Observation 01

The Absorption Problem

Time from introduction to economic rebalancing

Every technology wave displaced workers. Every economy eventually absorbed the shock. But the window keeps shrinking — and the adjustment mechanisms haven’t gotten faster. The absorption trend is clear.

Click any row to explore · Drag the slider to test assumptions

How long will AI absorption take? 3 years
1 year 25 years

Observation 02

The Geometry of Knowledge

The sphere of knowledge is an old idea, dating back to Plato. The geometric insight — as our knowledge grows, so does our awareness of what we don’t know — is often attributed to Einstein. What’s interesting is when you apply the geometry of knowledge to network architecture.

Nodes in the network 6
There are effectively infinite domains of knowledge. How much frontier surface area does each architecture expose for exploration?
Distributed AI model
Frontier surface area
Exploration surfaces
6 nodes

Each domain or node creates additional frontier surface area for knowledge exploration. Reduces dependency with each node added.

Centralized AI model
Frontier surface area
Exploration surfaces
1 hub

The exploration surface is internal to the provider. Creates dependency with each query.

The Implication

When you become the knowledge surface

In the distributed AI model, each node’s frontier surface area faces outward — toward undiscovered knowledge. The node explores, the node discovers, the node keeps the map.

In the centralized AI model, the exploration surface is internal to the provider. Its users provide the knowledge surface area for the hub owner to explore.

Distributed AI model

Your queries expand your own frontier. Each node maintains sovereignty over its exploration surface. The network gets smarter. So do you.

Centralized AI model

Every query you send gives the provider new surfaces to explore. Specifically:

What you asked
What you don’t know
What you’re researching
How you think

Your areas of ignorance become the provider’s new knowledge surfaces.

We watched what happened when a handful of companies captured the social graph. We watched what happened when they captured search intent. Centralized AI captures the cognitive frontier itself — the boundary between what people know and what they’re trying to figure out.

The most valuable thing about a thinking person isn’t what they know. It’s what they’re trying to figure out. In the centralized AI model, that signal — the surface area of your ignorance — routes to a single owner. And unlike social media, there’s no way to scroll past it. This is the architecture of thought, and right now, four companies are building it as a star graph.

Sources & further reading
Knowledge sphere geometry: Gleiser, M. The Island of Knowledge (Basic Books, 2014). The geometric formalization of knowledge as volume and awareness-of-ignorance as surface area. The insight dates to Plato; the quote “as our circle of knowledge expands, so does the circumference of darkness surrounding it” is widely attributed to Einstein but unverified.
Network topology & scaling laws: Reed, D.P. “The Law of the Pack” Harvard Business Review (Feb 2001). Metcalfe’s Law scales as n²; Reed’s Law shows group-forming networks scale as 2n. Star topologies collapse both to n×1.
End-to-end argument: Saltzer, J.H., Reed, D.P., Clark, D.D. “End-to-End Arguments in System Design” ACM Transactions on Computer Systems 2(4), Nov 1984. The architectural proof that intelligence at endpoints structurally outperforms intelligence in the network core.
Technology absorption: Brynjolfsson, E. & McAfee, A. The Second Machine Age (W.W. Norton, 2014). Perez, C. Technological Revolutions and Financial Capital (Edward Elgar, 2002). Historical absorption timelines and institutional adaptation lag.
Elite overproduction: Turchin, P. End Times: Elites, Counter-Elites, and the Path of Political Disintegration (Penguin, 2023). Credential surplus as predictor of societal instability across ten millennia of data.
Confidential Preview
Material beyond this point is shared
under NDA with invited reviewers.
Press Enter to continue