Advent of Code 2023
Each year dozzens of developers particpate in Advent of Code, a series of riddles and challenges that are intended to be solved by writing code. In 2022 at least 290,510 participants completed a challenge1.
Interestingly if a particpant completed the first star of a challenge, they were extremely likely to complete the second star. Even on the final challenge only 35% of people that got the first star on the challenge didn’t go on complete the full challenge:
Challenge | Completed Entire Challenge | Partially Completed | Pct Stalled |
---|---|---|---|
25 | 12922 | 7098 | 35% |
24 | 16081 | 237 | 1% |
23 | 18225 | 250 | 1% |
22 | 15118 | 5058 | 25% |
21 | 22712 | 3017 | 12% |
20 | 21850 | 992 | 4% |
19 | 17205 | 935 | 5% |
18 | 25101 | 4871 | 16% |
17 | 21194 | 6207 | 23% |
16 | 22912 | 5557 | 20% |
15 | 39187 | 5713 | 13% |
14 | 48461 | 965 | 2% |
13 | 50893 | 1202 | 2% |
12 | 58551 | 1048 | 2% |
11 | 68131 | 9154 | 12% |
10 | 84182 | 5446 | 6% |
9 | 81592 | 11306 | 12% |
8 | 106078 | 7526 | 7% |
7 | 111992 | 2555 | 2% |
6 | 154157 | 1580 | 1% |
5 | 155600 | 3056 | 2% |
4 | 182179 | 3657 | 2% |
3 | 197396 | 11196 | 5% |
2 | 229512 | 12071 | 5% |
1 | 276012 | 14498 | 5% |
I’ll be participating again this year and I’m hoping to make it past my current record of… 7 days. I have good intentions but things always get in the way.
This year I’ll be trying Maciej Caderek’s framework for running the challenges. I’m thinking if there’s a bit more abstraction I’ll have better luck. Last year I tried to make some tooling but ended up spending more time fighting Jest than writing AoC code.
I’m going to stick with TypeScript as much as I’d like to pick up Golang or Rust I don’t have as much time this year.
This year AoC has added a special note asking contestants to not leverage AI to quickly solve problems and get onto the leader board.2
Can I use AI to get on the global leaderboard? Please don’t use AI / LLMs (like GPT) to automatically solve a day’s puzzles until that day’s global leaderboards are full. By “automatically”, I mean using AI to do most or all of the puzzle solving, like handing the puzzle text directly to an LLM. The leaderboards are for human competitors; if you want to compare the speed of your AI solver with others, please do so elsewhere. (If you want to use AI to help you solve puzzles, I can’t really stop you, but I feel like it’s harder to get better at programming if you ask an AI to do the programming for you.)
Footnotes
-
It’s possible to skip challenges so it’s impossible to tell which participants only completed one of the later challenges so the true number is likely higher. https://adventofcode.com/2022/stats ↩