STMC draws the terrain: the binding constraint and the five-node dependency map. Path Painter reads it, generates and structurally stress-tests routes through it, and commits to one with eyes open about what committing forecloses. A fatal assumption at pre-mortem doesn't just get logged — it sends the route itself back for revision, which is what makes this a circuit rather than a pipeline. Hover any stage to read it, or play the loop to watch a route travel from terrain to committed decision — clean the first time, revised the second.
[ increasing returns › learning › coordination › expectations › sunk cost ]
The same content as the interactive circuit above, written out for reading straight through — this is also what the captions above are built from.
Path Painter reads two things from STMC's completed circuit: the binding constraint — the domain most likely to prevent landing — and the five-node dependency map itself: technology, IP, partner, investor, market. Everything downstream is a reading of this terrain, not a fresh survey of it. Read the full circuit at STMC.
Generates complete strategic narratives, not branches — each filtered for suitability, feasibility, acceptability, distinguishability and completeness. Feasibility doubles as a silent ruin gate: any route that cannot reach the next decisive point before runway runs out is discarded here, before it costs another minute of analysis. Produces a short list of feasible, genuinely distinct routes.
A matrix of nodes and their dependencies, from Steward and Eppinger's engineering work. One baseline is built once, from STMC's five-node map; each surviving route records only its delta against it — edges added, removed, or reversed. Reordering the matrix reveals coupling: which nodes depend on each other tightly enough that a failure in one propagates through the whole block.
Identifies which changed edges carry a reinforcing loop — a self-feeding commitment where taking the edge makes leaving it progressively harder — and reads what drives it: four sources of increasing returns from the path-dependence literature — learning effects, coordination effects, adaptive expectations, and sunk cost.
Reads each identified loop three ways: polarity (compounding an asset, or digging a hole), stickiness (how fast reversal cost climbs), and door type — one-way or two-way (Bezos). A one-way door is not a separate thing from a sticky edge — it is a sticky edge whose loop has finished winding, with reversal cost effectively infinite. Produces a structural profile per route: coupling, loop polarity, and door status.
Assumes the route has already failed and works backward through its worst-coupled dependency cluster to find why — Gary Klein's method. Red team joins only if STMC's IPB names a genuinely reactive adversary. Any fatal assumption surfaced here sends the route back to COA for revision or elimination — the loop that makes Path Painter a circuit, not a pipeline.
Runs no new procedure — only a reading of what the prior stages already produced. Each one-way door is severity-rated, trivial to existential; critical and existential closures get a second-order question: what happens downstream once this option is permanently gone. When every surviving route forecloses something, that is the operating condition, not a failure.
The chosen route is committed to deliberately, with a signed pre-commitment record of which doors close, why this route was still the right trade, and what would have made an alternative viable. Its thresholds become the trigger conditions STMC's Monitor layer watches for — the same dependency graph, reused as a surveillance scaffold rather than rebuilt.