Artifact: Cellular Space Structure. The tile anatomy and the field-population model are in there.
What changed in my thinking
After the puck, I looked at the other end of the problem: not the hands, but what the hands would handle. The cellular tile is my first attempt at an invariant worth mass-producing — edge rails, corner nodes, cross braces, power and data passthroughs — with every function demoted to content: solar, battery, radiator, antenna, controller, or nothing.
The argument that mattered to me was not mechanical but procedural: qualify the frame once. Contents can then evolve, fail, or upgrade without re-qualifying the structure that carries them. This is the first appearance of a rule that will keep returning in different clothes — in the rack’s permanent rails, in the PV field’s replaceable tiles, in the pod’s refusal to merge failure domains. I did not know that yet. That is what a ledger is for.
One placement decision is worth recording: non-solar functions spread outward from the center rather than clustering. Critical infrastructure should not have a center of gravity.
Carried forward, unresolved
- The 1.35 m tile size is asserted, not derived. It will later be replaced by a dimension earned from an energy ledger, and I will not mourn it.
- Connector standards, extraction envelopes, and who exactly installs a content tile: undefined. The puck and the tile have not yet met.
Recalled
- Look to Windward (Iain M. Banks, 2000). The Culture’s General Systems Vehicles are the fantasy version of my tile: a standardized, immensely capable frame hosting arbitrary contents — populations, factories, fleets — while the hull and its systems endure. Where Banks is wrong for my case: a GSV works because its frame is a genius. The Culture can afford monolithic intelligence at the structural layer because Culture manufacturing is post-scarcity. I have a launch ledger and a build script, so my invariant must be dumb: mass-producible, inspectable, and incapable of resenting its contents.