The contest ended a few days ago, some teams have already published their writeups. Usage of LLM-based tools appear to have mixed results. For example, I think #23 place claimed to have used nothing but AI[1], while the #24 place seems to be mostly manual code[2]. These are very good results considering they solved all problems, only 40 out of ~268 teams managed to do that[3].
I can confirm that I used Fable exclusively (I am link [1] above). I even got sunburn on one of the contest days because I spent it mostly at the pool, as mentioned in the article.
Levity aside, I remain deeply impressed by all of the extraordinary efforts - including yours! - and of course, the contest's organizers' work.
Based on the timestamps, I guess it's over now?
I'm curious to know how LLM-resistant it ended up being in the end.
The contest ended a few days ago, some teams have already published their writeups. Usage of LLM-based tools appear to have mixed results. For example, I think #23 place claimed to have used nothing but AI[1], while the #24 place seems to be mostly manual code[2]. These are very good results considering they solved all problems, only 40 out of ~268 teams managed to do that[3].
[1] https://blog.greg.technology/2026/07/26/doing-okayish-with-a...
[2] http://www.sawicki.us/icfp/2026/
[3] https://icfpcontest2026.com/standings
I went through the 3 days without using any AI, working mostly in the language designed by the judges. I placed somewhere in the middle.
I can confirm that I used Fable exclusively (I am link [1] above). I even got sunburn on one of the contest days because I spent it mostly at the pool, as mentioned in the article.
Levity aside, I remain deeply impressed by all of the extraordinary efforts - including yours! - and of course, the contest's organizers' work.
I am but a poster child for laziness :-) but, as Larry Wall said... https://thethreevirtues.com/ :-)
Planner did not understand fat man or wall clock
Reminds one of Orca.
https://100r.co/site/orca.html