Topic
Systems
10 essays and 7 notes, newest first.
Essays · 10
The Model Is a Centroid Pump. So Is the Review Loop.
Sameness comes from two places: training narrows what a model will produce, and a review loop narrows what survives. I've built something against both and don't yet know if either holds.
The Org Chart Is the Missing Constraint
Most of the new advice on running coding agents validates an older idea: constraints do the work. At multi-agent scale, the constraint that matters most is who holds which decision rights.
The Oracle Problem
Everyone wants to be the source of truth; almost nobody builds the conditions that let truth survive. From the Delphic Oracle to Grokipedia, the failure is structural, whatever the intent.
Defaults Are Decisions
Every system has defaults, and AI makes them more powerful and less visible. So ask whose assumptions your system is running on.
What Are Test Scores Actually Worth?
I built a school data product for parents. What I built, how my thinking evolved, what an axiom actually is, and where the data runs out.
Your Infrastructure Spec Already Moved Into Your Code
Infrastructure config is collapsing into application code, as build tooling did a decade ago. Void shows where that ends; the choice is whether to ride the platform or steal the pattern.
Your Vault, Your Rules: Password Managers, Sovereignty, and Agents
On the quiet ethos connecting Buttercup, Enpass, pass, and VeraCrypt — and why it matters more now that AI agents need your credentials too.
Pervasive AI: What Happens When Your Assistant Never Logs Off
I ran a personal AI agent on a Mac Mini for a month and wrote down what I saw. Seven months on: what held up, what didn't, and what a Meta launch taught me about architecture and trust.
Why AI Can't Shop for You Yet
The properties that matter most in fashion aren't properties of the product — they're properties of the relationship between the product and the person. No protocol fixes that.
The SaaSpocalypse
Anthropic wiped $285 billion off software stocks in a week. The market fears AI replacing tools; the bigger disruption is what it means for the people who build them.
Notes · 7
Everything New Is Old
Vector search in the 90s, badges that followed you through the office, and what happens when you build an AI system without knowing what's already been built.
The Dopamine Layer
A neuromarketing talk got me thinking about reward systems, GLP-1s, and what happens when AI becomes the most sophisticated rationalizer we've ever built.
Kipple at Scale
Large engineering orgs already solved coordination for thousands of humans. How much transfers to thousands of agents — and where does entropy win?
The Fork in the Stack
The agent era is splitting infrastructure into two camps: those building new primitives and those retrofitting old ones. The bet you make reveals what you think agents are.
Named Things: A Glossary of Compressed Wisdom
Named laws, razors and principles that keep showing up in engineering and strategy. Each compresses a concept into a name: human-language embeddings, built over centuries.
The Frontend Testing Gap
Backend testing is a solved problem. Frontend testing for stateful, streaming, visual UIs is not. That's exactly where AI coding agents are generating the most code.
Toward AI-Native Analytics for Personal Publishing
An RFC for a small, legible analytics system — privacy-first, edge-native, AI-queryable — built on Cloudflare Workers.