AQ-DML-001 — Demonstration maturity ladder DM-0 to DM-6
PURPOSE
This document defines seven demonstration maturity levels, DM-0 to DM-6. The ladder governs what may be shown, to whom it may be shown, what may be said about it afterwards, and what must be true before the next level is claimed. It is a defined process. No element of Akasha-Q has reached Lab validation, and no demonstration above DM-0 has taken place; every level below is a specification of a future act, not a record of a past one.
The ladder is independent of the six-rung reach ladder and of the Part 12 evidence ladder. The three are orthogonal axes. A position on one never implies a position on another, and the most common misuse of this document will be to treat them as a single scale.
A DM level is never a property of the programme, a layer or a technology. It attaches to (article, reach rung, claim, evidence level, date of record). The sentence “Akasha-Q is at DM-4” is malformed and is corrected wherever it appears, including in material the programme did not write, because an uncorrected malformed level becomes the level everyone quotes.
LADDER INVARIANTS
- IV-1 A level attaches to the tuple above, never to the programme as a whole.
- IV-2 A demonstration cannot exceed the evidence level of its weakest load-bearing element. If one input on the path to the outcome is MODELLED, the whole demonstration is bounded at Simulation regardless of how much of the remainder is MEASURED.
- IV-3 Observers do not set the level. A large public audience does not raise it; an empty room does not lower it. Evidence and observation conditions set it.
- IV-4 Degradation is monotonic without re-establishment. Any change to the article, protocol or assumption set returns the affected claim to DM-1 until the sequence is re-run. This mirrors DC-4 and is not a formality to be waived for a minor change; the judgement that a change is minor is exactly the judgement DC-4 exists to distrust.
- IV-5
abstainanddenyare results. A demonstration in which the article refused to act is recorded, presented and counted with the same prominence as one in which it acted. - IV-6 L8 AI-assisted analysis is advisory only at every level and may never be the evidence on which a level advances.
- IV-7 A level transfers to nothing. DM-n on the optical bench rung says nothing about metro fibre, HAPS relay, CubeSat, LEO or constellation. Each rung begins again at DM-0.
THE LADDER
Figure 1 - DM-0 to DM-6, the DM-4 hazard band, and the three independent axes
DM-6 | high-maturity public capability | repeated independent execution
DM-5 | independent execution to protocol | third party at the controls
+------------------------------------------------------------------+
| DM-4 | unscripted third-party observation | observer picks inputs |
| HAZARD BAND - observation is recorded here as if it were |
| validation, by everyone in the room, unless prevented |
+------------------------------------------------------------------+
DM-3 | controlled public research showcase | AQ-SHOW-001 sits here
DM-2 | supervised external technical review| objections dispositioned
DM-1 | internal instrumented demonstration | article log + raw capture
DM-0 | internal walkthrough | no article required
three independent axes - a level on one implies nothing on another:
DM level : how it was shown, and to whom
reach rung : bench -> metro/free-space -> HAPS -> CubeSat ->
LEO -> constellation
evidence level : Concept -> Theory -> Simulation -> Lab validation ->
Field validation -> Dynamic platform ->
Flight/space -> Operational demonstration
LEVELS: OBSERVERS AND EVIDENCE
| DM | Name | Who may observe | Evidence that must exist | Ceiling |
|---|---|---|---|---|
| DM-0 | Internal walkthrough | Programme members only | Written design description and assumption register, read end to end by someone other than its author. No article required. | Concept |
| DM-1 | Internal instrumented demonstration | Programme staff and named internal reviewers. No external attendees, no recording released outside. | Article log entry, raw capture retained, DC-1 tagging applied to every value, and the run script as executed rather than as written. | Theory |
| DM-2 | Supervised external technical review | Invited technical reviewers under a written access record. Adversarial questioning is expected and welcomed. The record shared with Visvambhara stays out of scope. | DM-1 evidence plus a review pack issued in advance: assumption set, known failure modes F-*, conflicts CF-*, and the open-question register. | Theory |
| DM-3 | Controlled public research showcase | General public, under the claim discipline of AQ-SHOW-001 | DM-2 evidence plus replay capability: every consequential transition shown is reconstructable from the Evidence Fabric alone, satisfying DC-5. | Simulation |
| DM-4 | Unscripted third-party observation | Third parties who choose the inputs, including inputs the programme did not anticipate. Programme staff operate the article but do not select the case. | Session log naming the observer-selected inputs, full capture, and every abstain and deny outcome recorded with the same weight as execute. | Simulation |
| DM-5 | Independent execution to published protocol | An independent party executes; the programme observes and does not touch the controls. | Published protocol, the independent party's own record, the programme's record, and a written reconciliation of every divergence between them. | Lab validation |
| DM-6 | High-maturity public capability demonstration | Public, independent executors, and reviewers with access to the full evidence chain | Repeated independent execution, sustained operation across a defined window on one named reach rung, failure modes exercised deliberately rather than encountered, and a claim scope tuple recorded per result. | Lab validation or above, per rung |
LEVELS: CLAIM CEILING AND ADVANCEMENT
| DM | May be claimed | May NOT yet be claimed | Advancement criteria |
|---|---|---|---|
| DM-0 | That a design exists, is written down, and has been read end to end by more than its author. | Any behaviour whatsoever. No statement about what the design does, only about what it intends. | An article exists (bench rig or software article) and an instrumentation plan names, in advance, which values will be MEASURED, DERIVED and MODELLED. |
| DM-1 | That the article produced the following tagged observations, under the stated conditions, on the stated date, with the stated script. | Repeatability. Performance characterisation. Any evidence level above Theory. That an unexercised failure mode does not occur. | The run is repeated on a separate occasion by a second operator using the same script; all divergences are recorded, including a null result; failure modes triggered are entered as F-* candidates. |
| DM-2 | That external technical reviewers examined the article and the evidence, and recorded the following objections and unresolved questions. | Endorsement. Validation. That an objection was resolved because it was discussed. That silence from a reviewer is agreement. | Every objection is dispositioned in writing as accepted, rejected with reason, or open; open items enter the FR register with a named owner. |
| DM-3 | This is the intended design; this is what fails; this is what is unproven; here is the record behind each thing shown. | Capability. Performance. Readiness of any kind. Any prohibited term. That public interest constitutes evidence. | Exit-test instrumentation collected; every replay station reproduces from the record alone; a third party has been briefed sufficiently to choose their own inputs. |
| DM-4 | An observer selected inputs the programme did not choose, and the outcomes under those inputs are recorded here — including the outcomes the programme did not want. | Validation of any kind. Statistical coverage. That failure classes not exercised are absent. That the observer's selection was representative. See the hazard rule below. | A written protocol exists that a third party can execute with no programme staff at the controls, and a negative-result path is defined in advance of execution. |
| DM-5 | An independent party executed the published protocol and obtained the following, which agrees with or diverges from the programme record in these specific respects. | Operational status. That the article behaves outside the protocol's stated conditions. That reproduction on one reach rung transfers to another. | Two independent executions; at least one negative or partially negative result published with equal prominence; a sustained-operation window defined with its instrumentation. |
| DM-6 | A bounded capability statement naming article, reach rung, conditions, assumption set, measured quantities and evidence level. | Anything about a rung not demonstrated. Unhackable, impossible to intercept, information-theoretically secure at system level, space-proven, or operational as a general statement. That DM-6 on one rung licenses the next rung. | None. DM-6 is terminal for its tuple. The next reach rung begins again at DM-0 with its own article and its own evidence. |
THE DM-4 HAZARD
DM-4 is where programmes of this class conflate observation with validation, and they do it in good faith. An observer choosing the inputs feels like a test: the programme did not pick the case, the outcome was not rehearsed, and the room experiences the result as proof. It is not proof. A third party who selects ten cases has exercised ten cases. Every failure class not selected remains entirely unexercised, and under DC-2 the absence of a failure in an unexercised class is undetermined, never benign. The observer's selection is not a sample; it is a selection, and its coverage is unknown to both parties.
No DM-4 result may use the word validated, or any inflection of it, in material prepared for any audience — internal note, reviewer pack, funding conversation, partner briefing, press response or public page. Also prohibited at DM-4: verified, proven, confirmed, shown to be safe. Permitted verbs: observed, recorded, exercised, selected, reproduced under the same script. The prohibition is on the word appearing, not merely on the intention behind it, because the word is what survives into the reader's summary after the qualification around it is discarded. A single audience-specific relaxation of this rule voids the DM-4 record for every audience.
GOVERNANCE OF THE LEVEL
- The L9 systems engineering and safety governance function is the only authority that records a DM level. A level not written into the L9 record does not exist, whatever occurred in the room and whoever witnessed it.
- No level is awarded on the day of the demonstration. The record is written after the capture is reconciled against the article log, so that the reaction in the room cannot become part of the evidence.
- Any deliverer, evidence steward or independent executor may refuse a level. Refusal requires a written reason; it does not require anyone's agreement, and it is not overridden by seniority.
- Demotion under IV-4 is recorded with the same prominence as promotion, in the same register, with the same distribution. A demotion is not a failure of the programme; an unrecorded demotion is.
- Where material is prepared for different audiences, the DM level and its full tuple appear in identical wording in all of them. Audience-specific phrasing of a level is prohibited, and its detection is grounds for withdrawing the level.
- Funding context does not enter the ladder. External capital, partnership and government collaboration accelerate the rate at which levels are attempted; they never lower the evidence required to record one.