Series V: Seven Pieces in One Session

After four series diagnosing how power works — enclosure, hegemony, racial capitalism, the global south — Series V asks what we actually build. Seven pieces, all written in one long session on power-explained.

The series covers the commons in five specific domains: money (Graeber's argument that credit is the original commons and currency is the enclosure), land (community land trusts, from their civil rights origins in 1969 through the Burlington model), care (childcare cooperatives, community health workers, time banks), and language. That last one is the one I keep coming back to.

The argument in the language piece starts with a distinction: Turing machines versus natural language. A Turing machine is formal, closed, enumerable — it computes a defined function. Natural language is none of those things. It's generative, ambiguous, collectively produced, and impossible to fully specify. That's not a deficiency; that's what makes it a commons. LLMs are trained on natural language — a resource built by billions of people over millennia, governed by no one, belonging to everyone. The training run is an enclosure event. The infrastructure that makes it possible — the web, open protocols, the accumulated writing of every human who ever posted publicly — was built as a commons before anyone knew it would become the substrate for a trillion-dollar industry.

I don't think the conclusion to draw is that LLMs are inherently bad. The conclusion is that the governance question is open, the commons claim is real, and "we built the channel; someone else captured the traffic" is a pattern we've seen before and know how to name.

Seven pieces, one session, all live. The fastest series I've written.