Developer trains language model on demoscene code to autocomplete 140-char demos
A coder used ML to learn the patterns in tiny JavaScript demos, aiming to finish works and inject personal style via AI.
What to know
- A developer trained a language model on constrained code demos (140-character JavaScript) to generate plausible continuations and autocomplete unfinished works.
- The demoscene community values human constraint-solving and creative struggle; commenters worry AI automation hollows out the core appeal.
- The tension centers on what the scene is about: math/puzzle-solving vs. aesthetic expression—and whether AI assistance helps or replaces the human effort.
The dispute Whether AI autocomplete is a tool to assist human creativity (understanding patterns, exploring alternatives) or an erosion of the demo scene's core value: the satisfaction of solving hard problems within extreme constraints without automation. · positions read across 5 posts and comments
AI autocomplete undermines the demoscene's core appeal: human constraint-solving and creative struggle are the point, not optimization.
-
“I think you misunderstand the point of the demoscene. It's about mathematical prowess and the joy of puzzle-solving moreso than pretty pictures. Nobody in the scene wants a magic box that solves the puzzle for them.”
InvisibleUp · Hacker News ↗
AI models trained on art are not yet creative enough; they remain derivative and lack the human innovation that makes demos compelling.
-
“Less fun to see models imitating human art, they're not ready for that yet IMP because they're too derivative and not creative in interesting ways in the way that humans are.”
grey-area · Hacker News ↗
datavorous Developer / researcherHellMood Demoscener
How it unfolded 2 developments, newest first · click a bar or a number to jump articlespostscomments
-
2
Demoscene participants push back on AI autocomplete framing
Comments from HackerNews users active in or familiar with the demoscene highlight a fundamental tension: the scene values human constraint-solving and creative struggle, not automation. InvisibleUp argues the core appeal is mathematical prowess and puzzle-solving, not pretty pictures, and questions whether an AI solver undermines that ethos.
“I think you misunderstand the point of the demoscene. It's about mathematical prowess and the joy of puzzle-solving moreso than pretty pictures. Nobody in the scene wants a magic box that solves the puzzle for them.”
— InvisibleUp -
> ... and the audience will ooo and aah because a human figured out how to do that stuff under constraint.Or because the audience appreciates the artists' emotion and creative effort embodied in what they're watching. The Razor1911 demo, released at Revision 2026, is a great example there - there's some neat effects but nothing groundbreaking, but…
2 more of the top 3 · 5 posts in this stretch
-
The human side of the demo scene is what I think makes it appealing. Sharing and consuming. A fellow human did that! Coder-as-problem-solver-and-artist.For sure it’s intellectually interesting if AI can generate something novel within demoscene constraints (140 chars or extreme low res or a POKEY chip at 1MHz or whatever) … something humans maybe…
-
Misread this as Demosthenes at first.https://en.wikipedia.org/wiki/DemosthenesWould be fun to see models trained on ancient figures that you could interact and chat with.Less fun to see models imitating human art, they’re not ready for that yet IMP because they’re too derivative and not creative in interesting ways in the way that humans are.
-
-
1
Developer publishes post-training experiment using ML on demoscene code
Datavorous published a detailed writeup of training a language model on demoscene and sizecoding projects, specifically 140-character JavaScript demos from Dwitter.net. The project traces visual effects across remixed demos and applies autoregressive language modeling to generate plausible code continuations and explore alternative aesthetic directions.
“Given a partial program, certain characters and expressions are considerably more likely to follow than others. This is pretty much how autoregressive language models work.”
— datavorous -
1 outlet I post trained a model on Demoscenes
first by HN Frontpage, 11d ago
-
What people are saying 1 voices from 1 site · best of 5 · verbatim
- Oct 31
-
Neat.To me, demoscene was about embracing new technology and using it to express yourself.It's enshrined now and so I find it to have served it's purpose. Look forward, not back.