bert-compose: "Troncale mode" — a meta-lens where Systems Processes are the template

[rsthornton]

The idea (Shingai, this session)

Today's lens packs (commit 7954426) rename Mobus primitives in domain vocabulary (crypto/gov/neuro/ecology). The next move: a mode switch where Troncale's Systems Processes themselves become the template — not a domain skin, but a different organizing vocabulary for the whole tool.

Imagine flipping a switch and suddenly:

• The palette is no longer "Buffering / Modulating / Sensing" but Troncale's process families (Flows, Cycling, Feedback, Networks, Potential Fields, Decay…).
• Picking a Troncale process drops in the primitive circuit that demonstrates it — the sweep rungs (src/sweep.rs / troncale-sweep/) become the building blocks. Compose at the level of systems processes, not atomic work-processes.
• The examples menu becomes the machine-demonstrated Linkage Propositions; building a model = chaining LPs ("coupled feedback → oscillation") that each expand to a runnable circuit.

Why it's more than a lens

A domain lens is a renaming (a functor on labels). A Troncale mode is a change of basis: it composes at the SP layer (Mobus = process physiology HOW; Troncale = the process patterns), with each SP backed by the verified primitive circuit beneath it. The sweep is the bridge that makes this possible — every Troncale process now has a constructive primitive realization, so the mode can expand an SP into its circuit deterministically.

Dependencies / prior art

• Built on the Troncale sweep (this session): the SP→primitive-circuit mapping, machine-demonstrated.
• Composes with lens packs (bert-compose: export the four-lens homeostat as a publishable artifact bundle #83/bert-compose: lenses auto-apply domain substances (lens suggests, user overrides) #84): a domain lens × Troncale mode = "this governance quorum IS a Feedback SP."
• Connects to the Lean bridge: if SPs are derivable from the K≅2 kernel + flow/storage, the mode is the UI over that theorem.
• The boundary buckets matter: only (a)-constructible SPs get a template; (c)/(d) ones are shown as "needs structure / needs agents."

Parked as a horizon idea, not next-up. Food for thought after the sweep + lens followups (#83/#84/#85).

Context: lens packs 7954426; Troncale sweep this session; bert#81 (cross-model inference) is the inference-layer cousin.