Sharing a bit of backstory on why I decided to work on this; Firstly, “for fun” but primarily because I felt like I started losing the childlike wonder/whimsy I once had with programming.
So I started this new hobby where I ask myself “can I hack on this?” upon getting/seeing something.
For instance, I got this new Aula F75 keyboard (really good keyboard for the price btw, it sounds good too!) and it only has dedicated control software for Windows. So I downloaded the driver files, software executable, and manual sheet and reverse engineered the full protocol/packets and rebuilt it for my Mac. Then played snake with the backlights. Fun.
Anywho, happy to see my blog on the front page. Would love to hear if anyones going through something similar or working on silly little projects! :)
When I saw the title, I thought of Lambda Calculus[0] and SKI combinators[1]. Given that there are "only six useful colors", I wonder if M&Ms could be used to implement them.
Funny you mention that, because yes, a combinator-style encoding is probably a cleaner fit for the “only six colors constraint than my stack machine. I hacked together a tiny SKI-flavored M&M reducer as a proof of concept: B=S, G=K, R=I, Y=(, O=), and N... is a free atom, so `B G G NNN` reduces to `a2`.
Fun project! I had a similar project a while back, but my medium of choice was the Uno card game. I called it UnoScript [1] and it had similar mechanisms as color was an important factor. I also ended with a stack as the main part of the language, where different colors/combinations of cards could read from/modify the stack. Interesting how similar constraints can lead to some similar design choices!
This makes the world a better place. I got a little oxytocin hit just from the thought that somewhere on this world, someone is working on this problem. Now I'll be kinder to old ladies and give those poor puppies a pass.
it actually sounds like a fun idea, but i have one question. do you think a lightweight CNN trained on synthetic candy layouts would outperform the deterministic decoder for messy real world photos?
Yes, for messy real-world photos a lightweight CNN would probably outperform the deterministic decoder, but I’d still use it in a hybrid pipeline with classic CV for blob detection and deterministic logic for reconstructing the actual program.
It gave me strong vibes too. It’s the writing style. I’ve seen OpenAI write just like this. Doesn’t mean it’s bad. There’s a few other markers. Note “silly” in quotes and over use if that word. Once would be enough. But also this is very very typical. The bolting and short quite direct and a bit repetitive statements “it absolutely does not solve “dumped a bag of candy on a messy kitchen table and took a dramatic iPhone shot.”
Real example programs are where the joke becomes a language
I didn’t want this to stop at “hello world with candy colors.”” The over use of quoting. The bold. It’s not like a human wouldn’t write this. But it’s unusual for a human to do this imho. All the same - it feels novel. And at the end of the day it’s a neat idea. It’s just we enter this new brave world where things written like this give you the ick. “Where do the ai learn this from?” Well I wouldn’t mind betting the author asked it to be written in a hn style post.
That's interesting! But confusing as well. The test they reference (https://en.wikipedia.org/wiki/Psychopathy_Checklist) includes lack of empathy. Are they saying that this criterion could be dropped from the test with no effect on the classification?
How I understand the article, is that they understand why others act in certain ways, they know the mechanism of empathy, but nothing here confirms that they are empathetic themselves. I think this article's conclusion is misleading.
Author of this silly project here!
Sharing a bit of backstory on why I decided to work on this; Firstly, “for fun” but primarily because I felt like I started losing the childlike wonder/whimsy I once had with programming.
So I started this new hobby where I ask myself “can I hack on this?” upon getting/seeing something.
For instance, I got this new Aula F75 keyboard (really good keyboard for the price btw, it sounds good too!) and it only has dedicated control software for Windows. So I downloaded the driver files, software executable, and manual sheet and reverse engineered the full protocol/packets and rebuilt it for my Mac. Then played snake with the backlights. Fun.
Anywho, happy to see my blog on the front page. Would love to hear if anyones going through something similar or working on silly little projects! :)
Great post, thanks for sharing it!
When I saw the title, I thought of Lambda Calculus[0] and SKI combinators[1]. Given that there are "only six useful colors", I wonder if M&Ms could be used to implement them.
0 - https://en.wikipedia.org/wiki/Lambda_calculus
1 - https://en.wikipedia.org/wiki/SKI_combinator_calculus
Funny you mention that, because yes, a combinator-style encoding is probably a cleaner fit for the “only six colors constraint than my stack machine. I hacked together a tiny SKI-flavored M&M reducer as a proof of concept: B=S, G=K, R=I, Y=(, O=), and N... is a free atom, so `B G G NNN` reduces to `a2`.
Gist: https://gist.github.com/mufeedvh/db930a423fdce8c1d8e495c7a3f...
Fun project! I had a similar project a while back, but my medium of choice was the Uno card game. I called it UnoScript [1] and it had similar mechanisms as color was an important factor. I also ended with a stack as the main part of the language, where different colors/combinations of cards could read from/modify the stack. Interesting how similar constraints can lead to some similar design choices!
[1](https://github.com/berlinquin/UnoScript)
This makes the world a better place. I got a little oxytocin hit just from the thought that somewhere on this world, someone is working on this problem. Now I'll be kinder to old ladies and give those poor puppies a pass.
Does this work with real candy?
Yes! Just make sure to take a photo on a plain white surface is all.
With:
It’s funny until one guy spills his bag of M&M’s and accidentally deletes the production database.
Wanted to fix this bug but I ran out of green M&M's
Will read this later, but in the meantime, if you're into “physical programming”, check out efghij: https://esolangs.org/wiki/Efghij
What color is your function?
That's one language that doesn't need an external IDE for syntax highlighting.
It's all fun and games until some fat bastard like me decides he wants a snack. Incidentally, which flavor? Asking for a friend.
This is what HN is all about
Thank you.Awesome project!
Unfortunately I think you have given my kids one more hard to refuse reason to get more candy: I need more storage.
Looking forward to the first LLM&M coding agent
I can't see it explained, why/where is 000000 defined as newline?
it actually sounds like a fun idea, but i have one question. do you think a lightweight CNN trained on synthetic candy layouts would outperform the deterministic decoder for messy real world photos?
Yes, for messy real-world photos a lightweight CNN would probably outperform the deterministic decoder, but I’d still use it in a hybrid pipeline with classic CV for blob detection and deterministic logic for reconstructing the actual program.
but can it run Doom?
oh man the program just melted in my mouth
this is so cool!
thank you! :)
This is AI slop but mildly amusing. Brainfuck did it first.
Which part made you conclude there's AI involved?
the bold section headers and bullet points. but who cares. i don't.
I literally write like the article on similar write ups, where do you think the AI's learned to write this way from.
I really don't get the AI vibes from the actual writing of it
It gave me strong vibes too. It’s the writing style. I’ve seen OpenAI write just like this. Doesn’t mean it’s bad. There’s a few other markers. Note “silly” in quotes and over use if that word. Once would be enough. But also this is very very typical. The bolting and short quite direct and a bit repetitive statements “it absolutely does not solve “dumped a bag of candy on a messy kitchen table and took a dramatic iPhone shot.”
Real example programs are where the joke becomes a language I didn’t want this to stop at “hello world with candy colors.”” The over use of quoting. The bold. It’s not like a human wouldn’t write this. But it’s unusual for a human to do this imho. All the same - it feels novel. And at the end of the day it’s a neat idea. It’s just we enter this new brave world where things written like this give you the ick. “Where do the ai learn this from?” Well I wouldn’t mind betting the author asked it to be written in a hn style post.
Yeah, the execution is stronger than the idea and design choices as well, which suggests heavy AI support
Please re-read your comment and tell me you're not grasping at straws just to accuse someone of using AI to write.
You have a problem. I would suggest thinking about it.
Are you serious? We can't make headers bold or use bullet points anymore without people like you instantly calling it AI slop?
Not really the same, but made me think of this classic xkcd: https://xkcd.com/505/
Am I allowed to use the term psychopath in the most loving, even inspired, way?
Psychopath implies lack of empathy so I don't think that's quite the word you want. You could maybe repurpose "psychotic" though!
Maybe lack of M&Mpathy?
“ As it turned out, there is nothing special about psychopaths when it comes to understanding or feeling empathy with others. ”
https://www.psychologytoday.com/us/blog/fulfillment-at-any-a...
But maybe it is like so often more about the contradictory definitions of “empathy”, and capability vs. willingness.
That's interesting! But confusing as well. The test they reference (https://en.wikipedia.org/wiki/Psychopathy_Checklist) includes lack of empathy. Are they saying that this criterion could be dropped from the test with no effect on the classification?
How I understand the article, is that they understand why others act in certain ways, they know the mechanism of empathy, but nothing here confirms that they are empathetic themselves. I think this article's conclusion is misleading.