Hub readiness scorecard
Model v0.2Fifteen questions across the five layers of the model. They are not a test you might fail: they are the standard a hub is measured against, published so it can be argued with before anyone is held to it. Answer them, generate the reading, and keep it. There is nothing here to request and nothing to send.
Every criterion scores 0 (no), 1 (partial), or 2 (yes), thirty points total. The four bands are unchanged from v0.1: under 40% Not ready, 40–64% Foundational, 65–84% Certifiable, 85% and above Routable. The weakest layer is still called out, because readiness is limited by the weakest layer, not the average.
v0.2 adds a reading of the same fifteen answers, and the whole rule is here so it can be done by hand. Facility plus people is the floor, F, out of 12: whether this place can make things, with people who know how. Factory OS plus quality is the system, S, out of 12: whether what happens on the floor is versioned, measured and auditable. Cells and tooling is the hinge, C, out of 6: whether the floor can change shape. The gap D is F minus S, and the gap rather than the sum is what characterises a hub.
No threshold is invented; the published band edges are mapped onto each axis. On a 12-point axis, 65% rounds up to 8 and 40% to 5, so F and S are high at 8 and low at 4 or less. On the 6-point hinge, high is 4 and low is 2 or less. On one 6-point layer, 3 is the Foundational edge and 4 the Certifiable one. The halves diverge at a gap of 5, the smallest whole-point gap at or above the 40% edge of a 12-point axis.
Branches are tested in order, first match wins: exactly one layer below 3, the uneven build; D of 5 or more with F at 8 or more, the undocumented shop; the same divergence toward the system, the managed plant; both halves at 8, the systematized floor; both at 4 or less, the early floor; two or more layers below 3 with one at 4 or more, the spiky profile; then the remainder, the partial system, which is unguarded, so all 14,348,907 answer sets land in exactly one branch. A hinge clause is added from C alone. Weighting every answer set evenly, those seven fire at 29, 3, 3, 3, 5, 43 and 13 percent: no hub has ever been assessed, so that is a property of the arithmetic, not of a cohort.
Model v0.2, 2026-08-23. The band edges are untouched, so a link shared under v0.1 still lands in the same band and the result says it was shared under the old rules. Two band descriptions were rewritten because they promised what the model cannot deliver: one said certification was within reach, with no certification body in existence, and one called capacity routable, with zero routing decisions ever made. The reading, the layer bars and the commitment column are new.
Still a deliberately simple draft, published so it can be corrected in the Stage 0 expert interviews. Weights are flat on purpose. The floor and system split is a design choice, not a derivation, and it is arguable both ways: cells could be argued onto the floor, and a repeatable training program is documentation, so people could be argued onto the system. The thresholds are mapped from the band edges and have never met a real hub. Whether the gaps should also be split by what closing them costs, discipline against capital, is a v0.3 question. Corrections: shk@moduloa.com.
This tool answers one half of the question: whether a site can run a given tier. The other half is which tier a product actually needs, which is what the production tier calculator asks, from five questions about the product. Same v0.1 caution, opposite direction.
To register interest, or to say that a hub should be assessed properly one day, say so to the founder. The link lands on /contact, which states the address in prose and offers a mail already addressed and titled for this tool. There are no certified hubs yet and no certification programme to apply to, so this is a conversation, not an application. The assessment does not travel with the link: press Copy assessment link above and paste it in if you want it to.