IDA recovery and programme synthesis

### PDF page 16 IDA and the Boundedness Engine — v1.0 8. Conjecture death: α = 1/12, the sextic lever, or birth-death diagnostics fail in the systems where the conjecture claims them. Firewall discipline. Local deaths are survivable only because the claim types are firewalled: a failed microscopic conjecture does not kill Aczél composition; a failed IDA trial does not kill recovery metrics; a failed recovery metric in one dataset narrows the model rather than destroying every bounded-domain result. Conversely, survival in one domain is never proof in another. 12. Conclusion The corrected boundedness engine is not a claim that every phenomenon is the same residue. It is a disciplined way to ask whether bounded systems are being misread because the lawful coordinate, the return path, or the unresolved remainder has been projected away. The chain is now cleanly typed: descriptions carry redundancy; physical content is the invariant residual; finite systems represent residuals through bounded coordinates; lawful composition selects generator coordinates; wrong coordinates create measurable residual errors; adaptive systems reveal failure through return and recovery residue; addressed systems require writeability for corrective updating; and IDA tests whether recovery geometry can be controlled by respecting residue rather than forcing state. The older Theory of Everything returns only as a possible microscopic completion, never as the burden placed on the reader at the door. If a companion result survives review or testing, this synthesis explains why it matters. If none do, this paper is the map of where and why the programme failed. Named contributions. Three constructs in this work are offered as named handles for others to adopt, test, or refute. The residue–negentropy correspondence (Section 6.4): under the stated operational definition, the recovery residue is exactly an accumulated negentropy deficit, so a residue-falling gate is a sign test on the rate of order export. Residue-gated control (Sections 9–10): the control principle that a bounded adaptive system should advance only when unresolved load is falling — an inversion of symptom-gated intervention. And ANDY, the attractor-normalised drift yield (Section 6.2): the apparent recovery that is really baseline drift, the error term a frozen baseline exists to expose. None is claimed as a law of nature; each is a defined object with a death condition. Final wager. IDA is not a machine that adds awareness from outside. It is a bounded-domain controller that asks whether writeable corrective updating improves when unresolved perturbation is no longer amplified and cleaner return can be written back to the state that actually needed repair. The name states the intent plainly: I Develop Awareness. Should the wager survive the ladder set out here, the intent is that IDA be built — simply, safely, and openly — so that the help it may offer is available rather than merely argued. The schema itself cannot be proven; its typed instances can be, and that is the whole of the claim. The honest hope is the same as the honest test: let someone build this and show it works, or build it and show exactly why it does not. Either outcome is a gift. Page 16 of 35 ### PDF page 17 IDA and the Boundedness Engine — v1.0 Appendix A. Reference implementation (real, runnable) The following is the complete, dependency-light reference engine, reproduced verbatim from the accompanying file ida_reference.py (Python 3.9+, numpy only). It runs offline on recorded or simulated signals; the actuator is a logging stub. Run ‘python ida_reference.py --demo’ to reproduce the predicted clean-versus-lingering signature described in Section 9.5, or ‘--csv yoursignal.csv’ to analyse a recorded single channel. This is the literal specification: an engineer can build the read-side from this alone. Page 17 of 35

Independent Researcher, Melbourne, Australia Final synthesis manuscript v1.0 — June 2026 Status note. This is a public synthesis and prediction document, not a conventional journal submission, and it makes no clinical claim. It maps a research programme across bounded composition, recovery geometry, memory updating, psychiatric measurement, redox biology, prime- residue projection, and an optional microscopic conjecture. Its central methodological commitment is typing: quotient residuals, coordinate residuals, recovery residues, addressed residuals, and definition, and additivity testing. Where bounded adaptive systems are perturbed, the decisive variable is not a static value but the geometry of return: clean return resolves load, unresolved displacement leaves a measurable residue, and repeated residue is hypothesised to shift the destination of return. This manuscript contributes four things: a typed-residue framework that prevents category errors across domains, a recovery-geometry metric for return after perturbation, a named drift-error term (ANDY), and a falsifiable closed-loop control prediction with a runnable reference implementation. It organises its observations as a typed-residue ladder rather than a single unified object. A quotient residual, a coordinate residual, a leaky recovery integral, an address-indexed organismic state, and a conjectural birth-death record instantiate one schema — hidden lawful object → bounded projection → lost information → typed residue → held-out test — at different type levels. Each rung carries its own death

IDA and the Boundedness Engine — v1.0 The empirical wager is placed first: return-geometry features must beat static baselines on held-out prediction, or the programme has no valid input. IDA — a proposed read-first, residue-gated headband, supplied here with a complete, runnable reference implementation — then tests a single write-side prediction: if recovery geometry is controllable, residue-gated closed-loop actuation should move a system toward cleaner return more effectively than matched open-loop actuation. IDA is a falsification instrument, not a treatment. A short, exact result shows that — under one operational definition — the recovery residue is identical to an accumulated negentropy deficit, so the same conclusion is reached by a second reading of the system as a flow of exported order, which points to a well-posed next problem in physics. The instrument is offered openly so that others may build it and either confirm the wager or show precisely why it fails. Keywords: bounded domains; typed residue; Aczél representation; lawful coordinates; Euclidean error; recovery geometry; allostatic drift; writeability gate; closed-loop control; falsification. Central claim. Typed residues are not the same object. They are locally defined tests of whether a projected coordinate misses recoverable structure. Schema recurrence across domains is not itself evidence; held-out predictive gain in each domain is the evidence. What is new here. 1. The typed-residue ladder. Residues recur across domains but are not one mathematical object; each rung is typed, with its own death condition. 2. ANDY — attractor-normalised drift yield. A named, measurable error term: apparent recovery produced by baseline drift rather than true return (Section 6.2, Figure 2). 3. Residue-gated control. A control principle — advance only when unresolved load is falling, inverting symptom-gated intervention (Sections 9–10). Plus: the residue–negentropy correspondence (Section 6.4, exact under one operational definition), a recovery-geometry metric, and a falsifiable closed-loop control prediction with a complete, runnable reference implementation. Reader’s route. Mathematical core: Sections 3–5. Recovery-geometry model and ANDY: Section 6. Cognitive extension: Section 7. Microscopic conjecture (fenced; may fail without killing the rest): Section 8 only. Device prediction and falsification ladder: Sections 9–10. The core (Sections 4–6) is self-contained and depends on none of the speculative material. Page 2 of 35

distribution, and a microscopic record do not live in the same mathematical category. The unifying object is schema recurrence, and the scientific content lives entirely in the typed instances and their separate death conditions. Missing structure made explicit. Schema: hidden lawful object → bounded projection → lost information → typed residue → held-out test. This schema is a search heuristic and filing system, not a theory. It is true of much of statistics and physics (sufficient statistics, latent-variable models, renormalisation) and therefore cannot itself be falsified. Only the typed instances below can live or die. 1.1 Dual framing: SSRN programme document with a self-contained journal core This document serves two readers. For the SSRN reader it is an architecture and falsification map for a multi-paper programme. For a journal editor, Sections 4–6 (bounded composition, the Euclidean error, and recovery geometry) are written to be liftable as a self-contained methodological contribution that depends on none of the speculative material. The cognitive model (Section 7) and the device proposal (Sections 9–10) are clearly marked as forward-looking and are not required by that core. Definition- and theorem-level statements: quotient by redundancy; Aczél representation under its hypotheses; the conditional semigroup time normal form; the Hill-1 matched-diagonal discrimination point. Definitions and proof. Model / effective Recovery residue, allostatic drift, mode gates, addressed-residual updating. Held-out falsification; not claimed as necessities of nature. Device prediction If recovery geometry is controllable, residue-gated closed-loop actuation beats matched open-loop actuation. Sham / open-loop / closed-loop trials.

3. Define the computable typed residue without reference to the outcome. 4. Declare the static or native-coordinate baseline it must beat. 5. State the held-out death condition: the predictive gain that, if absent, kills the local claim. 2.2 The typed residue ladder The word residue marks a genuine recurrence — something non-removable or unresolved survives a projection, quotient, perturbation, or retrieval — but it is always typed. The ladder below makes the type level and verification state explicit. The verification-state column is the second-order application of checked here R1 coordinate observed − generator prediction Error made visible when a bounded coordinate is fit in the wrong chart. Statistical Standard;

saturating effects, viability fractions, finite budgets, normalised sensor outputs. The native bounded value is frequently not the lawful coordinate. Theorem (Aczél representation, stated with its hypotheses). Let I be an open interval and : I × I → I a ⊕ binary operation representing lawful composition. If is continuous, strictly monotone in each ⊕ argument, associative, and closed on I, there exists a strictly monotone generator ψ, unique up to positive affine rescaling, with ψ(x ⊕ y) = ψ(x) + ψ(y), so x ⊕ y = ψ (ψ(x) + ψ(y)).⁻¹ and nothing more. It does not assert that any particular transform (artanh, logit, log-hazard) is universal, and it does not derive recovery geometry, memory updating, IDA, or any microscopic claim. 4.1 When the hypotheses fail — a decision tree 1. If strict associativity holds on I: use the full Aczél generator. 2. If only local one-parameter flow composition holds: use the local infinitesimal generator ψ(x) = ∫ dx / v(x) as an empirical chart, not a global law. 3. If neither holds: do not claim lawful composition; treat the coordinate as a fitted model only.

IDA and the Boundedness Engine — v1.0 4. In every case: if the generator coordinate does not improve held-out prediction or compositional consistency over the native coordinate, the coordinate claim dies. Use case Native Generator ψ Reading Signed bounded effect / velocity u (−1,1)∈ artanh(u) rapidity-type composition

IDA and the Boundedness Engine — v1.0 Figure 1. The Euclidean error made visible. Two effects composing by independent action. Predicting the composition additively in the native bounded coordinate (dashed) diverges catastrophically as effects approach the saturation boundary, while the generator-chart prediction (dotted) tracks the true composition (solid) exactly. The lawful coordinate is not the native one. 6. Adaptive systems: return, recovery residue, and drift A bounded adaptive system is defined by a viable domain, a defended attractor or trajectory, perturbations, returns, and energy-limited control. Static baselines can miss the object that matters Model claim vs. mathematical claim (made separable). The mathematical claim is only that Rβ is well-defined and computable under a frozen metric. The biological claim — that repeated unresolved displacement predicts slow drift of the defended reference, so that chronic illness can be successful return to a displaced destination — is an effective-model hypothesis whose death condition is held-out failure to predict drift, relapse, recovery time, or vulnerability beyond acute displacement. The first is proven; the second is killable. 6.1 Timescale separation The model is identifiable only if its layers are separated: τ_x τ_R τ_b τ_E (fast state; residue ≪ ≪ ≪ accumulation/decay; slow attractor drift; energy-capacity adaptation). If these scales collapse, the model becomes unfalsifiably flexible. Page 8 of 35

minimal factorised model is ΔZ_A = η · δ · s(A,C) · b_t − χ_A, with learning rate η, prediction error δ, address–context overlap s(A,C), update/writeability gate b_t, and decay/extinction χ_A. Other functional forms are possible; this one is useful because it separates retrieval, overlap, access, and write-back into independently testable factors. The load-bearing variable is b_t, the writeability gate — not the word “awareness.” Operational definition (deliberately narrow). The model does not require a theory of consciousness. It requires only that retrieved residual states differ in their writeability under corrective prediction error. “Awareness,” in this framework, is the system-level condition under which a retrieved residual becomes writable. IDA may stand for I Develop Awareness, but the formal claim is the narrower one: residue-gated support improves writeable corrective updating. Interference caveat. If repeated updates are empirically found to corrupt prior records (catastrophic forgetting), updates may be constrained toward the orthogonal complement of the non-target context

question above: birth/death dynamics as a substrate analogue of record creation and erasure; well- posedness discipline (finite state spaces, positivity, Doeblin-style coupling, explicit acceptance gates); invariant records as the physical content surviving relabelling; and tabletop falsifiers (metamaterial dispersion, Josephson-junction switching histograms, biological switching assays) that can kill the extension without any clinical interpretation. Corrected status of α = 1/12. The α = 1/12 dispersion coefficient and the sextic lever are preserved strictly as conjectural UV gates. They are never used to support Aczél composition, recovery to induce a state. It tests whether recovery geometry can be made controllable by gating gentle, non- novel perturbations on residue reduction. No therapeutic claim is made; throughout, IDA is a control- prediction and falsification platform. 9.1 Device definition A proposed soft headband with dry EEG, optional peripheral sensors, an estimator, a controller, conservative actuators, and a full data logger. Candidate actuators — rhythmic audio, bone-conduction, gentle haptic cues, low-intensity visual pacing, paced-breathing cues, or other established modalities

• The controller cannot expand its own safety envelope. • The target is cleaner return, not a larger felt response. • Every read, decision, and write is logged for offline falsification. 9.4 The exact claim IDA makes IDA prediction. If recovery geometry is controllable, a residue-gated closed loop should move the organism toward cleaner return faster, more safely, or more durably than a matched open-loop schedule at equal or lower exposure. If it does not, IDA dies as a mechanism even if the read-side metrics remain useful. Safety statement. No unsupervised clinical use follows from this manuscript. Any prospective study would require independent safety review, locked stimulation limits, sham and open-loop controls, adverse-event monitoring, pre-registration, and appropriate regulatory classification. 9.5 Reference implementation — real, runnable, and deliberately safe This paper does not describe the engine in prose and leave the building to faith. Appendix A is a leaky residue integrator (the same object as the negentropy deficit of Section 6.4), the multi-window return score, the residue gate with hysteresis, debouncing, rate-limiting, a confidence gate, and a hard safety cap. It runs. On synthetic perturbation-and-recovery data it reproduces the predicted signature with no tuning beyond the stated defaults: during a cleanly resolving perturbation the residue stays low and the controller advances; during a lingering perturbation the residue rises past the retreat gate and the controller holds and reduces. A reader can run it today on recorded signals — that is Gate 0 — with no hardware at all.

nothing in it points an unreviewed stimulator at a person. Simple, auditable, and — if the wager is right — experimentally informative; and if it is wrong, the same code is how someone shows it. 10. Empirical programme and falsification ladder Each claim is earned in order. No write-side claim is licensed until the read-side input carries independent, held-out information. Gate Pass condition Death condition 0 — read-only Return features d(t), Rβ(t), S_Δ(t) beat static EEG/physiology features on held-out prediction. Return features add no held-out information beyond static baselines. 1 — directed coupling history. Only co-movement (shared startle) with no directed, predictive coupling. 2 — closed-loop advantage Residue-gated actuation beats matched open- loop and sham at equal or lower exposure. On Gate 1. Co-movement is mere correlation and would let a flawed metric survive: a generic startle perturbs EEG and HRV together. The strengthened condition is directional — the recovery residue must carry transfer entropy toward the future state of the independent marker, predicting its trajectory beyond the marker's own history. (Transfer-entropy estimation is data-hungry, which reinforces the Gate-0 data requirement below.) 10.1 Gate 0 pre-registration skeleton Item Specification

The first real deliverable is a dataset, not an analysis. Gate 0 requires within-subject perturbation- and-recovery epochs with clean enough signal to estimate d(t) and Rβ(t) and an outcome that return geometry should predict. Most public EEG corpora are not built this way. Until such a dataset exists, the empirical wager cannot be placed — so the programme's next concrete step is one clean perturbation–recovery dataset with held-out structure. Everything downstream is conditional on it. Required data specification (illustrative minimums, not prescriptions). A first decisive dataset need not be clinical; the initial target is metric validity, not treatment response. Outcome An independently measured return/drift proxy or validated performance scale at, e.g., 0, 15, and 60 min post-perturbation. Primary question Do d(t), Rβ(t), and S_Δ(t) predict future return or drift beyond static band power and variance, under subject-wise held-out validation? 11. What would kill the programme A synthesis this broad must carry death conditions at every level, and must not be protected by rhetorical migration from one domain to another. 1. Mathematical death: a claimed composition law fails the representation hypotheses, or the proposed generator fails to linearise composition. 2. Coordinate death: lawful-coordinate models do not improve held-out compositional prediction over native-coordinate baselines. 3. Recovery death: return metrics add no held-out information beyond static baselines. 4. Residue death: Rβ does not predict drift, relapse, recovery time, or vulnerability beyond acute displacement. 5. Awareness-model death: addressed-residual variables fail to distinguish retrieval, access, prediction error, and write-back. 6. IDA read-side death: the headband cannot estimate stable return coordinates with adequate reliability. 7. IDA write-side death: residue-gated actuation fails to beat matched open-loop and sham. Page 15 of 35

symptom-gated intervention. And ANDY, the attractor-normalised drift yield (Section 6.2): the apparent recovery that is really baseline drift, the error term a frozen baseline exists to expose. None is claimed as a law of nature; each is a defined object with a death condition. Final wager. IDA is not a machine that adds awareness from outside. It is a bounded-domain controller that asks whether writeable corrective updating improves when unresolved perturbation is no longer amplified and cleaner return can be written back to the state that actually needed repair. The name states the intent plainly: I Develop Awareness. Should the wager survive the ladder