Adjacency Research Program · Hardware Program
Build + fund the test
An execution roadmap from today to independent replication. We have designed the scoreboard. We have not built the machine.
01 · Mission
Build the smallest auditable Golden Metrology Rig capable of testing whether post-arming unpredictable information can be decoded at endpoint B before the conservative lower bound of every inventoried physical causal path between A and B. Smallest, because every additional component is another unaudited channel. Auditable, because the claim is only as strong as the channel inventory behind it. The expected outcome is no anomaly or an architecture effect; the programme is designed so that outcome is still valuable.
Imagination chooses the question; measurement chooses the answer.
02 · System architecture
Independent reference + TDC at endpoint A.
Independent reference at B. Different calibration model to A.
No live control path. Raw unsmoothed capture, hashed.
Shielded enclosures, filtered feedthroughs, separate supplies.
Temperature, voltage and drift logging throughout every run.
Deliberate bypass installer used as a positive control against the audit.
Schematic of a proposed apparatus. Nothing shown here has been built.
03 · Golden Rig vs Engineering Rig
Minimal, auditable, and frozen for the duration of evidence collection. Firmware, analysis and topology are hashed. It is never optimised, never upgraded mid-campaign, and never modified before its baseline is archived and reproduced. Its only job is to be believable.
Forked only after verification clears, and only ever as a separate machine. This is where optimisation, scaling, IP and productisation live. Work here can never feed back into the Golden Rig while a campaign is open.
04 · Build phases B0 – B8
Each phase also carries full stage-gate QA — entry criteria, procedure, raw outputs, acceptance, failure and re-test rules, responsible role, hashes and calibration IDs — on the process validation page, alongside the synthetic campaign sanity tests.
Process validation →B0 · Clock loopback
planned- Inputs
- Two endpoint FPGAs, both timing stacks, short reference cable.
- Test
- Timestamp a signal that leaves and returns along a known path; compare both timing systems against each other and against themselves over hours.
- Pass criteria
- Both stacks agree within stated uncertainty; drift characterised and logged; no unexplained offset.
- Failure response
- Stop. No later phase means anything until the clocks are trustworthy. Rebuild the timing chain, not the experiment.
- People
- Timing/metrology lead, FPGA engineer.
- Equipment
- 2x FPGA endpoints, White Rabbit-class references, independent time-interval instrument.
- Deliverable
- Timing uncertainty budget measured on our hardware, not quoted from datasheets.
B1 · Post-arming entropy and decode
planned- Inputs
- Hardware TRNG at A, decoder at B, arming sequence.
- Test
- Generate an unpredictable payload only after arming; verify it decodes at B over an ordinary path; log entropy health and hash the payload at creation.
- Pass criteria
- Decode success above the preregistered error threshold; payload demonstrably unpredictable before generation; entropy health nominal.
- Failure response
- Fix the payload chain. A correlation without a decode is not information and will never count.
- People
- FPGA/embedded lead, statistician.
- Equipment
- Hardware TRNG, encoder/decoder firmware, raw capture.
- Deliverable
- Validated information-transfer primitive with a frozen decode rule.
B2 · Delay bank calibration and drift
planned- Inputs
- Selectable fibre spools / delay lines, environmental sensors.
- Test
- Characterise every delay segment bidirectionally, across temperature and over time; OTDR every fibre; repeat with spools swapped.
- Pass criteria
- Group delay known per segment with conservative uncertainty; drift bounded and logged; nominal length never used as an input.
- Failure response
- Re-characterise or replace the segment. An uncalibrated delay is the fastest route to a false anomaly.
- People
- Metrology lead, RF/EMI engineer.
- Equipment
- 1 km-class SMF spools, matched transceivers, OTDR, thermal logging.
- Deliverable
- Calibrated delay bank with per-setting empirical envelope.
B3 · Hidden-shortcut detection challenge
planned- Inputs
- Deliberate bypass injection fixture; blinded operator.
- Test
- A colleague secretly installs conventional bypasses. The channel-inventory and audit pipeline must find every one before unblinding.
- Pass criteria
- 100% detection. Anything less and the apparatus is not able to support any claim.
- Failure response
- Hard stop. Extend the inventory, add RF/optical leakage scans, repeat until detection is total.
- People
- RF/EMI engineer, red-team statistician, facility partner.
- Equipment
- Shielded enclosures, RF scanner, optical leakage detection, injection fixture.
- Deliverable
- Audit pipeline with a demonstrated 100% detection record on known positives.
B4 · Ordinary C0 propagation
planned- Inputs
- 16-node oscillator substrate, conventional routing.
- Test
- Measure ordinary transport across the substrate under the strongest permitted conventional signalling; build the empirical minimum-over-paths bound D_tau.
- Pass criteria
- Reproducible baseline consistent across sessions with a conservative lower-confidence bound.
- Failure response
- The substrate is not characterised. No comparison is possible without a baseline.
- People
- Experimental physicist, FPGA lead.
- Equipment
- Custom 16-node oscillator/coupling PCB, DAQ.
- Deliverable
- Empirical causal envelope D_tau for the apparatus as built.
B5 · Nonlinear / moving-front analogue
planned- Inputs
- Substrate driven into the nonlinear regime; two-phase control.
- Test
- Run the moving-front architecture and measure transport relative to the ordinary route C0.
- Pass criteria
- A reproducible architecture effect relative to C0. This is an ARCHITECTURE result and carries NO causal claim whatsoever.
- Failure response
- Record the negative result and publish it. The programme has published every failure so far.
- People
- Nonlinear dynamics expert, experimental physicist.
- Equipment
- Substrate, control firmware, DAQ.
- Deliverable
- Characterised analogue transport behaviour, explicitly bounded by D_tau.
B6 · Blinded mixed controls
planned- Inputs
- Randomised run schedule mixing real, null, sham and bypass conditions.
- Test
- Operators and analysts are blind to condition; the frozen analysis assigns verdicts without knowing labels.
- Pass criteria
- Verdicts match conditions at the preregistered rate; no condition leaks through operator behaviour.
- Failure response
- The blinding is broken. Redesign the run protocol before any confirmatory data.
- People
- Independent statistician, full team.
- Equipment
- Blinding harness, sealed condition schedule.
- Deliverable
- Demonstrated blinded pipeline ready for a confirmatory campaign.
B7 · Frozen confirmatory causal-bound campaign
gated- Inputs
- All prior phases passed; firmware, analysis and protocol hashed.
- Test
- The full delay-amplified test across multiple delay settings, with cable-swap nulls, topology permutation and dual timing.
- Pass criteria
- Delta = D_tau,lower − T_decode,upper positive by the preregistered guard band at every delay setting, with all quality gates green.
- Failure response
- Report NO ANOMALY or ARCHITECTURE EFFECT — the expected and entirely publishable outcome.
- People
- Full team plus external observer.
- Equipment
- Complete Golden Rig.
- Deliverable
- A published dataset with hashes, whatever it says.
B8 · Independently assembled second apparatus
gated- Inputs
- Published protocol and parts specification only.
- Test
- A second rig, built from the spec rather than cloned, runs the same frozen campaign.
- Pass criteria
- Preregistered result reproduced with no re-tuning.
- Failure response
- The first result was apparatus-specific. That is a finding, and it is reported as one.
- People
- Second team, facility partner.
- Equipment
- Full second rig.
- Deliverable
- Replication dataset, or a documented failure to replicate.
05 · Hardware & bill of materials
| Category | Item | Note | Cost / status |
|---|---|---|---|
| Endpoints | Two independent FPGA endpoint boards | Separate power domains, separate timestamp hardware, no shared control bus during a run. | ~$2.5k subtotal (from our current BOM) |
| Timing | White Rabbit-class timing references | Reference architecture for distributed sub-ns synchronisation. Performance must be established on our hardware. | REQUEST QUOTE |
| Timing | Independent calibrated time-interval instrumentation | Second stack with a different calibration model — deliberately not sharing the reference chain. | REQUEST QUOTE / PARTNER FACILITY |
| Payload | Hardware TRNG | Post-arming payload generation with entropy health monitoring. | REQUEST QUOTE |
| Delay | 1 km-class single-mode fibre delay spools / selectable delay bank | Multiple switchable segments so delay scaling is a configuration change, not a rebuild. | REQUEST QUOTE |
| Delay | Matched optical transceivers | Characterised bidirectionally; asymmetry is a first-class error term. | REQUEST QUOTE |
| Substrate | Custom 16-node electronic oscillator / coupling PCB | Programmable weighted, directed coupling. Requires design, fabrication and revision cycles. | DESIGN REQUIRED |
| Isolation | RF shield enclosures and filtered feedthroughs | Directly gates B3. Under-specifying here invalidates everything downstream. | REQUEST QUOTE |
| Isolation | Isolated / battery power domains | Separate supplies per island, or fully characterised coupling. | REQUEST QUOTE |
| Environment | Temperature and environment sensing | Continuous logging; timing claims without environmental records are not auditable. | REQUEST QUOTE |
| Data | Raw DAQ and storage | Offline and blinded; raw waveforms preserved unsmoothed with published hashes. | REQUEST QUOTE |
| Calibration | Calibration standards and cabling | Traceable standards, OTDR access, matched and characterised cable set. | REQUEST QUOTE / PARTNER FACILITY |
06 · Capital plan — planning envelopes
Budget A
Partner-Assisted MVP
$75k – $150k
Leans on university or national-lab metrology where available. We supply endpoints, substrate and protocol; the partner supplies timing instrumentation, fibre plant and calibration.
Reaches: B0 through B3, and B5 if facility time allows.
Budget B
Independent Golden Rig
$250k – $400k
Dedicated instrumentation and personnel. Own timing stacks, own delay bank, own shielding, no dependency on a host facility's schedule.
Reaches: B0 through B7 — a complete frozen confirmatory campaign.
Budget C
Replication + Engineering Program
$1.0m – $1.6m
Second independently assembled rig, external replication support, custom PCB and firmware revisions, IP work, publication, and the engineering fork.
Reaches: B8 and the Tier 2 handoff, if and only if the verification ladder clears.
07 · Use of funds — $305,000 Phase I target
- Scientific & engineering personnel30% · $91,500
- Metrology, timing, fibre, instrumentation25% · $76,250
- Oscillator, FPGA, custom electronics15% · $45,750
- External facility & testing10% · $30,500
- Fabrication & calibration8% · $24,400
- Data, software, security7% · $21,350
- Contingency, IP, admin5% · $15,250
08 · Funding ladder
Founder / pre-seed
Reach B3 (hidden-shortcut detection) and, with facility help, B5.
NSF SBIR/STTR — Scientific Instrumentation
Phase I: instrumentation build and the audit pipeline.
Strategic deep-tech / quantum / instrumentation angels
Bridge to a dedicated Golden Rig.
NSF Phase II / Fast-Track
Only if eligible and only if milestones justify it.
Strategic / venture capital
Only after credible B5/B7 data exists. Not before.
NSF SBIR/STTR — displayed deadline
November 4, 2026
69d
Deadline shown for planning only. Program terms, topic eligibility and submission windows must be rechecked directly with the agency before any submission.
09 · Partner targets
Tier 1 target · Brookhaven National Laboratory Quantum Network Facility + Stony Brook NYS Quantum Internet Testbed
- The BNL Quantum Network Facility is open to the user community for benchmarking and validating concepts — which is exactly the kind of engagement this programme needs.
- BNL/Stony Brook classical control infrastructure uses White Rabbit timing over long-distance fibre.
- Stony Brook reports a testbed of six nodes across roughly 300 km of dark fibre with sub-nanosecond synchronisation.
We are NOT claiming our experiment is quantum, and we are not proposing quantum work. What we need from a partner of this class is precision timing expertise, characterised long-fibre plant, calibration instrumentation, and people who have already been burned by timing artefacts. Their scepticism is the deliverable.
Principal experimental physicist
Owns the physical claim and the right to stop the programme.
Precision timing / metrology lead
Owns the uncertainty budget — the most load-bearing number in the whole design.
FPGA / embedded lead
Owns firmware freeze, timestamping and the arming sequence.
Nonlinear dynamics expert
Owns the oscillator substrate and the moving-front regime.
RF / EMI engineer
Owns the channel inventory and the hidden-shortcut challenge.
Independent statistician / red team
Owns blinding, exclusions and the frozen analysis. Paid to falsify us.
Lab / facility partner
Supplies fibre plant, calibration standards and institutional credibility.
10 · Outreach funnel
Partner scorecard
11 · Data room checklist
| in progress | 2-page technical brief | Drafting from the frozen protocol pages. |
| ready | Theory hypotheses H1–H3 | Published on this site and versioned. |
| in progress | Frozen causal-bound protocol + hash | Frozen-draft v1.0; hash issued at build start. |
| in progress | Bill of materials | Endpoints spec'd; metrology lines out for quote. |
| ready | B0–B8 test matrix | Defined on this page. |
| ready | Synthetic red-team results | Complete and clearly marked SYNTHETIC. Not evidence about hardware. |
| ready | Alternative models & kill board | Maintained publicly, including everything that failed. |
| missing | IP invention log | Not started. |
| in progress | Budget | Planning envelopes only. |
| in progress | 12-month Gantt | Derived from B0–B8; dates pending funding. |
| missing | Team gaps | Every experimental role is currently unfilled. |
| missing | Partner letters | None yet. |
| missing | Funding application | Not submitted. |
12 · The kill board
Exact-fabric concurrency unresolved
The toy architecture's concurrency behaviour on the exact fabric remains open. It is not cleared by any later phase.
Activation throughput unresolved
Throughput debts from the concurrency front are still outstanding and still red.
Local nonlinear mimicry not proven unique
We have not shown that the moving-front behaviour cannot be produced by an ordinary local mechanism.
Complete physical-channel inventory requires hardware
The inventory cannot be completed on paper. Until an apparatus exists, the channel list is provisional.
Absolute timing uncertainty requires hardware
Datasheet precision is not measured accuracy. Our uncertainty budget does not exist yet.
Scaling beyond 64 nodes unverified
Nothing above N = 64 has been validated even in simulation at the fidelity the claim would need.
Independent replication: none
Zero external replications of anything in this programme.
Causal anomaly evidence: none
There is no evidence of a causal anomaly. None has been observed, because no measurement has been taken.
13 · Investor story — two-sided value
If no causal anomaly appears — the expected outcome
The programme still produces a precision causal-path instrumentation stack: an empirical channel-inventory and audit methodology, programmable nonlinear oscillator networks with characterised transport, local sparse transport and control architectures with published cost accounting, and timing/audit tooling for anyone who needs to prove that a signal did not take a hidden route. Those are real instrumentation and methodology assets, and they exist regardless of what the physics says.
If an anomaly candidate survives
It proceeds through the verification ladder — independent timing, an independently assembled apparatus, then external replication — before any extraordinary claim is made by anyone. There is no version of this programme in which a single rig's result becomes a announcement.
14 · What we will not claim
15 · Origin · why we asked the question
The question was provoked by information-first framings of reality — the Thomas Campbell-style idea that what we experience is better modelled as information than as substance. Those ideas are SPECULATIVE and NOT ESTABLISHED PHYSICS. They are not evidence, they are not cited as support, and nothing in this programme depends on them being true.
Our scientific translation is narrow and testable: imagination chooses the question; measurement chooses the answer. A speculative framing is allowed to suggest what to measure. It is never allowed to influence what the measurement says.
16 · SEEKING
If you can falsify this design faster than we can, that is the most valuable thirty minutes anyone could give us.
Contact the programme →