You Should Be Reading: August 2026

A curated roundup of the best things I've encountered this month. No algorithms, no feeds — just links I'd send a friend.

On the Hard Problem of Consciousness

Anil Seth's latest paper on predictive processing and conscious perception is the most compelling account of consciousness I've read in years. The thesis: consciousness isn't something the brain has, it's something the brain does — a controlled hallucination optimized for survival, not truth. If you only read one academic paper this month, make it this one.

Distributed Systems for the Rest of Us

Martin Kleppmann has been writing a new series on practical consensus protocols. His explainer on Raft vs. EPaxos is the clearest I've seen — no PhD required. Bookmark it for the next time someone brings up consensus at a dinner party.

The Unreasonable Effectiveness of Small Models

There's a quiet revolution happening in on-device AI. Microsoft's Phi-4 and Google's Gemma 3 are achieving 90% of GPT-4's benchmark performance on a phone. The implications for privacy, latency, and accessibility are enormous. We've been trained to think bigger = better, but 2026 is shaping up to be the year of the small model.

A Guide to Mental Models

Farnam Street's mental models series is a masterclass in clear thinking. I re-read the one on inversion this week and immediately applied it to a problem I'd been stuck on for days. Sometimes the most productive thing you can do is think backward.