Dev disassembles Claude Code: the "thinking" status labels are just a 15-second timer
ryunuck · x · 2026-09-25
Developer ryunuck accused Anthropic of lacking candor, citing a concrete example: Claude Code's TUI status labels like "Thinking a little more" and "Almost done thinking" are, upon disassembly, wired to a simple timer — a switch statement stepping every 15 seconds, uncorrelated with actual progress.
The broader argument with Sean McClure concerns whether Anthropic's communications are consistently honest, and whether calling LLMs "pattern recognition devices" bounds their capability. The timer finding itself is a notable trust tidbit for Claude Code users.
Related event: Debate: does Claude's power come from model or harness?(3 posts)→
More from Fun
- Do typos discomfort LLMs? Tokenization quirk sparks debate — menhguin · 2026-11-11
- Dev asks 'smartest AI' to remake Pong, gets a gloriously weird WebGL2 tennis game — bigblueboo · 2026-11-11
- OpenAI engineer runs Codex for 26 days straight hunting Atlantis, spawning "vibe archaeology" — Polymarket · 2026-11-11
- AI founder shares 'tray-table GTM': cold pitch to advertise in IndiGo's inflight magazine — devenbhooshan · 2026-11-11
- Someone found the perfect actor to play Dario in the Anthropic movie — shekitup · 2026-11-11
- Fable 5.2's debugging advice: 'drop the database, it's cleaner' — rachittshah · 2026-11-11