AQ-SHOW-001 — First public research showcase architecture
STANDING
This document defines the architecture of the first public research showcase for Akasha-Q. It describes an intended process, not an event that has occurred. Nothing has been built, and no element of the programme has reached Lab validation. The showcase therefore exhibits four things and nothing else: the intended design, the method by which it will later be evidenced, the ways it is expected to fail, and the questions that remain open. The showcase is the DM-3 instance defined in AQ-DML-001; its claim discipline is inherited from that document and is not renegotiable on the floor.
Every exhibit carries an evidence-level chip from the Part 12 ladder — Concept Theory Simulation. No exhibit may carry Lab validation or above until an article log entry exists for it. Under DC-1 every displayed value is tagged MEASURED, DERIVED or MODELLED. There are no MEASURED values from AQ-NODE-01 at this showcase; any value shown as MEASURED is an instrumentation reading of the demonstration rig itself and is labelled a rig measurement, never a node measurement.
THE TEN SECTIONS
| ID | Section | What the attendee does | Delivered by | Evidence ceiling |
|---|---|---|---|---|
| S1 | Entry and framing | Hears what Akasha-Q is and is not: a communications platform on a six-rung reach ladder, the rung it currently occupies, the funding basis, and the Visvambhara boundary | Programme lead | Concept |
| S2 | L0 Physical Reality | Reads each physics claim beside its assumption set, and marks which assumptions are not yet measurable on the current rung | L0 physics lead | Theory |
| S3 | The ten layers L1–L9 | Walks the stack board from Sensing & Entropy Trust to Systems Engineering & Safety Governance and locates where a value becomes an inference | Systems architect (L9) | Theory |
| S4 | AQ-NODE-01 trust boundaries | Traces one signal across TB-1 to TB-5 and sees where determinism ends (TB-2, FPGA) and where a physical effect is hardware-gated (TB-4) | Node engineer | Concept |
| S5 | Command authority | Submits a command and watches Q1–Q4 resolve; sees Q4 reduce, and never expand, what Q1–Q3 already allowed | Command authority engineer (L5) | Simulation |
| S6 | Failure wall | Reads F-1 to F-14 and CF-1 to CF-7 stated as effects, unmitigated, before any mitigation is described | Failure steward | Theory |
| S7 | Evidence & Provenance Fabric (L6) | Chooses an incident from S6 and reconstructs the decision at a replay station from the record alone | Evidence steward | Simulation |
| S8 | Decision boundaries B-1–B-5 | Drives the article to abstain and to deny, and sees both recorded as outcomes rather than as errors | Decision boundary engineer | Simulation |
| S9 | Unproven register | Reads the FR and LH tier questions with no answer attached, including the ones the programme cannot presently frame as an experiment | Systems architect (L9) | None |
| S10 | Exit | Answers the two exit questions and receives the takeaway record | Programme lead | None |
AUDIENCE JOURNEY
Figure 1 - entry to exit, with the two staffed gates
ENTRY
|
v
[S1] framing: what it is / is not | funding basis | Visvambhara boundary
|
v
[S2] L0 physical reality --> [S3] ten layers L1..L9
|
v
[S4] AQ-NODE-01, TB-1..TB-5
|
v
[S5] command authority Q1..Q4
+-------------------------------+
|
v
GATE A (staffed): attendee names one displayed value that is MODELLED
|
v
[S6] FAILURE WALL F-1..F-14 / CF-1..CF-7 <-- failures BEFORE evidence
|
v
[S7] EVIDENCE FABRIC L6 + replay stations RS-1 RS-2 RS-3
|
v
[S8] decision boundaries B-1..B-5 (live abstain, live deny)
|
v
[S9] UNPROVEN REGISTER (FR / LH tiers)
|
v
GATE B = [S10] EXIT TEST: one failure named + one unproven item named
|
v
EXIT (takeaway record issued)
WHY FAILURES (S6) PRECEDE EVIDENCE (S7)
The ordering is load-bearing and is the single most reversible mistake in this architecture.
- Ordering decides the reading. An evidence fabric shown before any failure is read as proof that nothing fails. The identical fabric shown after the failure wall is read as the record that makes a failure reconstructable. The artefact does not change; the claim the attendee forms does.
- The replay stations have nothing to replay otherwise. RS-1 to RS-3 reconstruct incidents the attendee has just read at S6. Reversed, the operator must narrate a failure verbally and then produce its record, which is precisely the pattern that invites the conclusion that the record was written to fit the story.
- DC-4 is unintelligible in the other order. Trust falling on a single observation, and rising only through the full recovery sequence, only reads as a design commitment once the observation class is concrete rather than hypothetical.
- It sets up the exit test. An attendee who meets capability first and failure second remembers the first. The exit test at S10 requires the opposite retention profile.
- It removes the demonstration-grade reading. The programme is seen raising its own failure modes before anyone in the room raises them for it, which is the only durable way to establish that the evidence layer is not decoration.
The inverse ordering (evidence, then failures) is prohibited even under time pressure. If the floor runs late, S3 and S9 are compressed and the compression is announced. S6 is never compressed below the full F-* and CF-* wall and is never moved after S7. A showcase that drops or relocates S6 is not a DM-3 showcase and must be relabelled in every subsequent reference to it.
REPLAY STATIONS
Three stations sit inside S7. Each takes an incident the attendee selects and reconstructs the consequential transitions from the Evidence Fabric alone, satisfying DC-5. The operator's narration is not the evidence and is not permitted to supply anything the record does not contain.
- RS-1 — single-observation degradation. The attendee selects one sensing or timing anomaly from S6. The station shows the trust state moving
NORMAL→RESTRICTED→CRITICALon that one observation, and shows that no sequence of further favourable observations returns it directly toNORMALwithoutRECOVERING. - RS-2 — command under reduced trust. The attendee submits a command that passes Q1, Q2 and Q3. The station shows Q4 narrowing the outcome to
execute-restricted,safe-holdordeny, and shows that no trust state can widen an outcome that Q1–Q3 did not already permit. - RS-3 — absence and abstention. The attendee removes an input. The station must render the input as
undeterminedunder DC-2 and reachabstainunder DC-3. A station that renders a missing input as benign is defective by definition and is closed.
A replay station that cannot reproduce a decision from the record is closed by the floor marshal for the remainder of the day, and the closure is announced at S7 rather than worked around by narration. No station presents an L8 output as a decision: AI-assisted analysis is advisory only, and wherever it appears the human decision it informed appears beside it, with the human decision named as the operative one.
STAFFING AND DELIVERY AUTHORITY
| Role | Delivers | Standing authority and duty |
|---|---|---|
| R-1 | Programme lead — S1, S10 | Owns the framing sentence, the funding statement and the Visvambhara boundary. Answers every out-of-scope question with the refusal wording rather than an improvisation. |
| R-2 | L0 physics lead — S2 | States each assumption that cannot be measured on the current reach rung, unprompted, at least once per delivery. |
| R-3 | Systems architect (L9) — S3, S9 | Holds the open-question register. Adds a question raised on the floor to the register in the attendee's hearing when it is genuinely new. |
| R-4 | Node engineer — S4 | Owns TB-1 to TB-5 and must be able to name what crosses each boundary and what is refused at it. |
| R-5 | Command authority engineer (L5) — S5 | Demonstrates Q4 reducing an outcome at least once per session, including to deny. |
| R-6 | Failure steward — S6 | Presents failure modes without accompanying mitigation narrative. Holds authority to refuse any request to soften the wall. |
| R-7 | Evidence steward (L6) — S7, RS-1..RS-3 | Certifies each morning that every station reconstructs from the record alone; closes those that do not. |
| R-8 | Decision boundary engineer — S8 | Drives B-1 to B-5 to abstain and deny on request and treats both as successful outcomes in the wording used. |
| R-9 | Floor marshal — all sections | Enforces claim discipline, closes stations, and may stop any delivery mid-sentence for a prohibited claim. Reports to no section owner. |
Before the doors open, each deliverer states three things about their own section without prompting: its evidence level, its refusal condition, and one value in it that is MODELLED rather than MEASURED. A deliverer who cannot is replaced. If no replacement exists, the section is withdrawn and its absence is announced at S1 as a gap rather than concealed by redistributing its material. No section is delivered by a person reading another person's script.
EXIT TEST
At S10 every attendee is asked, and answers before leaving: name one thing that fails, and name one thing that is unproven. Answers are tallied anonymously at the exit desk as counts by category. No attendee is identified and no answer is attributed. The test measures the showcase, not the attendee.
No pass threshold is asserted here. The metric is the proportion of departing attendees able to name a specific failure mode by its effect and a specific unproven element by name. The first showcases are instrumented to establish the baseline distribution of answers; the threshold, and the precise category-coding rule that makes the proportion computable, are set afterwards from that measurement and recorded in the showcase log. Asserting a number before the first measurement would violate DC-1 in the document that exists to enforce it.
Remediation is directed at the architecture. If attendees name capability but not failure, the defect lies in S6 duration, S6 staffing or the S1 framing, and the remedy is re-scripting those sections. If attendees name failures but no unproven element, S9 is under-delivered and is moved earlier in the compression order. An attendee who answers neither question is evidence about the floor plan, never about the attendee.
The takeaway record is a single page: per-section evidence levels, the reference for the open-question register, and the plain statement that no element has reached Lab validation. It carries nothing drawn from the restricted-access record shared with Visvambhara.
CLAIM DISCIPLINE ON THE FLOOR
- The words secure, resilient, autonomous, high-precision, low-latency and quantum-secure are not used as unqualified descriptors. Where the last is used at all, the key-distribution assumption set, the threat model and the evidence level are named in the same sentence.
- Never claimed, in any wording, by any deliverer: unhackable; impossible to intercept; universally quantum-secure; information-theoretically secure at system level; satellite-ready; space-proven; operational.
- No geopolitical, sovereign-capability, national-security or defence framing. Questions of that class are answered with a fixed sentence — the programme does not make claims of that kind — and the delivery continues.
- Akasha-Q is described as a communications platform advancing along six rungs: optical bench → metro fibre and free-space → 20–30 km HAPS relay → CubeSat → LEO satellite → constellation. The rung the programme currently occupies is named at S1 without softening.
- Visvambhara is described as a separate aerospace programme and as the first platform designed to carry an Akasha-Q terminal — never as the reason Akasha-Q exists and never as its parent. The shared record is restricted and does not appear on the floor.
- Funding is stated plainly: EduRankAI revenue and disciplined reinvestment. External capital, partnership and government collaboration accelerate the programme; no station implies they are preconditions for its existence.
- No numeric performance target is displayed anywhere. Where an attendee asks for one, the answer names the metric and the method by which its target will later be established from measurement.