Writing.

Thoughts on building, learning, and the future.
Essays, notes, and reflections.


2026

Jun 20

Can Agents Learn to Look Trustworthy?

About reputation, relationships, and the gap between who we are and who we appear to be.

7 min read
Jun 03

Sycophancy and the Optimization of Approval

About people-pleasing machines, uncomfortable truths, and the price of agreement.

6 min read
May 18

Reward Hacking: When Optimising the Wrong Objective Works Too Well

About loopholes, unintended incentives, and the surprising creativity of goal-seeking systems.

8 min read
Apr 27

Trust as a Proxy Objective in Multi-Agent Systems

About trust, credit scores, five-star ratings, and the metrics that quietly run the world.

9 min read