Future Shock essays
The durable pieces, pulled out of the newsletter pile.
The daily archive moves fast. This page is the calmer shelf: essays worth reading after the news cycle has moved on.
Series
The Shape of the Next Decade
A running map of how AI changes coordination, labor, identity, rehearsal, and the boring systems underneath the spectacle.
The Race to AGI
Why the race frame produces worse incentives than the models themselves.
The Scaffolding Is the Intelligence
The systems around models are becoming the real capability surface.
The Boring Transformation
AI change arrives through dull workflow plumbing before it looks like science fiction.
The Belonging Gap
What happens when institutions automate contact faster than they rebuild belonging.
The Rehearsal Problem
Cheap simulation makes consequences visible without making choices easy.
Systems
Agents, swarms, and coordination
The useful agent stories are not about one brilliant chatbot. They are about handoffs, conflict, memory, and authority.
When Your Swarm Disagrees
Multi-agent systems fail in the space between competent parts.
What If Your Personal Devices Formed an Agent Swarm?
The personal-agent future starts with devices negotiating around you.
The Primer Prediction
A sci-fi lens for agent handoffs, memory, and autonomy that actually works.
Dead by April, Part 2: The Agent That Can't Agree With Itself
A concrete look at internal disagreement as an agent failure mode.
Reality check
Infrastructure and power
AI still cashes out in power, chips, water, law, procurement, and management. The abstractions have addresses.
The Cloud Has a Physical Address
AI infrastructure is land, power, water, jurisdiction, and bottlenecks.
AI Didn't Replace Workers. It Outran Their Managers.
The management layer often breaks before the labor layer disappears.
Building Moats, Not Bridges
What enterprise AI reveals about access, lock-in, and institutional incentives.
The Ban-Build Cycle: Enterprise vs. Agentic AI
Organizations ban what they cannot govern, then quietly rebuild it inside the walls.
Control surfaces
Security, governance, and accountability
The interesting safety question is increasingly mundane: who can authorize what, under which receipt, with what rollback path?
The Security Industry Is Solving the Wrong Problem
Agentic systems need authority boundaries, not just better risk language.
The Suppression Problem
What gets hidden when systems optimize for acceptable outputs.
Promise Nothing, Claim Everything
A clean read on AI marketing, responsibility, and the gap between demos and obligations.
Palantir, Privacy, and Rights
The surveillance story is less cinematic and more administrative than people want.