Former CP contestant rejects "raising difficulty to counter AI": pointless
anpaure · x · 2026-08-27
Responding to suggestions of raising contest difficulty to counter AI, a former competitive programming skeptic expressed strong disagreement, arguing:
a) By 2027, models will be smarter than all human contestants and organizers combined, making it a poor benchmark.
b) Raising the ceiling is pointless as the current score distribution looks fine.
c) AI affects programming contests like Deep Blue affected chess: it doesn't, except for cheating.
More from AGI Musings
- David Brooks: Replacing human opinion writers with AI leads to predictable, repetitive content — zacharynado · 2026-08-27
- Debate on whether Nvidia's billions in profit equals new GDP growth — EigenGender · 2026-08-27
- Ex-OpenAI Staffer: Moral Duty Near Power Is Not To Be a Cog — RichardMCNgo · 2026-08-27
- Weeks-long Sol agent swarm burned billions of tokens and invented its own private vocabulary — Sauers_ · 2026-08-27
- Chinese Models Go Anonymous: OxAlpha Hype and Identity Mystery — 创业邦 · 2026-08-27
- AI Money Surge Leaves Researchers Feeling Poorer and Anxious — pratyushmaini · 2026-08-27