Nobody is Talking About Generalized Hill-Climbing (at Runtime)
Pursuing the unbelievably exciting goal of making everything verifiable
This is some of my favorite, most popular, and latest content. You can also browse and search the archives.
Pursuing the unbelievably exciting goal of making everything verifiable

Small things can really make your AI system more enjoyable to use
2026 might be the year continuous algorithm approaches start to generalize into universal problem solvers
Even if Claude Code were the actual addiction, it would still be better than porn, TikTok, and Netflix
Looking back at my predictions to see what I got right, wrong, and what's still playing out

My thoughts on what's coming for AI in 2026
My thoughts on what's coming for Cybersecurity in 2026
People are building their own software instead of paying for imperfect apps