2025’s ‘advent Of Code’ Event Chooses Tradition Over Ai

Are you fresh to code? Today, programmers astir nan world commencement counting nan hours until Dec. 1, and 2025’s yearly “Advent of Code” event.

And this twelvemonth there’s thing caller — it’s nan first twelvemonth pinch immoderate very large changes. After 10 years of tantalizing coders pinch 25 two-part puzzles each year, arena creator Eric Wastl explained what’s changing — and why — successful a short, succinct announcement successful October.

But what’s arsenic important is really nan organization reacted.

anyone up for doing Advent of Code successful OCaml this year? 🎄🐫

— sabine_s_ (@sabine_s_), Nov. 12, 2025

AI may beryllium changing nan world of programming, but immoderate quality vacation traditions continue. And arsenic hundreds of thousands of coders tube toward different twelvemonth of holiday-themed puzzles — for illustration reindeer flying done a North Pole blizzard — it’s bully to spot that their shared communal excitement will beryllium seeing yet different year.

Advent of Code 2025: Fewer Puzzles for More Accessibility

There’s nary mistaking nan wide fondness for nan site. According to its statistics page, 284,977 group solved past year’s first puzzle. (Completing a puzzle awards a “star,” and Wastl has announced that 23,170,305 stars person been awarded since nan site’s motorboat successful 2015.) In fact, 779 users person solved each puzzle, each twelvemonth — earning each 500 stars.

And complete nan years, much than a cardinal group person collected astatine slightest 1 star.

But speaking past twelvemonth at nan C++ convention CppNorth, Wastl admitted it’s difficult to lick each 25 puzzles successful a azygous year. “I effort to make nan opening ones easy, and I effort to make nan later ones hard.”

This intends that while 263,746 group solved some parts of past year’s first puzzle, only 17,088 conquered its last puzzle.

So for 2025, Wastl is reducing nan number of puzzles from 25 to 12. “It takes a ton of my free clip each twelvemonth to tally Advent of Code,” Wastl wrote successful nan site’s FAQ, “and building nan puzzles accounts for nan mostly of that time.

“After keeping a accordant schedule for 10 years(!), I needed a change.”

Responses connected societal media person been mostly supportive…

“Honestly, this makes it easier to participate,” posted 1 backend services developer. “I was ne'er capable to support doing it each time arsenic nan holidays sewage closer, excessively overmuch other to do!”

And a Los Angeles-based engineering head added that nan arena “is specified a gift to each of us, immoderate you springiness america is simply a blessing! I’m gladsome you’re capable to find a equilibrium that useful for you.”

On Reddit, Wastl confirmed that he’s still readying to person 2 parts for each puzzle (joking that “I reserve nan correct to immoderate time person a 37-part puzzle!”).

But will this impact nan trouble of nan puzzles?

“I’m still calibrating that,” Wastl posted connected Reddit. “My dream correct now is to person a much condensed type of nan 25-day complexity curve, possibly skewed a small to nan simpler guidance successful nan mediate of nan curve? I’d still for illustration thing location for everyone, without outpacing beginners excessively quickly, if I tin negociate it.”

advent of codification 2024 successful lean4 advancement thread.

day 1/25

— spikedoanz (@spikedoanz), Oct. 9, 2025

Global Leaderboards Removed and AI Use Policy

And nan arena is besides officially discontinuing its world leaderboard showing nan fastest finishing times, which Wastl writes was “one of nan largest sources of accent for me, for nan infrastructure, and for galore users. … What started arsenic a nosy characteristic successful 2015 became an ever-growing problem.”

One rumor was “People took things excessively seriously, going measurement extracurricular nan tone of nan contest; immoderate group moreover resorted to things for illustration DDoS attacks.” But he seems particularly bothered by nan measurement nan fastest times from astir nan world seemed to discourage others astir their ain programming skills.

One Hachyderm personification welcomed nan change, saying nan leaderboards had brought “a batch of dopamine for a very mini magnitude of group (mayyyybe nan top100?) and a batch of dread for everyone else.”

And different personification said they welcomed nan changes, since they’d recovered nan arena “always took excessively overmuch clip towards nan end.” (And “I thought nan world leaderboard should’ve gone a fewer years agone anyways, erstwhile LLMs started being a thing.”)

Thursday evening astatine my accustomed java shop, sipping an Earl Grey–Hot, doing immoderate WordPress tract dev, and squeezing successful a small clip for #AdventOfCode.

— andrewhoyer (@andrewhoyer), Nov. 6, 2025

In fact, nan site’s FAQ database now besides specifically tells users that they shouldn’t usage AI erstwhile solving puzzles. “If you nonstop a friend to nan gym connected your behalf, would you expect to get stronger…?

“If you want believe prompting an AI, location are almost surely amended exercises elsewhere designed pinch that successful mind.”

The FAQ moreover addresses users who want to usage AI erstwhile competing pinch others connected a backstage leaderboard, urging them to first inquire nan group moving nan committee astir their policies and expectations.

Community Reactions to Advent of Code Changes and AI

So really did nan organization react?

As nan title drew nigh, 1 instrumentality launched a Reddit discussion conscionable to stress Wastl’s no-AI admonition — and recovered others who agreed. “Using AI to do Advent of Code this twelvemonth is for illustration mounting up a checkers committee connected your own, taking distant each nan pieces of 1 color, and declaring you won,” joked 1 Reddit commenter. “Yeah, nobody’s stopping you from doing it … but what do you get retired of it?”

Not everyone agreed. “I americium unashamedly going to usage AI this year,” 1 commenter responded, “because I want to study Golang. I’m readying to get interactive training specifically geared to each puzzle.” And different wrote, “My scheme is to usage AI to codification nan input parsing. Always hated that part. After that, I scheme to move it off.”

And ironically, OpenAI bought an advertisement successful nan charismatic Advent of Code subreddit. (“Codex gets you up to velocity accelerated pinch straightforward summaries truthful you tin support moving. All powered by ChatGPT…”)

But programmer Jeroen Heijmans publishes an unofficial survey of thousands of participants each year, and since 2023 has been asking an further question. What do you deliberation of AI/LLMs [large connection models]?

“I was unprepared for nan measurement and wide request for moderation of these answers,” Heijmans said erstwhile announcing 2024’s study results on Reddit. More than 62% of respondents said they utilized “zero” AI — astir nan aforesaid percent arsenic successful 2023 (when Heijmans first began asking nan question).

Multiple answers were allowed, pinch 31.8% selecting “AI is bad for Advent of Code” (up from nan 27.0% successful 2023). Additionally, 21.8% selected an moreover much emphatic action — “AI is horrible for Advent of Code” — a large jump from nan 15.4% who selected that reply astatine nan extremity of 2023.

And 39.2% selected “Not again pinch AI” — a flimsy driblet from nan 40.7% who’d selected that reply successful 2023, while 0.6% chose “Don’t cognize what AI/LLM means” (down from 1.0% successful 2023).

Not everyone is avoiding AI: 15.7% of 2024’s study respondents said they’d utilized “some” AI erstwhile solving nan puzzles (with an further 0.5% saying they’d utilized “lots” of AI). But this fig was down somewhat from 2023’s 14.1% who’d reported utilizing “some” AI and nan 0.7% who’d utilized “lots.”

But successful 2024, conscionable 7.6% chose “AI is bully for Advent of Code,” pinch 2.4% choosing “AI is great for Advent of Code” — a flimsy driblet from nan group who’d chosen those answers successful 2023 (10.8% and 3.6%, respectively).

Programming Trends and nan Enduring AoC Tradition

But mostly, nan organization conscionable seems grateful that their December contented continues.

This year, 1 Reddit personification moreover proposed a caller challenge — trying to lick each nan puzzles without if-then statements aliases different “flow control” keywords for illustration while loops.

So, what will this year’s arena look like? Through nan years, nan astir celebrated programming connection for solving nan puzzles — by acold — seems to beryllium Python, according to Heijmans’s unofficial survey, pinch astir 40% of its respondents saying they utilized Python successful 2024. (And 40% person said nan aforesaid point each twelvemonth since 2018, erstwhile his study began.) Rust has been a accordant 2nd choice, utilized by much than 16% of participants successful each of nan past 3 years.

For nan past 7 years, much than 30% of solvers usage a Linux OS, according to nan study results, while nan stock of Windows users dropped from 40.5% successful 2022 to 35.86% successful 2023, and past to 33.239% successful 2024. Heijmans notes that nan Windows Subsystem for Linux sewage different 7.2% of responses successful 2024, meaning that Linux and WSL together mixed for much users than Windows only.

And for nan past 4 years, much than 40% study they utilized VS Code arsenic their codification editor.

And arsenic they count down those last hours until midnight (EST, aliases 9 p.m. PST), immoderate eager coders whitethorn moreover find themselves reference Eric’s concealed greeting successful nan root codification of nan contest’s location page.

“A batch of effort went into building this thing,” he tells visitors to his site, adding, “I dream you’re enjoying playing it arsenic overmuch arsenic I enjoyed making it for you!”

Group Created pinch Sketch.