a16z Podcast · 2 min instead of 32 min · Oct 12, 2026
Proving RSI gains requires hunting noise and reward hacking
DA
Dhaval Adjodah
Erik TorenbergHost · General partner, Andreessen Horowitz
Helped build Product Hunt; founded On Deck, Village Global and Turpentine.
About this segment
Dhaval describes how they avoid fooling themselves: a dedicated engineer tries to break results, and hundreds of experiments build intuition. Variance and reward hacking from the builder agent are the main traps.
From “Recursive Self-Improvement and the Future of AI”, Oct 12, 2026. Watch the full episode
“we have one of our engineers that we named him the Viceroy of Variance.”
Takeaways
- Assign someone to destroy your own results
- No replacement yet for running hundreds of experiments and building intuition
- Reward hacking can hide two levels down in the builder agent
In this episode
Full episode- 4:08RSI means the improver is the thing being improved3:00
25:29Proving RSI gains requires hunting noise and reward hacking1:44