrink it down given I only need to support a limited vocabulary domain...
I just pretended I'm playing piano to a megamix of tracks which popped into my head and did 'well'. Only 2 people more random. Typing though, think the site said that's "easier to win" on. Cba to speak 200 numbers out loud :/
Haha, that’s great. Yeah, I totally get what you mean about speaking out loud.
You only really need to vocalize about ~40 numbers total, but you can actually get a similar level of RNG by "subvocalizing the numbers" before you press them on the keypad.
The main issue is that using the keypad kind of lends itself to just pressing it randomly without really thinking about the actual numbers themselves.
"You spread your picks more evenly than a random draw would. Your most-used number made up 23% of your picks, where a random draw's busiest number gets about 23%."
I have a little "mad-lib templatized" set of sentences for reporting on various stats, and that one was supposed to be sort of related to the law of small numbers. For example, shorter runs mean there is a larger probability of getting lopsided distributions (different from pure 11%).
It's supposed to demonstrate that even though you used a single number a disproportionate number of times, it's not necessarily statistically significant since shorter RNG runs can produce the same effects.
This is a nice spin on the Aaronson oracle:
> In a class I taught at Berkeley, I did an experiment where I wrote a simple little program that would let people type either “f” or “d” and would predict which key they were going to push next. It’s actually very easy to write a program that will make the right prediction about 70% of the time. Most people don’t really know how to type randomly. They’ll have too many alternations and so on. There will be all sorts of patterns, so you just have to build some sort of probabilistic model. Even a very crude one will do well. I couldn’t even beat my own program, knowing exactly how it worked. I challenged people to try this and the program was getting between 70% and 80% prediction rates. Then, we found one student that the program predicted exactly 50% of the time. We asked him what his secret was and he responded that he “just used his free will.”
Must have been looking at some external random source
Haha thought the exact same
I used the digits of pi, transposing and changing the order, and got 14% on the original game. Professional poker players will use external sources of randomness to make sure they’re perfectly balancing their positions and not exploitable.
like whether a number is prime should be enough, but for making it even more random I would try inverting the result of each second number.
There's another version where a stats professor divides people into two groups. One group flips a coin 50 times and writes on a whiteboard the results. The other comes up with 50 results.
They leave the room and when they come back immediately know which is which. Specifically because of the over-alternation that yours mentioned. A string of HHHHHH is exactly as likely as HTHTHT, but people don't intuit that.
> A string of HHHHHH is exactly as likely as HTHTHT, but people don't intuit that.
this contradicts the comment you're replying to. the premise they are having is that HTHTHT is "less random" than HHHHHH.
The strings are equally likely if they're being purely randomly selected, but not if they're being selected by people with a poor understanding of random. In the latter case, people are more likely to select HTHTHT.
I wonder what thought process led you to believe that even though the strings are equally likely and random, people who don't choose the one you think is more random have somehow inferior understanding of randomness
The issue the anecdote is reflecting is that across the data of 50 naive made-up coin tosses there are too many alternating sequences and probably zero HHHHHH.
and it's different from 50 randomly selected true coin tosses because you think it's impossible for all of them to alternate?
I understand your point, but the context is that repeatedly, experiments have shown that humans inject too much pattern to be good rngs. It turns out that is visible with training after the fact.
Yes, it is possible that any two 50-sequence patterns are actual rng. But some of them are much more likely, and we can tell.
Here it is:
https://aaronsonoracle.com/
I got 64% at 50 steps in, then dropped down to 53%.
The secret? I distracted myself with the website text and kept pressing the buttons quickly without thinking about it.
I have managed to get 38-48% the times I tried. I believe I might have been influenced by the article I just read. It would be interesting to see if by giving people advice on statics it would produce, on average, a measurably different outcome.
I got it down to 39% after 66 presses. It would be harder if it didn’t show you the rate. I was definitely learning from the feedback
A simple way to score 50%ish is to take a sequence of text and press the left button if even length, right button if odd.
This is what I also use for rock, paper scissors (mod 3). You can also use second hand of a watch for random samples, e.g splitting 0-30 vs. 31-60.
Similar to me, when I have to choose randomly I look for a number anywhere around. If it's odd, one choice, even is the other.
My strategy for rock paper scissors is to always go rock, and tell everyone that's my strategy. It's more fun for me seeing how people deal with that information than the win would be. Not very random though
I think Zipf's law means you're biased towards odd in that case (unless you allow yourself to choose empty strings)
I don't think Zipf's law gives you this. It mostly says that common words tend to be short, but it doesn't tell you what proportion of the text will be common words. For example, in English writing two-letter words are found much more than one-letter words.
However, I think you're correct anyway. The right distribution to look at is something like the orange series in the bottom graph on [0]. Eyeballing it, it does look like 3/5/7/9/11-letter words appear more in aggregate than those of 2/4/6/8/10 (49% Vs 44.5%), and longer words are a rounding error.
[0] https://medium.com/@davidxu90/relative-frequencies-of-words-...
I don't get the fuss? 163 presses in and the machine scores 48%. You just ... randomly press the buttons? Should I play for even longer?
> (...) Then, we found one student that the program predicted exactly 50% of the time. We asked him what his secret was and he responded that he “just used his free will.”
- Scott Aaronson, Quantum Computing Since Democritus.
Just use your free will, bro!
yeah i did 329 presses and got 55%. Doesn't seem that hard i wasn't even being very random, i just sort of zoned out and mashed them
I'd never heard of the Aaronson oracle so thanks for the link!
I actually built this on a whim after watching that episode of TNG where Data locks everyone out of the Enterprise.
https://youtube.com/watch?v=rERApU26PcA
That audio sample was used in a song by Pogo! I wondered which episode that was from.
https://youtu.be/bl5TUw7sUBs
Someone shared a small game based on this a few months back: https://news.ycombinator.com/item?id=48598051
(archived now): https://web.archive.org/web/20260625105410/https://brainfrog...
The implementation on that website (https://aaronsonoracle.com/) is uncanny, I recommend everyone to check it out
And it inspired me to build a little game based on it. See if AI can read you mind: https://jeffyclassify.com/blog/mind-reader/
Rather than attempt to be random, knowing that I'm not, I wanted to see how well it can predict Math.random:
This isn't very interesting, but it scratched an itch I had to see what would happen.
I really want to know who was more random than a RNG
It might not be that someone is more random than an RNG, but that the distribution of numbers in your 190 vs. theirs appears more "random". It's a limited sample size.
If yours happened to contain consecutive digits or common ones (ending in zero, significant numbers like 911 or 42 or 69), that would imply less randomness even if it's actually the result of "real" RNG.
i actually was, it said no one was more random than me! guess i just got lucky.
tldr; "im just like totally so random xD"
be honest, you held up spork, didn't you?
They were probably also a RNG… one of them has to win, just on chance.
It got 8% of mine.
Getting 0% isn't the goal though. You want to sit right at 11.11%, the blind guessing rate. If the prediction algo always predicts you wrong, it actually always "predicts" a number you won't guess. That too is a feat, reducing the pool of numbers you're going to choose from, therefore making you more predictable.
It looks like it's a fair implementation of guessing. Giving it digits of pi gives a very high (low) score - most times. To get a lower score I think you would have to be adversarial and/or get a lucky run. Based on the house doing better/worse on some runs I guess it not deterministic.
Seems I'm perfectly random.
I was. But if you're more random then you're probably not random. I was actively aiming to exploit the game - try to figure out which pattern detector it's keyed on based on its guesses, and then swap to a different pattern. Ended up getting 6% over a decent sample. You could probably get zero with a bit of luck!
The problem I had was somehow the double-pick algo randomly clicking on at seemingly illogical times. I'm guessing that maybe it's something like if you're not hitting any pattern matcher then the next guess uses a random pattern guesser, because it nailed me multiple times with the doubles even after not having one in 10-20 numbers prior. Though maybe that is itself a pattern that it checks for? It wasn't mentioned in the docs.
Dilbert
"Over here we have our random number generator."
https://web.archive.org/web/20230305130159im_/https://assets...
Thats technically true. I miss peak dilbert
xkcd with a similar joke: https://xkcd.com/221/
Does someone know whether people saw the "predictions" in the actual test? Because saying random numbers blindly is not what I was doing. After my next number was predicted correctly, I changed my "strategy" to increase randomness (be less predictable).
So I don't know if this result is comparable to the one from the study.
My results were:
The house called 10 of your 87 scored numbers: 11%.
Blind guessing would average about 9.7 of them (11%).
Only 5 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
You’re right. If you notice the instructions in the game, it encourages you to go as fast as you can. I keep the numbers up there just to make sure the voice recognition is working properly and to make the game seem more engaging.
But yeah, I agree: I think this does influence the results, since it makes it easier to be more random because you can essentially try to play in an adversarial fashion by adjusting your results according to feedback from the predictor itself.
In the original study, that wasn’t the case. They had a three-by-three grid of numbers and a metronome as well, which meant that they couldn’t slow down or speed up and had to go at a very specific rate with zero feedback until the very end.
Maybe I should add the equivalent of a "Hard" mode which makes the game more faithful to the original study.
I know pi to 100 decimal places and so that was pretty easy to cheat it
You should prove that digits of pi are randomly distributed :)
That is easy. I entered all of the digits into this website and it determined it was random.
I only have a few digits of pi memorized, so I had better luck using digits from fibonacci's sequence. I can derive them in my head, albeit slowly as the numbers get large.
> The house called 15 of your 113 scored numbers: 13%. Blind guessing would average about 12.6 of them (11%). The house called 1.2 times as many. Only 5 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
I did a PhD in a stochastic-related field though, so I know about some “tricks” (in that most humans are too random and eg. dont sample long running sequences)
An article was posted a few months back about a competition for the best rock-paper-scissor program, where submitted programs played each other. Point was to build the program that would be able to predict best the opponent program's moves (which itself is trying to predict you program's moves).
It was a very interesting read, and it explained a really "simple" strategy which I found very memorable: build 3 strings representing "all previous moves + one of the 3 possible next moves", run a text compression on them, pick the move that yielded the smallest compressed result. That's the one where the compression algorithm identified the strongest pattern.
I found it really smart, instead of modelling the game it went one step higher in abstraction and works directly at the level of pure pattern recognition.
That keeps coming back to mind when thinking about things like randomness, the "bitter lesson" and LLMs.
Is anyone sufficiently paranoid that their first thought was, "Are you going to train a model on my voice?"
(Yes, I saw the "type the numbers" option.)
Yes, my mind went directly here
Why would they when there are thousands of hours of radio and podcasts do train on?
And all that GOOG-411 data!
Maybe they are not training to make a TTS, but training on accents to make STT.
"Hey its me ur grandson"
because they need more random
It could be the first step of a very long social engineering process, who knows.
Ah, you see they are doing something far more random…
No, I thought of the penguin of doom...
Holds up spork
Many years ago, as a side project to help me apply to tech companies, I made this little site [0] where the premise is trying to pick the number that you think other people won't pick very often. I still think it's pretty neat.
Also, in part because of this project, I got to go from being unemployed in my hometown in the South to working in SF on some pretty neat projects. It's a good memory.
[0] https://www.humanrandomness.com/
Nice~ The website made me recall an old mentalist force trick we learned as kids where you ask somebody to "quickly think of a two-digit odd number between 1 and 50 where each digit is a different number".
Most people will say: 100101 (binary to avoid spoilers)
I absolutely said 100101
I went for 13, and thought it'd be the common one!
The pattern of holes is really interesting there. Everyone avoiding nice round numbers, so a bunch of those are very low frequency - except weirdly at the very ends of the range
The house called 12 of your 116 scored numbers: 10%.
Blind guessing would average about 12.9 of them (11%).
Only 2 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
What about the following strategy? You first think of a random integer n1 from 1 to 9, then you think of another one n2, perhaps one that you have not used very often, and finally say (n1+n2) mod 9 + 1. I think this strategy could make your selection more random . Perhaps someone could try it and report the results.
How would making your input deterministic make it more random?
Well clearly there’s a modulo operation in there and nobody knows how they work
Depends on the algorithm they use to find patterns.
You're mixing up unpredictable and random
Fair, but so did the person I replied to. For the purposes of what we are trying to do, unpredictable is what matters.
Imagine that instead of two numbers you select a million numbers, and apply a zip algorithm to that sequence to obtain a file then you map the first byte of that file using (first byte) mod 9 + 1, intuitively there must not exist any correlation between consecutive tries of this strategy. Looking for this example I realized that instead of using n1+n2 you should use a non commutative operation to improve the strategy, but when using just two numbers the addition suffices.
I found that saying two digit numbers like 82, 63, 45 etc was more random... maybe throw a couple double digits being the same in there like 77.
Or just use a PRNG? https://www.hillelwayne.com/post/randomness/
Somehow this reminds of Von Nuemann's trick. Take a coin, even a biased coin, and then (with H = heads, T = tails) :
HT => 0
TH => 1
HH => discard
TT => discard
It's pretty surprising at first, but this leads to a perfectly random and balanced distribution of 0s and 1s.
The essence of von Neumann's trick is that all permutations of independent events have the same probability. So the trick generalizes to any other set of possible results.
> The house called 3 of your 27 scored numbers: 11%.
> Blind guessing would average about 3.0 of them (11%).
Admittedly I gave up before the 30 threshold but I think I did well enough at "beating the house"
Edit: tried again, 14/118, which is 12%. Maybe it's because I'm typing, and that's easier to do that speaking up.
This is pretty interesting. I've found it's fun to throw off the algorithm. Use your warmup to repeatedly guess the same number, and then don't use it for a while.
Typing, so more random. Got a nice 9%, more random then most... And yes, did e.g. type 5-5-5 repeated.
Realize I said it wrong: 11%, 9 correct guesses (expectation if random would have been 9.3), three people more random in some other experiment.
That's quite a creative game idea. The house called 4 out of 40 number picks of mine. I did not use the speaking mode, so I do assume my randomness was inflated by that.
How are you sealing the predictions?
The house called 9 of your 117 scored numbers: 8%. Blind guessing would average about 13.0 of them (11%). None of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
Nice!
> The house called 1 of your 32 scored numbers: 3%. Blind guessing would average about 3.6 of them (11%). Only 1 of the 206 people in a psychology study was more random than you. The house found you extremely hard to read.
I found it interesting when I felt my focus shift from the number-generating task to noticing (and responding to) the house’s guesses.
As soon as I noticed it repeating its guess, I had a good long run of adversarially avoiding that number… it’s kind of fun being on the machine side of the slot-machine-type racket!
The first time I did badly. But the second time I got really far by deciding on a number instinctively and then picking something next to it. Maybe we instinctively pick similar numbers, but if we fight our instincts we end up with something truly random.
> Only 7 of the 206 people in a psychology study were more random than you. The house found you extremely hard to read.
I started just lagging the guesses of the machine, and it never seemed to catch on.
Didn't work for me, it started guessing them with 100% rate after just few of my replied guesses.
I did mix it up. So, 3 or 4 trailing guesses and then some random clicks and then trailing is what I did.
I just entered same number 5-6 times and it never caught it up.
Alternating and advancing from both ends seems to work, it calls n-1, and n+1 correctly if the previous guess was n, but got no calls in a sequence like - 1, 8, 3, 7, 4, 7, 1, 7, 3, 8, etc
Very cool, I really like that you have a typed and voice version
I'm reminded of the Dilbert Comic with Accounting's Random Generator.
I tried it. Several 9s later, it figured me out.
Too many alternations is exactly how people give themselves away.
The house called 19 of your 190 scored numbers and only 2 people were more random than me.... Hmmm
I guess I've skewed the totals a little, but the site agrees that the digits of π are random.
https://xkcd.com/1210/
I always thought I was random, but now it's conclusive!
Haha I was thinking of the Dilbert RNG strip when I made this
https://imgur.com/a/0azWegW
I was able to beat this quite easily by just saying the first 100 digits of pi out loud. The house guessed 1/10, no better than random. Maybe you should add a predictor for "well-known mathematical constants".
I got to 2/30 then it just guessed 7 from 30-50.
Did I break it?
Yup! Most of the way I put the system together was by having it build several n-gram Markov models of your numerical sequences on the fly.
But we also have a couple of predictors that aren’t markov chains and are based on simpler heuristics. Without seeing your record, I'm guessing that the "numerical avoider" predictor got overweighted and just kept trying to guess the same number.
It should be fixed now!
When I see the world “sealed”, I immediately know Claude generated the site copy.
Curious to know what model they use for speech recognition?
Not familiar with the sota if someone can chime in.
OP here - I went with Vosk because it's pretty trivial to lock input down to a set vocabulary (in this case numbers) and from my testing works relatively well on most browsers. (webkit, firefox, etc)
https://github.com/alphacep/vosk-api
this is such a poor model, I've tried it before. Try yapsnap it has better word error rate and runs super fast on CPU. https://github.com/kouhxp/yapsnap/
Thanks for the suggestion - I've been looking yadsnap over but it seems like the ONNX model for english is significantly larger than the one used by vosk (30mb versus 150mb) so that would be my only hesitation to use it.
Maybe there's a way to shrink it down given I only need to support a limited vocabulary domain...
Yup, it's q8. I would try q4 and q2 and if you are still unhappy with MB size, limited vocab.
Similar idea: https://aaronsonoracle.com/
I just pretended I'm playing piano to a megamix of tracks which popped into my head and did 'well'. Only 2 people more random. Typing though, think the site said that's "easier to win" on. Cba to speak 200 numbers out loud :/
Thanks, btw, actually enjoyed that. :)
Haha, that’s great. Yeah, I totally get what you mean about speaking out loud.
You only really need to vocalize about ~40 numbers total, but you can actually get a similar level of RNG by "subvocalizing the numbers" before you press them on the keypad.
The main issue is that using the keypad kind of lends itself to just pressing it randomly without really thinking about the actual numbers themselves.
"You spread your picks more evenly than a random draw would. Your most-used number made up 23% of your picks, where a random draw's busiest number gets about 23%."
??? huh?
That's an explanatory bug on my part.
I have a little "mad-lib templatized" set of sentences for reporting on various stats, and that one was supposed to be sort of related to the law of small numbers. For example, shorter runs mean there is a larger probability of getting lopsided distributions (different from pure 11%).
It's supposed to demonstrate that even though you used a single number a disproportionate number of times, it's not necessarily statistically significant since shorter RNG runs can produce the same effects.
blobrand dot com
House called 0 of 50. Huh?