What is a stabilisation stall?

A stall is a behavioural substitution — a cluster doing something observable (X) instead of something harder (Y). Stalls are locally rational responses to incentives, not failures. There are 9 canonical stall types identified across 75+ diagnostics.

What is a stabilisation stack?

A stack forms when two or more stalls mutually reinforce each other. Once a stack forms, removing a single stall is absorbed — the stack compensates. Stack analysis is why single interventions consistently fail to shift cluster regimes.

What are the 9 canonical stalls?

Re-proving instead of narrowing; Coordinating instead of deciding; Forgiving instead of redesigning; Extracting without reinvesting; Mediating instead of coupling; Stabilising around incumbents; Narrating instead of testing; Scaling activity instead of throughput; Waiting for permission.

How does the ClusterOS diagnostic work?

The diagnostic runs a 5-stage AI pipeline: evidence collection (50–70 items), pattern detection, stall identification, stack analysis, and leverage hypothesis generation. It combines public data with actor questionnaire responses for higher confidence stall identification.

The Diagnostic

The substrate flows freely.
Until it doesn't.

On a healthy substrate, every actor's selfish action feeds the system. Signals travel. Value compounds. Connections form without anyone managing them.

Then the system stabilises. Actors adopt locally rational behaviours that prevent compounding. Not because they fail — because they're responding sensibly to the incentives they face.

These stabilisations are called stalls. When stalls reinforce each other, they form stacks. Stacks are why your cluster works hard and goes nowhere.

See the 9 stalls Build a stack →

Complex adaptive systems

The system isn't broken.
It's stabilising.

"Every behaviour makes sense from the actor's perspective. Stalls are not failures — they are the system finding equilibria. Understanding them is the precondition for changing them."

ClusterOS Diagnostic Framework

Actors are locally rationalEvery behaviour that looks like dysfunction makes perfect sense from inside the actor's constraints. Stalls are sensible responses, not mistakes.

The system learns what reduces discomfortCoordination reduces friction, so the system coordinates more. Narrative restores confidence, so the system narrates more. Each stall is a learned response that became a default.

Stalls serve interests, not just learningOnce a stall entangles with roles and reputations, it is actively defended. Removing it doesn't just change behaviour — it removes someone's position.

Leverage means small perturbations that shift regimesNot large programmes. The goal is identifying the minimal intervention that changes the operating conditions — not fixing everything at once.

Stage 3 · Stall detection

Nine ways a cluster
stops compounding

Each stall is a behavioural substitution — the system doing something observable (X) instead of something harder (Y). Click any stall to see the full pattern. Then add it to the stack builder.

On the stack builder below: Stalls rarely operate alone. Once you understand each one, add two or three to the stack builder — and see what kind of cluster that combination describes, why single interventions fail to break it, and what the diagnostic looks for.

Stage 4 · Stack analysis

Stacks are why single fixes fail.
Build one and see why.

A stack forms when one stall lowers the cost of another. Once that happens, stabilisation becomes anticipatory — pressure enters, the same configuration activates, and the system doesn't consciously choose the response. It's already structured to produce it.

Select 2–3 stalls from the cards above (use the "Add to stack" button), or click any stall name below.

Stack composition

0 / 3 stalls

↑ Add stalls from the grid above

Select a stall...

Select a stall... (optional)

Quick pick

⬡

Add at least 2 stalls to see
the stack analysis and reinforcement logic.

From 75 diagnostics

Stacks we've seen
in the field

These three configurations appear repeatedly across geographies and sectors. They have different names in different ecosystems — but the reinforcement logic is the same. Recognising your stack is the first step to targeting it.

Narrative × Activity

S7 Narrating S8 Scaling Activity

External scrutiny rises. The system responds with strategy documents and programme launches. Each legitimises the other. The ecosystem becomes excellent at demonstrating vitality — while deferring the learning that would collapse choice.

Seen in: Science clusters, innovation districts

"More programmes producing more activity — but what's actually converting?"

Coordination × Mediation

S2 Coordinating S5 Mediating

Alignment mechanisms create demand for intermediaries. Intermediaries make alignment manageable. Friction is absorbed through process rather than forcing decisions that would concentrate responsibility. The system coheres but never narrows.

Seen in: Manufacturing clusters, regional partnerships

"Everybody's coordinating. Nobody's deciding."

Incumbent × Extraction

S6 Stabilising S4 Extracting

Innovation pathways route through large incumbents, which absorb and filter demand signals. New entrants that can't attach to the incumbent orbit find their value (talent, capital, companies) exits the ecosystem rather than reinvesting.

Seen in: Defence, aerospace, energy clusters

"We scale to Series A, then they leave for London."

Stage 5 · Leverage

Your cluster has a stack.
The diagnostic names it.

The ClusterOS diagnostic runs evidence through all five stages. It doesn't score your cluster or rank it. It identifies the specific behavioural substitutions operating in your ecosystem — and the stack they form. That's the starting point for interventions that actually shift regimes.

Request a diagnostic ← Back to The Substrate

Testable leverage hypotheses · Calibrated to your specific configuration

The substrate flows freely.Until it doesn't.

The system isn't broken.It's stabilising.

Nine ways a clusterstops compounding

Stacks are why single fixes fail.Build one and see why.

Stacks we've seenin the field