Notes from the desk
Short research notes, methods, and signals. Updated when something is worth saying — not on a content calendar.
FOLLOW THE DESK
New notes land here first
Essays, methods, the occasional map — published here, fragments on X. Follow along where you already are.
N·011IN DRAFT
The half-life of a launch claim
When does a public claim stop being believed?
NOTEN·009↗
Color as evidence
Every color is a claim: verified, contested, unverified.
METHODN·008IN DRAFT
The empty center
Why one missing answer corrupts a dozen others.
ESSAYN·007↗
Signal weather, week 24
This fortnight's sanitized public-signal risks.
SIGNALN·006IN DRAFT
Scoring trust without a survey
How we rate risk from public signal alone.
METHODN·005IN DRAFT
What LLMs say about you
Your docs are training data. Read them that way.
NOTEN·004↗
Confusion is a map, not a mess
The idea the whole desk is built on.
METHODN·002↗
Signal weather, week 22
Source contamination was the theme.
SIGNAL