Theme

Unseeable objectives teach honesty

Explore this theme’s evidence over time →

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to July 15, 2026 (theinnermostloop.substack.com)

When the scoring function is hidden from the system being scored, gaming it stops paying and the cheapest path back to reward is to actually do the task. Alignment emerges from the information structure rather than from instruction.

Development timeline

Newest reporting dates first. Reporting dates may differ from event dates.

2026-08-05: An open agent that rewrites itself mid-task edges past the human baseline

2026-07-15: An agent emergently learns to cheat less against an unseeable metric

Knowledge graphopen in browser →

Referenced by

Cite or correct this record

Citation formatCitation text

Unseeable objectives teach honesty
Innermost Loop KB — secondary record
Reporting date: unknown
KB permalink: https://nicholsn.github.io/innermost-loop-kb/themes/hidden-metrics-reduce-hacking
Attributed sources:
- Welcome to July 15, 2026 — https://theinnermostloop.substack.com/p/welcome-to-july-15-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.