What Rules Can't Capture
Writing 247 rules made my information system worse. Some judgments can be specified precisely; others need interpretation. Working with AI helps me test where that boundary lies.
Writing
I write long essays on how organizations work, and shorter notes on AI, cognition, and what I’m learning along the way.
Essays
The Right Direction
Markets coordinate within rules they cannot choose. Better measurement sharpens the question of which objectives matter, without settling it.
Strategic Time
Strategy needs time to pursue a commitment and learn whether it works. Organizations must protect that time from both rapid market signals and pressure to survive at any cost.
The Architecture of Commitment
Organizations need stable purposes and methods that adapt to evidence. The question is which commitments to protect, which to revise, and what should trigger each change.
Notes
The Category Confusion
Climate change is a phenomenon, not an aspiration. Confusing the two obscures how it threatens objectives the military already holds.
strategyinstitutionsaspiration-problem-solution
Not Enough Problems
AI is making solutions cheaper. We still lack the institutions and practices that turn conditions worth addressing into questions we can act on.
AIeconomicsstrategycoordination
Where You Want Variance
As we get better at specifying things, the question shifts from 'can we reduce variance?' to 'should we?'
AIsystemsexpertisedesign
The Third Enlightenment
What would organizations look like if AI handled routine work and made more room for human judgment?
AIworkenlightenmenthuman flourishing
Agent Polities
What political philosophy can teach us about authority, disagreement, and adaptation in systems of AI agents.
AIorganization theoryagentsgovernance
The Dot Collector
On discovering that machines can be peers, and on the division of cognitive labor that makes partnership possible.
AIpartnershipcognitionpersonal
The Tacit Knowledge Problem
On the strange process of extracting what you know but cannot say, and why it requires watching more than asking.
AIcognitiontacit knowledgepersonal
Multiple Minds
On discovering that a single perspective—even a smart one—is never enough to see what you're actually thinking.
AIthinkingdebatecognition