In the current game, the rational move is to lie. We don't build better data pipes. We design better games — where truth-telling is the dominant strategy. Five pathologies. Three tells. One question that makes CFOs go quiet.
These aren't hypothetical. These are real conversations between different layers of an organisation — failing because each layer speaks a different geometric language. Read them and count how many you've had this week.
Diagnosis: S5 broadcast a SIEVE (quality-first filter) and a SLOT (who we are). S1 needed a STACK-compatible instruction — what happens next, in what order. The bridge works because S5 translated their filter into slot-priority language that operations can act on.
Diagnosis: S3 asked for a SACK (aggregate). S1 only produces STACKs (sequences) and SLOTs (status). The bridge works by asking in S1's language — what happens next? — then aggregating the STACKs into SACKs at the right level.
Diagnosis: S1 asked for a SLOT (yes/no). S3 answered with SACKs and SCALEs. The bridge: S1 asked for slots — green, red, open, closed — and got slots. Language operations can act on.
Diagnosis: S4 presented a SCALE (tension). S5 responded with a SIEVE (filter) and a SLOT (who we are). The bridge used Clean Language to discover the real shape of the filter. It wasn't "no recurring revenue" — it was "no exploitative lock-in." The policy was narrower than the word.
Diagnosis: S3 asked for a SACK (a number). S4 only produces SCALEs (tensions, comparisons). The bridge: S3 asked in S4's language — "what's the tension?" — then translated the scales back into sack-compatible numbers themselves.
The pattern is always the same: ask in the shape the listener can answer, then translate back to your own. That's what ZQL federation does automatically. These conversations are the human version of the same translation.
Every dysfunctional organisation falls into one of five patterns. Each has a signature, a tell, and a prognosis. You'll recognise yours.
"The iPhone didn't really do it."
"Poster child for not doing anything but what we do."
"I don't look up… don't look down."
"Go where the money is." Engages with market changes, plans multi-year scenarios.
But: "Could not be redeployed due to skill-location mismatch." 12,000 engineers eliminated — not because the skills didn't exist, but because there was no SOCKET between units.
Healthy. The game rewards surfacing.
Friction. Analysis paralysis before action.
Toxic. Honesty is punished.
Broken. Nobody can see.
This bypasses what leadership says about culture and tests what actually happens to information. The answer reveals the DOSE.
Using only public signals — earnings calls, Glassdoor, job boards, SEC filings — classify an organisation in under three hours.
Read the earnings call Q&A. Three questions determine whether the organisation can sense and respond to external reality.
Can assumptions be challenged?
How rigid is identity?
What's the time horizon?
Cross-reference Glassdoor, job boards, and crisis language. Four questions reveal whether S2 functions or is absent, captured, or toxic.
How was the last crisis resolved?
What happens to early warners?
How is capacity shared?
Stress-test every classification. A suspected Statesman with disclosed S2 reforms may be viable in transition. Monitor the next crisis.
| Tell | What to Listen For | Diagnosis |
|---|---|---|
| Response to Threat | Dismisses → Engages without internal mechanism → Engages AND shows response | Blind → Statesman → Viable |
| Crisis Language | "eliminated / restructured" → "found responsible" → "changed incentives / learned" | Statesman → Soviet → Viable |
| The Missing Topic | No coordination reforms → No dissent → No cross-unit collab → No scenarios | Statesman → Soviet → Darwinian → Blind |
You've seen the pathologies and the conversations that fail. Here's the structural model that explains them — and makes diagnosis systematic rather than anecdotal.
We're not building software. We're building a treaty framework. Shape defines assets. World defines state. Context defines which game. Dose defines the rules that make cooperation rational.
What geometry of answer? Assets on the board.
What epistemic status? State of play.
What type-specific rules? Which game?
Therapeutic or toxic? Rules of engagement.
Chaos is variety you refused to manage. Collapse is variety you tried to hide. ZQL surfaces variety at the rate the system can absorb it.
You've seen the conversations that fail, the archetypes that emerge, and the theory that explains them. Here's the good news: federation fixes this — without anyone giving up control.
ZQL is read-only. It queries systems where they sit. No data lake. No replication. No ETL pipeline. No political battle over who "owns" the data — because nobody moves it.
No hallucinations. No probabilistic guessing. Every answer traces back to the source system, the query that produced it, and the vocabulary that defined it. Auditors can follow the chain.
No rip-and-replace. No data migration project. ZQL federates across the systems you already have. First value in weeks because we're not rebuilding — we're translating.
No AI. No data lakes. Just the truth — traceable, auditable, and owned by the people who created it.
An AI answering "do we have 50 units?" gives you a Q¹ answer — a plausible-sounding fact with no provenance. Under the EU AI Act, that answer creates liability. Under DORA, it fails audit. Under real operational pressure, it hallucinates exactly when you can least afford it.
ZQL answers the same question at Q³ — with context (which customer, which compliance regime, which priority) and dose (who should see this, who set the expectation, what happens when there's a gap). The answer isn't probabilistic. It's traceable to source.
The difference isn't technical sophistication. It's whether you can stand behind the answer when the auditor arrives.
Swagelok's distributed network: 200+ points of sale, multiple carriers, handover darkness between systems. Each carrier tracks brilliantly — but "delivered" means four different things in four different systems.
Pathology: Darwinian. Units fend for themselves. Sandra reconciles by phone, 15 minutes per call.
Cure: Federate the disagreements. Don't merge the systems — make the differences visible. 15 minutes becomes 30 seconds.
42 Integrated Care Boards. Each trust defines "bed available" differently — clinically ready, discharge-pending, cleaned, staffed. A bed in Lewisham is not the same shape as a bed in Leeds.
Pathology: Statesman. Sophisticated strategy nationally, coordination absent locally.
Cure: Preserve local definitions. Surface the variance as signal. "Why does Trust A discharge 2× faster?" becomes a learnable question.
50,000 EU financial entities must demonstrate ICT risk oversight, third-party dependency mapping, and operational resilience — across systems that were never designed to talk to each other.
Pathology: Soviet. Centre demands the Register of Information. Edge fills it in to survive the audit, not to tell the truth.
Cure: Federate the Register from live systems. The RoI becomes a Q³ instrument — traceable, auditable, and impossible to game.
| Constraint | The ZQL Treaty Requirement | What It Eliminates |
|---|---|---|
| Read-only federation | Data queried where it sits. No replication, no movement, no lake. | Political battles over data ownership |
| Cryptographic provenance | Every answer traceable to source system, query, and vocabulary version. | Hallucinations, audit gaps, AI liability |
| Local vocabulary | Each system keeps its own definitions. Variance is signal, not noise. | Forced harmonisation projects |
| Symmetric visibility | All parties see the gap. No asymmetric advantage. | The Sucker's Payoff (and the lying it causes) |
| Existing infrastructure | Federates across systems already deployed. No rip-and-replace. | Multi-year transformation programmes |
We'll run this framework against your organisation — or your acquisition target, your supplier, your competitor — using only public signals. No access required.
Book a Call