I never have made it to the end of a season, just due to lack of time. It's great fun, though - already solved today's "warm up" puzzles...
Advent of Code 2024
131–140 of 580 posts
Re: Advent of Code 2024
#132I maintain a joke domain which I've found sadly applicable through the years. https://adventofrealizingicantread.com. I try to keep it updated pointing to the current day throughout the month
I've found especially as the month progresses it's just as much Advent of Reading Comprehension as it is coding :)
Re: Advent of Code 2024
#133Earlier quoted context omitted.
No stars are required at all. You can also skip some days if you don't want to do them. Or only to the first stars. However you like. First/silver stars will also increase your score in the leader boards.
I think you need all stars to do the final puzzle.
That might be true, I've only completed it once a few years ago. But if you don't want to the more complicated parts, then you want to skip the last one for sure.
Re: Advent of Code 2024
#134Last year I got stuck on Day 12 for a full week, and thinking about how to solve it consumed my every waking moment. I think this year, I'm going to be kind to myself and not participate so I can really enjoy the winter break from work.
All things in moderation! Now if I can just make it through the winter without installing the new Factorio DLC...
Re: Advent of Code 2024
#135Re: Advent of Code 2024
#136This is the first time I've seen Firefox's alternate stylesheets mentioned in the wild. Interesting.
Re: Advent of Code 2024
#137Earlier quoted context omitted.
The global leaderboard is so fast that any AI assistance would literally slow them down, here's one of the guys who tends to score highly solving today's puzzle. ( https://youtu.be/ym1ae-vBy6g ), and on the more complicated days that's even more pronounced because anyone who is even somewhat decent doesn't need to ask chatgpt how to write Dijkstra. Obviously if you're doing it recreationally you can cheat with AI but…
The thing is that the AI can read a puzzle faster than a human can. If someone put any effort towards an AI-based setup, it would easily beat human competitiors (well, up until the point the puzzles got too difficult for it to solve).
I've always done AoC "properly" but this year I've decided to actually use it as a learning experience for working with LLMs (and I don't get up early so will never sully the leaderboard) and trying some experiments along the way.
I think the strategy for the harder puzzles is to still "do" them yourself (i.e. read the challenge and understand it) but write the solution in English pseudocode and then have an LLM take it from there. Doing this has yielded perfect results (but less than perfect implementations) in several languages for me so far and I've learnt a few interesting things about how they perform and the "tells" that an LLM was involved.
Re: Advent of Code 2024
#138I’m using sqlite this year. Hoping that there won’t be any computational geometry or trie problems. Kind of hoping for a graph problem solvable with recursive CTEs, that would be cool.
Re: Advent of Code 2024
#139I’m attempting to make my own language for solving the puzzles as I go along.
I'd pretty much decided I wasn't going to participate because it would take up too much time, but now you've both made it way more complicated and way more appealing...
Re: Advent of Code 2024
#140Earlier quoted context omitted.
Are we using enshittification for everything we don't like these days? We invented calculators, those really enshittified manual arithmetic puzzles. Private boards for this stuff makes sense anyway, it's the Internet afterall.
Using LLM for this kind of thing is like using cheats or aim assist for online games. So yes, this is prime example of enshifittication.
I think the intent is important. Using LLMs to do well on the public leaderboard is like using cheats/aim assist. But learning how to use LLMs to solve complex puzzles independent of any sense of "competition" is more like when people train neural networks to drive a car in GTA or something - it's not hurting anyone and it can be a real learning experience that leads to other interesting byproducts.
But, yeah, don't use LLMs to try and get 9 second solve times on the public leaderboard, it's not in the spirit of the thing and is more like taking a dictionary to a spelling bee.