Agents raise the effective threshold for one-way-door decisions, says astronomer mjuric
mjuric · x · 2026-10-11
mjuric observes that the effective threshold for one-way-door decisions is far higher when working with agents: very few choices can't be walked back within a few hours. The implication is that agent-era decision-making can afford to be bolder and iterate faster.
More from coding & agent
- Claude Opus Rebuilt a Tesla Roadster Entirely in Code, Rendered with three.js — jasonkneen · 2026-10-11
- LoreKit: open-source deterministic memory layer for coding agents, no vector index — madsthines · 2026-10-11
- Devs push back on AI slop panic: codebases have always contained human-made slop — ctjlewis · 2026-10-11
- Shrike: open-source MCP server that gives Claude Code a recording of your app's crash — S_m_nt · 2026-10-11
- Grok agent gets its own email and proactivity, winning back skeptical users — Baconbrix · 2026-10-11
- Zed CEO's dashboard shows 295M tokens burned by 2 devs in 30 days, mostly cache reads — zeeg · 2026-10-11