Claude and John Carmack Debate the 'clamp'
ID_AA_Carmack · x · 2026-07-17
John Carmack mentioned asking Claude why a specific clamp was necessary. Claude casually replied that if an "always true" clamp makes you uncomfortable, it's merely an artifact of next-token prediction.
However, Claude's subsequent technical reasoning was spot on: while mathematically redundant, different reduction paths necessitate this guard to prevent issues like sqrt(-epsilon). Carmack described the exchange as being "gaslit," pointing out that kernel calls and parameter read/writes aren't entirely free, unlike a constant-true branch in standard C.
More from coding & agent
- Coding agents are heading toward an AI-writes, AI-reviews, human-approves workflow — aftahi_ai · 2026-07-22
- oMLX 0.5.2 adds Mac menu-bar stats, low-bit decode kernels, and faster downloads — awnihannun · 2026-07-22
- GitHub review bot hits its PR limit and forces a 39-minute cooldown — DanielLockyer · 2026-07-22
- Max reasoning effort appears to be mobile-only in Codex Remote, not desktop — GabGarrett · 2026-07-22
- A Reddit demo argues online stores should expose carts and pricing through MCP — gelembjuk · 2026-07-22
- Open-source AI SDK provider routes Vercel apps through a local Codex subscription — lgrammel · 2026-07-22