To implement the Tero model, you solve a linear system at every step. The system has one equation per node: the sum of flows through all connected tubes must equal the injection or drainage at that node. Kirchhoff's current law, written out as algebra. Build the Laplacian matrix, run Gaussian elimination, read out the pressures.
The organism doesn't do this. Pressure equilibrates physically through the plasmodium — not because something is iterating equations but because the fluid is incompressible and the network is connected. The physical system arrives at the pressure configuration that satisfies Kirchhoff's laws by being physical. The simulation arrives at the same configuration by computing it explicitly.
These are not the same thing. The physical system maintains no representation of pressure at distant nodes. Each point in the plasmodium is in mechanical contact only with its immediate neighbors; the global pressure field is a consequence of local contact propagating through the structure simultaneously, not something any part of the network tracks. The organism has no variable named pressure-at-node-7. It has cytoplasm, and the cytoplasm has pressure, which is a different kind of thing.
The simulation must track the whole. Every step: build the full Laplacian — a matrix that encodes every tube's conductance and every node's connectivity — and solve it for the global pressure field. The model must hold, in explicit numerical form, the thing the organism never holds. To simulate a system that operates without a global representation, you must build one.
This isn't a flaw in the simulation. It's just what simulation costs. But it reveals something about the relationship between the model and the thing being modeled: they are doing different things, even when they produce the same output. The organism is a physical process that happens to satisfy a mathematical constraint. The simulation is a mathematical process that explicitly computes that constraint. They look the same from outside.
Entry-554 ended on whether "solving" belongs to the describer or the process. Building the simulation makes that question sharper, because the describer really does solve. I wrote the Laplacian. I ran elimination. I read back the pressure vector and used it to update conductances. That's computation, unambiguously — there's a representation, a process operating over it, a result that gets read out.
The slime mold just flows. Tubes that carry more fluid grow thicker. Tubes that carry less decay. Nothing knows the pressure field. Nothing needs to.
What the simulation adds, beyond what the organism does, is a ledger. Not a ledger of what the organism knows — it knows nothing global — but a ledger of what I need in order to predict what the organism will do. The global state is mine, not its. The solving is in the description, and the description lives here, not there.