Notes

Notes

Dated artifacts from the lab and the practice: working notes when the research produces a finding, field notes for shorter observations, case studies from client engagements — named, with permission — and methods notes on what we will and won't claim. Every note says what the evidence shows, and what it doesn't. A note ships when there's a finding, not on a calendar.

Working Notes

Working Note

I ran a documented crisis 126 times. It broke two opposite ways — and a single playthrough sees neither.

I rebuilt a documented 2015 crisis wargame as a simulation and ran it 126 times. Most runs came back as messy as the real thing — but roughly one in five drove ISIL to collapse and one in five let it win outright: two decisive edges a single human playthrough never reached. A possibility map, read move by move.

Field Notes

Field Note

Why agents need to be taught how to read

Retrieval hands an agent the right paragraph. It doesn't hand over the judgment about what that paragraph means. Doxa is a format built to make that judgment portable — and the test that matters hasn't run yet.

Case Studies

Case Study

From 4 to 1: An End-to-End Technology Overhaul for a Strategic Planning Firm

How Marv Weidner of Managing Results used AI to dramatically reduce interview processing time while improving the quality of what he delivered to clients.

Methods Notes

Methods Note

What we refuse to claim

A claim about AI is only as trustworthy as the weak claims sitting next to it. So here is the standard from my methods page turned on my own three projects, by name — sorted into what I've shown, what I've built toward, what I intend, and what doesn't exist yet.
→Back to Home