Crossing at a ford — watercolor

Cross at a Ford

It means setting sail even though your friends stay in harbour, knowing the route, knowing the soundness of your ship and the favour of the day. — The Book of Five Rings

A ford is a moment: a route that is briefly legible and passable, before the water moves again. Facing complexity is learning to see those moments — and having the nerve and the instruments to cross.

Complex systems: what modulates behavior?

Not the org chart. Underneath every formal structure is a living network of trust, attention, and obligation — and that is what actually decides what happens next. It can be mapped, measured, and worked with directly.

A schematic networkENS airdrop network — a governance structure read from data
Finding order in the network of interactions that surrounds us — the schematic, and the real thing: 84,000 ENS wallets arranged by delegation and flow.

Structure hides in the data

A governance structure nobody drew — communities, brokers, and whales — read directly out of transaction data. The same genre of prototype has opened government-scale geospatial engagements.

Current study: narrative geography — mapping ballot-measure affinity across San Francisco and New York, asking of each neighborhood, would this story land here?

The fight is integrity

Confidentiality is about secrets. Availability is about uptime. Integrity is about correctness. Correctness enables trust.
— Bruce Schneier
1990s — 2000sOpen protocolsthe fight was availability2010sPlatformsthe fight was attention2020s —AI systemsthe fight is integrity1990s — 2000sOpen protocolsthe fight was availability2010sPlatformsthe fight was attention2020s —AI systemsthe fight is integrity

Most modern attacks against AI systems are integrity attacks — manipulated training data, inputs, outputs, feedback. Years of building investigation platforms inside security operations centers taught us what it takes to reconstruct why a decision was made and who was accountable. We bring that discipline to AI systems: audit trails, reasoning traces, and harness designs that keep human judgment as the grounding layer.

Where we help

Your data should be answering questions, and it isn't.Pipelines, analytics, and visualization that hold up at scale — a practice that started on YouTube’s executive metrics.
You are putting AI into workflows where trust matters.Agentic harness design, evaluation frameworks, and audit trails for systems that must stay in integrity.
Your organization faces a question models can’t settle.Complexity facilitation and narrative mapping — instruments for seeing where you actually are, before committing to cross.