I made a puzzle game called Spaghetti Network!
The goal is to build a graph that computes a given function.
You start with simple problems like A + B, and later get to multiplication, Fibonacci numbers, GCD, popcount, and more.
After solving a problem, you can also try to minimize the number of nodes and compete on the leaderboard.
You can play it here: https://spa-net.com/
Thanks to camypaper, DEGwer, hos.lyric, nuip, semiexp, sigma425, snuke, yosupo and Yupiteru for playtesting and feedback.
Feedback and discussion are welcome!
I should have read the tutorial...
Really fun to play! I solved up to Fibonacci (inclusive with some skips), really original idea for a game ;) I think the progression in hardness is great, because each problem requires thinking even after solving all the previous ones.
The best puzzle I've ever played
fun... but i can't play it on Win 7 :sob:
Its essentially DFA construction right?
Really awesome, tho site seems to be down rn
orz, but i cant see the solutions of my past submissions
Try the floppy-disk icon at the top right of the canvas — every network that beat your best cost is auto-saved there. Thanks for the feedback!
really really fun and peak , keep it up .
I loved it, but some suggestions:
Please add an option to view past submissions
Could you add an option to filter by friend or filter by country
how do you do sign? nvm i got it
congrats, it's so fun, loved it! small suggestion: I'd love to be able to label my nodes and reuse saved networks between puzzles
maybe add functions so u can reuse code from previous submissions idk if that sounds stupid
This is great I love it :D
It would be cool if there was a way for user to quickly see which tasks they havent perfected yet.
There is, just go to your profile and you'll see your score vs the best for all levels
Thanks! I’ve just added a “Show details” option to the Tasks page for logged-in users. It shows your cost and the global best side by side for each solved task. Thanks for the suggestion!
There’s a really stupid solution to Gate without self loops
Invert B, if this inverted value is 1 we use it as the base of a quickly growing sequence we can manually code in. I used the Fibonacci sequence as it was easy to do. Then when it reaches >100 we subtract A by the final value. Then just add A into X. Obviously we can also use this for Select.
I was wondering how cases are tested in the backend. Do you do anything smarter than simulation, or do you just simulate? If you simulate, when do you decide to abort the test?
Thanks
if you try a circuit like this: A self + to A, you will see Aborted, 1e6 ops (1e6 ops is max, you can see that in help too)
Can someone please give an hint for approaching the task Abs Diff while having the cost as 1. I solved it with cost as 2 but couldn't get any idea for cost as 1.
First, transfer a − b into x and b − a into a separate node. Then, if b − a is positive, add twice that amount back to x
Hmm I did the same except that I could'nt figure out how to add twice of (b-a) without additional node. Figured out later after so many tries. Thank you for the help!!!
Man, I love it. It's so fun. It would be better if you add dark mode though :)
can anyone give a hint for multiplying i can't figure out how to run it b times
Is it intended that you can see hidden test? I am not complaining because I solved a problem with that hidden test
Yes, that’s intentional. "Hidden tests" are really just additional test cases and aren’t meant to remain secret, so feel free to use failed cases as hints.
Very fun game! I want a sandbox mode, it can help me save my big projects and set inputs fastly.
Interesting game! Kudos to the developers.
I wonder if the game can be theoretically solved. We seem to want to know what functions are computable by the network.
A network is obviously a discrete object, since there are states, edge types, and the ordering of the states from left to right determines the computation. Therefore, we might be able to automate the procedure of designing networks / computing certain functions, or determining what's computable.
Or maybe not?!
You can very easily chain together basic operators to make any program, but the fun is in trying to make as small a solution as possible
Wow this is really fun!
Can you add a feature to go to a spesific step of the replay , I often find myself trying to find the exact spot where things went wrong while debugging the network multiple times
Yeah, a redo button would be helpful
Super fun