A General Boundary-First Framework for Scientific Measurement, Controls, Model Comparison, and Anomaly Claims

DOI: To be assigned

John Swygert

August 15, 2026

Abstract

Science frequently advances through discrepancies. An observation departs from an expectation, a detector reports an unexplained signal, a biological response persists after known mechanisms are modeled, a material behaves differently under apparently equivalent conditions, or an experimental apparatus produces a residual that the accepted model does not immediately explain. Such discrepancies are indispensable to discovery. They are also among the easiest scientific observations to overinterpret.

A residual is not automatically an anomaly. An anomaly is not automatically evidence for a new mechanism. A new mechanism is not automatically evidence for a new theory, and evidence against an existing model is not automatically evidence for any particular replacement.

This paper proposes a general boundary-first framework for determining how far a scientific discrepancy is entitled to travel along that inferential chain. The framework combines established principles of measurement uncertainty, calibration, metrological traceability, experimental control, model comparison, preregistration, equivalence testing, replication, and falsifiability with a relational analysis of experimental boundaries, pathways, receivers, timescales, and control interventions.

Its central proposition is simple:

Before an unexplained residual can be interpreted, the scientific architecture capable of producing that residual must itself be bounded, measured, challenged, and made capable of failure.

Every experimental control suppresses some relations while potentially creating others. Every detector measures through a finite receiver architecture. Every boundary is selective over some combination of carrier, energy, direction, phase, state, frequency, geometry, and timescale. Every model discrepancy is therefore conditional upon the declared system, measurement chain, comparator, uncertainty structure, and observational receiver.

A rigorous residual claim consequently requires more than subtraction between observation and prediction. It requires a fixed system boundary; a defined measurand; calibrated and sufficiently sensitive receivers; quantified uncertainty; explicit environmental and instrumental pathways; control-induced pathway analysis; the strongest relevant comparator; prospective decision criteria; bounded negligibility for excluded routes; visible null and failed outcomes; and replication under conditions capable of separating apparatus-specific effects from persistent structure.

The framework defines a hierarchy from observation to validated measurement, model discrepancy, robust residual, mechanism-specific anomaly, theory-discriminating evidence, and possible new physical effect. It introduces the concepts of Relational Control Accounting, Bounded Negligibility, Residual Qualification, and the No-Inference-Jump Rule. It also distinguishes a receiver-null from universal absence and a model-relative residual from unexplained physical ontology.

The framework is intentionally theory-neutral. It is compatible with established measurement science and may be applied without accepting TSTOEAO or any other proposed ontology. Within TSTOEAO, however, it provides a particularly important constraint: a proposed substrate, hidden pathway, displaced cost, or unregistered relation may never be introduced after observing a discrepancy merely to rescue the theory. If such a mechanism is scientifically real, it must eventually produce a prospectively distinguishable consequence.

The purpose is therefore neither to make anomalous claims easier nor to make them impossible. It is to establish what must happen before the residual earns the right to mean more than residual.

Keywords: scientific method; residual; experimental control; boundary conditions; measurement uncertainty; metrological traceability; model comparison; anomaly detection; falsifiability; preregistration; replication; equivalence testing; receiver architecture; TSTOEAO; bounded negligibility

1. Purpose

Scientific discovery often begins with the sentence:

Something remains unexplained.

That sentence is powerful precisely because it is incomplete.

An unexplained signal may indicate:

• measurement error;
• calibration drift;
• environmental contamination;
• an overlooked interaction;
• inappropriate preprocessing;
• an incomplete model;
• a poorly chosen model;
• a boundary effect;
• detector backaction;
• sampling variation;
• hidden dependence among variables;
• a control-induced artifact;
• a genuine but conventional mechanism not represented in the analysis;
• or, in rare cases, an effect that requires an extension of existing physical understanding.

The scientific problem is therefore not merely to detect a discrepancy.

It is to determine which inferential claims that discrepancy can support.

The framework developed here addresses that problem across scientific disciplines.

It does not replace domain-specific experimental standards. A particle detector, clinical trial, astronomical observatory, chemical assay, ecological field study, materials experiment, and computational model require different instrumentation, error structures, statistical methods, and causal assumptions.

What can be shared is a deeper methodological grammar:

system → boundary → pathway → intervention → receiver → measurement → uncertainty → comparator → residual → challenge → replication → inference

The integrity of the inference depends upon the integrity of the entire chain.

2. The Fundamental Distinction: Difference Is Not Explanation

Let a registered observable be O, and let a qualified comparator predict:

[ \hat{O}_0. ]

A raw discrepancy can be written:

[ r = O_{\mathrm{obs}}-\hat{O}_0. ]

The existence of:

[ r\neq0 ]

does not by itself identify its cause.

It establishes only that the registered observation and the registered comparator differ by the measured amount under the declared conditions.

Even when:

[ |r|>u, ]

where u represents an appropriate uncertainty bound, the scientific interpretation remains conditional.

The discrepancy may belong to the measurement architecture rather than the target phenomenon.

It may belong to an omitted conventional variable.

It may expose an incorrect boundary assumption.

It may reveal model inadequacy.

It may represent a genuine unknown mechanism.

The subtraction discovers a difference.

Science must still discover the reason for the difference.

This distinction is the foundation of Residual Qualification.

3. The Scientific Claim Ladder

A major source of scientific overstatement is movement across several inferential levels without establishing the intermediate steps.

The following levels should therefore remain distinct.

Level 0 — Observation

A detector, instrument, observer, database, or computational procedure produces a record.

No claim concerning its physical meaning is yet required.

Level 1 — Validated Measurement

The observation survives applicable calibration, uncertainty, quality-control, traceability, preprocessing, and measurement-validity requirements.

The claim becomes:

the registered quantity was measured within the stated conditions and uncertainty.

Level 2 — Model Discrepancy

The validated measurement differs meaningfully from a specified comparator.

The claim becomes:

the registered model does not fully reproduce this registered result under the tested conditions.

Level 3 — Robust Residual

The discrepancy survives registered challenges involving:

• calibration;
• instrumental response;
• boundary pathways;
• environmental variables;
• alternative preprocessing;
• conventional confounds;
• uncertainty treatment;
• comparator improvement;
• and applicable replication.

The result is now a qualified residual.

Level 4 — Mechanism-Specific Anomaly

The surviving residual follows a structure expected from one proposed mechanism more strongly than from qualified alternatives.

Direction, scaling, timing, pathway, receiver consequence, or intervention response begins discriminating mechanisms.

Level 5 — Theory-Discriminating Evidence

Competing theories generate different prospective predictions, and the observation preferentially supports one prediction under conditions established before outcome access.

Level 6 — Evidence for a New Physical Effect

Established explanations have been subjected to unusually strong attack, independent systems reproduce the effect, and the phenomenon displays a stable causal or quantitative structure not adequately represented within the existing framework.

Level 7 — Ontological Interpretation

Claims concerning a new substrate, field, entity, dimensional structure, or fundamental ontology require additional evidence beyond the demonstration of an unexplained effect.

The movement:

[ \text{Residual}\rightarrow\text{New Physics} ]

is therefore not one inference.

It is a sequence of scientific burdens.

4. The No-Inference-Jump Rule

The general rule is:

A result may support the next inferential level only after the requirements separating the two levels have been satisfied.

A model failure is evidence against the adequacy of that model under the tested conditions.

It is not automatically evidence for the preferred alternative.

Similarly:

[ \neg A ]

does not imply:

[ B. ]

This elementary logical distinction becomes critically important when studying anomalies.

Suppose conventional hypothesis H_0 predicts one outcome while a speculative hypothesis H_1 exists.

If observation D is inconsistent with H_0, the conclusion:

[ D\not\sim H_0 ]

does not establish:

[ D\sim H_1. ]

For H_1 to gain distinctive support, it should supply some consequence not already guaranteed merely by the failure of H_0.

The stronger structure is:

[ H_0\rightarrow P_0 ]

and:

[ H_1\rightarrow P_1, ]

where:

[ P_0\neq P_1 ]

under a prospectively defined test.

Only then can the result discriminate the hypotheses rather than merely embarrass one of them.

5. The Boundary Comes Before the Residual

Every measurement occurs within a declared or implicit boundary.

The experimental boundary may include:

• apparatus geometry;
• shielding;
• chamber walls;
• atmosphere or vacuum;
• electrical grounding;
• thermal environment;
• optical isolation;
• software filters;
• sample preparation;
• control conditions;
• temporal windows;
• detector acceptance;
• inclusion and exclusion criteria;
• or the mathematical boundary of the model itself.

A boundary should not be treated merely as scenery surrounding the experiment.

Operationally, a boundary may:

[ \text{admit}, ]

[ \text{suppress}, ]

[ \text{redirect}, ]

[ \text{delay}, ]

[ \text{convert}, ]

[ \text{attenuate}, ]

or

[ \text{differentiate} ]

possible interactions.

Consequently, the statement:

the system was isolated

is scientifically incomplete.

Isolated from what?

At what frequency?

For which carrier?

Across what direction?

At what energy?

For what duration?

Through what material?

At what sensitivity?

To which receiver?

No physical experimental boundary should be assumed absolute merely because it is highly effective against the interaction for which it was originally designed.

6. Every Control Removes Relations—and Creates Relations

Experimental controls are indispensable.

Yet controls are not metaphysically neutral.

Installing shielding may reduce electromagnetic coupling while altering:

• capacitance;
• grounding;
• thermal behavior;
• mechanical stress;
• vibration;
• reflected fields;
• geometry;
• gas flow;
• pressure;
• optical access;
• detector position;
• or electrical return paths.

Changing a control fluid can alter chemistry.

Adding a placebo can alter behavior.

Changing software filtering can suppress noise while altering temporal structure.

Moving a sensor can reduce one interference source while changing its environmental exposure.

Therefore:

A control intervention should be analyzed not only for the pathway it suppresses, but for the pathways it creates, modifies, or redirects.

This does not invalidate controlled experimentation.

It strengthens it.

The appropriate question is not:

Did the control change anything?

Of course it did.

The appropriate question is:

Were the unintended changes sufficiently measured, matched, modeled, randomized, counterbalanced, or bounded that they cannot plausibly produce the registered effect?

This is Relational Control Accounting.

7. Relational Control Accounting

Let the experimental architecture be represented by:

[ \mathcal{A}

(B,P,R,T,\tau,C), ]

where:

• B = declared boundary;
• P = relevant pathway set;
• R = registered receiver or receiver set;
• T = applicable transformations;
• \tau = relevant temporal structure;
• C = control architecture.

When a control changes from:

[ C_0\rightarrow C_1, ]

the experiment should not automatically assume:

[ P_0=P_1. ]

Instead the intervention may produce:

[ \mathcal{A}_0\rightarrow\mathcal{A}_1. ]

A rigorous control analysis therefore asks:

1. Which intended pathway changed?
2. Which other pathways could have changed?
3. Which of those pathways are measurable?
4. Which can be physically bounded?
5. Which remain unresolved?
6. Could any unresolved pathway generate the sign, magnitude, timing, or scaling of the observed result?

This converts control design from a binary vocabulary—

controlled / uncontrolled

—into a physically explicit account of what the control actually did.

8. Bounded Negligibility

Science cannot measure every conceivable interaction.

Nor should it be required to.

A methodology demanding the elimination of every logically possible alternative would make experimental inference impossible.

The solution is not exhaustive elimination.

It is bounded negligibility.

Let a possible pathway p_k generate some maximum receiver-visible contribution:

[ \Delta O_k. ]

For a registered domain of physically allowable parameters \Theta_k, define an upper plausible bound:

[ U_k

\sup_{\theta\in\Theta_k} |\Delta O_k(\theta)|. ]

Let the experiment define a scientifically meaningful decision scale:

[ \delta. ]

A pathway may be treated as negligible for the registered claim when:

[ U_k<\delta, ]

subject to the domain’s uncertainty and safety requirements.

This statement does not mean:

[ p_k=0. ]

It means:

[ p_k ]

cannot account for a scientifically consequential fraction of the registered effect under the stated assumptions and measurement limits.

This distinction is essential.

A scientist does not need to prove universal nonexistence of every alternative.

The scientist must show that qualified alternatives are too small, too slow, too weakly coupled, improperly directed, temporally incompatible, geometrically excluded, statistically inadequate, or otherwise incapable of explaining the registered observation.

9. The Receiver Is Part of the Scientific Claim

A physical event and a registered event are not synonymous.

Measurement requires a receiver.

A receiver may be:

• a photodetector;
• antenna;
• calorimeter;
• mass spectrometer;
• microscope;
• sequencing platform;
• human observer;
• questionnaire;
• imaging system;
• telescope;
• computer algorithm;
• particle detector;
• or another physical or informational measurement architecture.

Every receiver has finite:

• bandwidth;
• sensitivity;
• dynamic range;
• spatial resolution;
• temporal resolution;
• selectivity;
• noise;
• saturation behavior;
• preprocessing;
• and failure modes.

Consequently:

A receiver-null establishes non-detection by the qualified receiver under the registered conditions. It does not, by itself, establish universal physical absence.

The correct scientific statement may be:

[ |\Delta O|<\delta_R ]

for receiver R,

rather than:

[ \Delta O=0 ]

for reality universally.

This protects science from two opposite errors.

The first is claiming that an undetected phenomenon definitely exists.

The second is claiming that a receiver’s inability to detect something proves that nothing physically occurred.

Neither inference follows automatically.

10. Measurement Before Interpretation

A residual is meaningful only to the extent that the measurements from which it is constructed are meaningful.

Measurement science therefore precedes anomaly interpretation.

At minimum, applicable work should address:

• measurand definition;
• calibration;
• reference standards;
• uncertainty;
• repeatability;
• drift;
• linearity;
• sensitivity;
• detection limits;
• environmental dependence;
• data acquisition;
• preprocessing;
• traceability where relevant;
• and instrument-specific failure modes.

NIST measurement guidance emphasizes quantitative uncertainty as an integral component of measurement reporting, while metrological traceability requires an appropriate documented chain connecting measurements to recognized references [6,7].

ISO/IEC 17025 similarly establishes requirements concerned with competent and consistent testing and calibration and the production of valid laboratory results [8].

The general principle is straightforward:

A residual smaller than the unresolved architecture of the measurement cannot carry a larger scientific interpretation than the measurement itself can support.

11. Residual Qualification

A useful distinction should be drawn between a raw residual and a qualified residual.

For observable O_j:

[ r_j

O_{j,\mathrm{obs}}

O_{j,\mathrm{model}}. ]

The raw residual becomes a scientifically qualified residual only after applicable tests address the major attribution classes:

[ \mathcal{K}

{ \text{calibration}, \text{instrument}, \text{environment}, \text{boundary}, \text{control}, \text{analysis}, \text{sampling}, \text{model}, \text{unknown} }. ]

These categories should not automatically be treated as numerically additive. Different scientific domains contain different variables, interactions, nonlinearities, and error structures.

Where a physically justified additive decomposition exists, one may estimate contributions such as:

[ r

r_{\mathrm{cal}} + r_{\mathrm{inst}} + r_{\mathrm{env}} + r_{\mathrm{boundary}} + r_{\mathrm{model}} + \epsilon. ]

Where no such decomposition is justified, the terms should remain separate hypotheses rather than being forced into a false scalar accounting system.

A qualified residual is therefore not:

whatever the model failed to explain.

It is:

a registered discrepancy that remains after the applicable measurement, boundary, control, comparator, uncertainty, and analysis challenges have been passed.

Even then, its cause may remain unknown.

That is scientifically acceptable.

12. The Strongest Comparator Rule

Novel hypotheses must compete against the strongest relevant conventional explanation, not the weakest convenient alternative.

A comparator should therefore include, as appropriate:

• established causal variables;
• known nonlinearities;
• environmental covariates;
• instrument response;
• conventional boundary effects;
• established kinetics;
• recognized interaction terms;
• appropriate nuisance parameters;
• uncertainty;
• and accepted corrections.

A speculative model gains little from outperforming an intentionally impoverished null.

Model comparison should also distinguish descriptive flexibility from predictive power.

Increasing the number of adjustable parameters generally increases the ability of a model to accommodate observed data. Consequently, fitting the same observations used to construct a model is weaker evidence than successful prediction of untouched observations.

Akaike’s work on statistical model identification formalized one influential approach to balancing model fit and complexity [15]. Different disciplines may appropriately use other information criteria, cross-validation, Bayesian comparison, likelihood methods, held-out prediction, or mechanistically constrained comparison.

The universal requirement is not one statistical technique.

It is:

Do not call a theory scientifically superior merely because it can be adjusted to reproduce what is already known.

13. Before the Outcome

Exploratory science and confirmatory science are both legitimate.

They answer different questions.

Exploration asks:

What pattern might be here?

Confirmation asks:

Does a prospectively specified pattern survive an independent test?

The distinction becomes especially important for anomaly claims.

After seeing an unusual outcome, investigators may legitimately formulate a new explanation.

What they cannot legitimately do is silently convert that explanation into a prediction that supposedly preceded the observation.

Preregistration and related prospective methods help maintain the distinction between prediction and postdiction [11].

For a strong confirmatory anomaly test, applicable elements should be fixed before confirmatory outcome access:

[ P_d

(B,I,O,R,\tau,N,U,\Delta,F), ]

where:

• B = system boundary;
• I = intervention;
• O = registered observable;
• R = receiver;
• \tau = time window;
• N = strongest comparator;
• U = uncertainty treatment;
• \Delta = decision threshold or equivalence margin;
• F = falsification condition.

The more of these elements that remain adjustable after the outcome is known, the weaker the confirmatory interpretation becomes.

14. The Post-Hoc Rescue Prohibition

A theory that can absorb every possible result cannot meaningfully fail.

Suppose an experiment produces a residual.

A hypothesis may not respond by inventing after the fact:

• an invisible pathway;
• an unmeasured energy account;
• a hidden boundary;
• an unspecified receiver;
• a deferred cost;
• a new parameter;
• an unconstrained coupling;
• or an unknown correction term

and then count the resulting fit as confirmation.

Such mechanisms may legitimately become new hypotheses.

They must then face new tests.

The required sequence is:

[ \text{unexpected result} \rightarrow \text{new hypothesis} \rightarrow \text{new locked prediction} \rightarrow \text{new data}. ]

Not:

[ \text{unexpected result} \rightarrow \text{new adjustable explanation} \rightarrow \text{retroactive confirmation}. ]

This rule applies equally to conventional and unconventional theories.

No theory receives unlimited lives.

15. Null Results and Equivalence

A null result must also be interpreted carefully.

Failure to detect a statistically significant difference does not necessarily establish scientifically meaningful equivalence.

If the question is whether two conditions are sufficiently similar for a particular claim, a predefined equivalence margin may be more appropriate.

Let:

[ \Delta

O_A-O_B. ]

Instead of testing only whether:

[ \Delta=0 ]

can be rejected, the scientifically relevant question may be whether:

[ -\delta<\Delta<+\delta, ]

where \delta is a prospectively justified equivalence bound.

Equivalence-testing methods formalize this distinction [14].

This matters directly to boundary and anomaly science.

If an experimental control produces “no significant difference,” that alone does not necessarily establish that the control-induced pathway is negligible.

The experiment should be capable of excluding an effect large enough to account for the phenomenon under investigation.

16. Replication Versus Reproduction

Scientific confidence increases when an effect survives multiple forms of re-examination.

The National Academies distinguishes reproducibility, involving consistent results from the same data and computational procedures, from replicability, involving consistent results across studies addressing the same scientific question using new data [9].

For robust anomaly claims, both concepts matter.

An effect should ideally survive:

Computational reproduction

The same data and declared analysis reproduce the reported result.

Analytical robustness

Reasonable alternative analysis choices do not make the phenomenon arbitrarily appear or disappear.

Instrumental replication

Different instruments or receivers recover the relevant structure where scientifically feasible.

Environmental replication

The effect survives controlled changes in environmental conditions or behaves according to a predicted environmental dependence.

Laboratory replication

Independent investigators reproduce the phenomenon.

Architectural replication

A predicted relational structure appears in a materially different system where the hypothesized mechanism says it should.

The final category can be especially powerful.

If the effect disappears whenever one apparatus is replaced, the apparatus deserves renewed scrutiny.

If a quantitatively predicted structure survives across independent architectures, the space of conventional artifacts narrows considerably.

17. Residual Shape Matters

Not all residuals carry equal information.

A single unexplained offset is generally less discriminating than a structured residual possessing prospectively predicted:

• sign;
• scaling;
• threshold;
• periodicity;
• angular dependence;
• frequency dependence;
• phase relation;
• temporal ordering;
• spatial distribution;
• dose response;
• boundary dependence;
• receiver dependence;
• or conservation-linked counterpart.

Suppose a hypothesis predicts only:

[ r\neq0. ]

Almost any unexplained discrepancy can satisfy the prediction.

A stronger hypothesis predicts:

[ r(x)

f(x;\theta), ]

with locked structural properties.

Stronger still is a prediction in which an intervention reverses, suppresses, enhances, or redirects the residual:

[ I_1\rightarrow r_1, ]

[ I_2\rightarrow r_2, ]

with:

[ r_1-r_2 ]

specified in advance.

The more structured the prediction, the less room remains for retrospective reinterpretation.

18. Boundary-Swap Experiments

One particularly useful experimental strategy follows directly from the boundary-first framework.

If an unexplained effect may arise from either:

1. the proposed target phenomenon, or
2. an unintended property of the experimental boundary,

then change the boundary while preserving the target condition as closely as practicable.

Let:

[ B_1\neq B_2 ]

while:

[ E_1\approx E_2 ]

for the registered target inputs.

Then compare:

[ O(B_1) ]

and:

[ O(B_2). ]

A useful boundary swap might alter:

• material;
• thickness;
• geometry;
• grounding;
• shielding topology;
• orientation;
• thermal conductance;
• optical properties;
• magnetic permeability;
• electrical conductivity;
• mechanical mounting;
• or other relevant characteristics.

The hypothesis should state whether the anomaly is expected to:

• remain invariant;
• scale;
• vanish;
• reverse;
• migrate;
• or transform.

The boundary thus becomes an experimental variable rather than an invisible assumption.

19. Receiver-Swap Experiments

The same principle applies to receivers.

Where feasible, test the effect through:

[ R_1\neq R_2. ]

If two receivers operate through substantially different physical principles yet recover the same predicted structure, certain classes of instrument-specific artifact become less plausible.

Conversely, if an effect exists only through one receiver architecture, that dependence itself becomes scientifically informative.

The correct conclusion need not be:

the effect is false.

It may instead be:

the effect is receiver-conditioned.

The next experiment should then determine why.

20. Control-Swap Experiments

A control should also be challenged by alternative implementations.

Suppose control C_1 suppresses suspected pathway p.

A stronger design may introduce control C_2, which suppresses the same pathway through a different physical architecture.

If:

[ C_1\rightarrow O^* ]

and:

[ C_2\rightarrow O^*, ]

while their major unintended relations differ, confidence increases that O^* is not merely an artifact specific to C_1.

This is particularly valuable when studying very small residuals.

One control can accidentally manufacture an effect.

Two physically different controls manufacturing the same quantitatively structured effect becomes a more constrained explanation.

21. Scientific Burden Should Scale With Claim Distance

Not every scientific claim requires identical evidence.

The evidentiary burden should scale with the distance between observation and interpretation.

A routine calibration claim may require ordinary laboratory standards.

A claim of an unknown but conventional mechanism requires more.

A claim of a new interaction requires stronger independent replication.

A claim of a new fundamental substrate requires stronger evidence still.

This may be written conceptually as:

[ \mathcal{B} \propto D_I, ]

where:

• \mathcal{B} = required evidentiary burden;
• D_I = inferential distance between measured observation and proposed interpretation.

This is not a universal numerical law.

It is a methodological principle.

The farther the interpretation extends beyond what was directly measured, the more independent bridges must support it.

22. Application Across Scientific Domains

The framework is intentionally carrier-agnostic but not domain-indifferent.

Physics

Boundary fluxes, detector response, conservation accounting, field leakage, environmental coupling, calibration, and competing physical mechanisms must be addressed before a residual is interpreted as evidence of new physics.

Chemistry

Reaction pathways, contamination, solvent effects, catalysts, vessel surfaces, temperature history, phase boundaries, and analytical instrumentation may generate apparent discrepancies.

Materials Science

Interfaces, substrate preparation, strain, defects, encapsulation, geometry, process order, and measurement architecture may determine observable properties.

Biology

Population heterogeneity, batch effects, environmental history, biological variability, measurement platform, selection criteria, and hidden covariates require explicit treatment.

Medicine

Clinical populations, endpoints, treatment adherence, comparator selection, measurement validity, missing data, confounding, and clinically meaningful effect thresholds constrain interpretation.

Psychology and Social Science

Operational definitions, sampling, context, observer effects, measurement instruments, analytical flexibility, effect-size interpretation, and replication become part of the relevant boundary architecture.

Astronomy

Instrument response, calibration, background subtraction, selection functions, atmospheric effects, reconstruction algorithms, and competing astrophysical models constrain anomalous interpretation.

Computation and Artificial Intelligence

Training distribution, preprocessing, architecture, random seeds, evaluation datasets, leakage, metric selection, hardware, software version, and evaluator structure may function as experimental boundaries and receivers.

The common grammar survives.

The physical meaning of each term does not have to be identical.

23. Relationship to TSTOEAO

This methodological framework does not require acceptance of TSTOEAO.

Its measurement, calibration, uncertainty, model-comparison, preregistration, replication, equivalence, and falsification requirements are independently defensible scientific practices.

TSTOEAO nevertheless supplied an important motivation for making the relational architecture explicit.

Its foundational expression:

[ V=E\times Y ]

treats realized outcome V as conditioned not only by available Energy or Opportunity E, but by Encoded Equilibrium Y: the architecture of boundaries, pathways, transformations, receiver access, correction, cost, and relevant history.

Its recurring sequence:

[ \text{Gradient} \rightarrow \text{Boundary} \rightarrow \text{Correction} \rightarrow \text{Cost} \rightarrow \text{Equilibrium} ]

encourages investigation of where an observed difference is physically expressed and where compensating consequences may occur.

The scientific requirement, however, is crucial:

TSTOEAO may not use the relational richness of reality as permission to explain every residual.

If a hidden pathway is proposed, it must become constrainable.

If a displaced cost is proposed, it must become locatable.

If receiver dependence is proposed, changing the receiver should eventually produce a testable consequence.

If a boundary effect is proposed, changing the boundary should eventually matter in a predictable manner.

If substrate coupling is proposed, the hypothesis must eventually distinguish itself from conventional interactions.

The framework therefore constrains TSTOEAO at least as strongly as it constrains competing theories.

That is precisely what a scientific framework should do.

24. Relationship to Latching Onto Space

The paper Latching Onto Space: Asymmetric Electrostatic Propulsion, Relational Momentum Accounting, and a TSTOEAO Substrate-Coupling Hypothesis addresses a specific speculative possibility.

Its central scientific question is not whether unexplained thrust should automatically be interpreted as substrate coupling.

It is the opposite:

If a reproducible momentum or force residual remained after sufficiently complete conventional accounting, what experimental architecture could distinguish a substrate-coupling hypothesis from known electromagnetic, electrostatic, thermal, mechanical, environmental, and instrumental mechanisms?

The present framework establishes the methodological burden that precedes such an inference.

Thus:

If a true residual exists, Latching Onto Space asks what it might mean and how it could be tested.

But first the residual must become qualified.

25. Relationship to Inside Is Not Isolated

The companion paper Inside Is Not Isolated: Faraday Boundaries, Channel-Selective Shielding, and the Relational Architecture of Experimental Control Through TSTOEAO addresses the preceding problem.

A Faraday enclosure strongly suppresses particular electromagnetic relationships.

It does not remove the enclosed apparatus from physical reality.

The system may remain coupled through:

• gravity;
• vibration;
• heat;
• mechanical support;
• electrical grounding;
• acoustic transmission;
• residual electromagnetic leakage;
• material stress;
• particle penetration;
• gas or pressure pathways;
• and other domain-dependent interactions.

The scientifically useful lesson is not that everything interacts with everything in an experimentally important way.

It is that each plausible route should either be measured, controlled, modeled, or bounded below relevance.

Thus:

Before one dares call an unexplained difference a true residual, Inside Is Not Isolated asks how thoroughly the boundary and controls have been attacked.

The present paper generalizes that requirement beyond Faraday shielding.

26. The Three-Paper Methodological Hierarchy

The resulting architecture can now be stated clearly.

Paper I — General Method

Before the Residual

What standards must any scientific discrepancy satisfy before interpretation?

It establishes:

• measurement qualification;
• boundary accounting;
• relational control accounting;
• bounded negligibility;
• receiver qualification;
• comparator strength;
• prospective locking;
• residual qualification;
• replication;
• and inferential discipline.

Paper II — Boundary Challenge

Inside Is Not Isolated

How aggressively must a supposedly isolated physical system be challenged before its remaining signal can be called unexplained?

It specializes the general method to Faraday boundaries and channel-selective shielding.

Paper III — Residual Interpretation

Latching Onto Space

If a genuine residual survives those challenges, what might it mean, and what experiment could discriminate the proposed substrate-coupling hypothesis from conventional physics?

The papers therefore do not contradict one another.

They form an evidentiary sequence:

[ \boxed{ \text{Measure} \rightarrow \text{Bound} \rightarrow \text{Challenge} \rightarrow \text{Qualify} \rightarrow \text{Replicate} \rightarrow \text{Interpret} \rightarrow \text{Discriminate} } ]

27. Minimum Residual Qualification Record

A serious residual claim should preserve, at minimum, the following record.

System

What exactly is inside the claim?

Boundary

What separates the declared system from its environment?

Measurand

What physical or informational quantity is being measured?

Receiver

What instrument or observational architecture registers it?

Calibration

How was the measurement architecture validated?

Uncertainty

What uncertainty applies to the measured quantity and derived residual?

Pathways

Which conventional interactions can transmit influence into, out of, or across the declared boundary?

Controls

Which pathways are controlled, and what new relations do those controls introduce?

Negligibility bounds

Why may excluded pathways be considered incapable of explaining the effect?

Comparator

What is the strongest relevant established model?

Residual

What remains after the registered comparison?

Prediction status

Was the claimed structure specified before or after seeing the result?

Decision criterion

What magnitude, direction, structure, or equivalence region determines the outcome?

Falsifier

What result would count against the proposed explanation?

Replication

Does the result survive new data, independent receivers, independent controls, or independent laboratories?

Evidentiary level

Is the result an observation, measurement, discrepancy, robust residual, mechanism-specific anomaly, theory-discriminating result, or evidence for new physics?

This record should remain visible regardless of whether the experiment succeeds.

28. What a True Residual Means

The phrase true residual should be used carefully.

It does not mean:

a quantity proven to come from unknown physics.

It means something narrower and more useful:

A reproducible, receiver-qualified discrepancy relative to a declared comparator that persists after the applicable measurement uncertainties, known conventional mechanisms, boundary pathways, control-induced relations, analytical alternatives, and registered confounds have been either corrected, experimentally challenged, or bounded below scientific relevance.

The word true therefore modifies the status of the discrepancy.

It does not identify its ontology.

A true residual may still arise from an unknown conventional mechanism.

That is not a failure.

That is precisely where discovery begins.

29. Falsification and Failure

The framework itself must be capable of constraining claims.

It therefore rejects several forms of reasoning.

A claimed anomaly is weakened when:

• it disappears under improved calibration;
• it scales with an instrumental artifact;
• it follows environmental contamination;
• it depends on one undocumented preprocessing choice;
• it disappears under a stronger comparator;
• its magnitude lies within applicable uncertainty;
• an allegedly negligible pathway proves large enough to explain it;
• independent replication repeatedly fails under qualified conditions;
• or the proposed mechanism survives only through post-hoc modification.

A particular speculative explanation is weakened when:

• the residual survives but does not follow its predicted structure;
• changing the hypothesized pathway does not change the result;
• changing the boundary produces the opposite effect from prediction;
• a qualified receiver fails to observe a predicted consequence;
• or a conventional model predicts the same result without the additional mechanism.

These are productive outcomes.

A framework that cannot lose cannot learn.

30. The Larger Methodological Principle

Scientific rigor is sometimes described as skepticism toward extraordinary claims.

That description is incomplete.

Rigor is not merely skepticism toward novelty.

It is skepticism applied symmetrically to:

• the novel explanation;
• the conventional explanation;
• the measurement;
• the instrument;
• the boundary;
• the control;
• the analysis;
• and the assumptions connecting them.

The objective is not to protect established theory from anomalies.

Nor is it to protect anomalies from established theory.

The objective is to construct an experiment in which reality has fewer places to hide from the question being asked.

That requires boundaries.

But it also requires examining the boundaries themselves.

It requires controls.

But it also requires examining what the controls changed.

It requires models.

But it also requires identifying where those models stop.

It requires receivers.

But it also requires remembering that receiver limitations are not identical to limits on physical reality.

And it requires residuals.

But it requires earning them.

Conclusion

Science does not become conservative by demanding that anomalies survive rigorous attack.

It becomes capable of recognizing the anomaly that matters.

A raw discrepancy is easy to produce.

A qualified residual is difficult.

That difficulty is a feature of science rather than an obstacle to discovery.

Before an unexplained observation can support a new mechanism, the measurement must be qualified. Before a residual can be called robust, the system boundary must be declared. Before a boundary can be trusted, its relevant pathways must be examined. Before a control can eliminate an explanation, the relations created by that control must be considered. Before a pathway can be ignored, its possible contribution must be bounded below scientific relevance. Before a null can establish similarity, the experiment must possess sufficient sensitivity or an appropriate equivalence criterion. Before a theory can claim prediction, the distinguishing consequence must exist before the outcome is known. Before a new physical interpretation is justified, the residual must survive independent attempts to make it disappear.

The resulting hierarchy is deliberately demanding:

[ \boxed{ \text{Observation} \rightarrow \text{Validated Measurement} \rightarrow \text{Model Discrepancy} \rightarrow \text{Qualified Residual} \rightarrow \text{Mechanism-Specific Anomaly} \rightarrow \text{Theory Discrimination} \rightarrow \text{Possible New Physics} } ]

No step is forbidden.

No step is automatic.

This framework therefore establishes neither a presumption for conventional explanations nor a presumption for unconventional ones. It establishes a presumption for traceable inference.

Its deepest principle is equally applicable to ordinary laboratory science and to the most ambitious attempt at fundamental discovery:

The unexplained remainder becomes scientifically interesting only after we have worked hard enough to determine what, exactly, is doing the remaining.

A boundary is not merely the edge of an experiment.

A control is not merely the absence of a variable.

A detector is not reality itself.

A model is not its subject.

And a residual is not yet an explanation.

Before the residual can tell us something new about reality, science must first establish that the residual belongs to reality—and not to the architecture through which we asked reality the question.

References

1. Swygert, J. (2026). Pathways, Boundaries, and Phases: The Relational Expression of Reality.
   
2. Swygert, J. (2026). Before the Outcome: A Prospective Prediction Architecture for TSTOEAO.
   
3. Swygert, J. (2026). The Minimum Sufficient Relational Ledger.
   
4. Swygert, J. (2026). Inside Is Not Isolated: Faraday Boundaries, Channel-Selective Shielding, and the Relational Architecture of Experimental Control Through TSTOEAO.
   
5. Swygert, J. (2026). Latching Onto Space: Asymmetric Electrostatic Propulsion, Relational Momentum Accounting, and a TSTOEAO Substrate-Coupling Hypothesis.
   
6. Taylor, B. N., & Kuyatt, C. E. (1994). Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results. NIST Technical Note 1297. National Institute of Standards and Technology.
   
7. National Institute of Standards and Technology. NIST Policy on Metrological Traceability and Metrological Traceability: Frequently Asked Questions.
   
8. International Organization for Standardization & International Electrotechnical Commission. (2017). ISO/IEC 17025:2017 — General Requirements for the Competence of Testing and Calibration Laboratories.
   
9. National Academies of Sciences, Engineering, and Medicine. (2019). Reproducibility and Replicability in Science. Washington, DC: The National Academies Press. https://doi.org/10.17226/25303.
   
10. Munafò, M. R., Nosek, B. A., Bishop, D. V. M., Button, K. S., Chambers, C. D., Percie du Sert, N., Simonsohn, U., Wagenmakers, E. J., Ware, J. J., & Ioannidis, J. P. A. (2017). A manifesto for reproducible science. Nature Human Behaviour, 1, 0021. https://doi.org/10.1038/s41562-016-0021.
    
11. Nosek, B. A., Ebersole, C. R., DeHaven, A. C., & Mellor, D. T. (2018). The preregistration revolution. Proceedings of the National Academy of Sciences, 115(11), 2600–2606. https://doi.org/10.1073/pnas.1708274114.
    
12. Wasserstein, R. L., & Lazar, N. A. (2016). The ASA’s statement on p-values: Context, process, and purpose. The American Statistician, 70(2), 129–133. https://doi.org/10.1080/00031305.2016.1154108.
    
13. Lakens, D. (2017). Equivalence tests: A practical primer for t tests, correlations, and meta-analyses. Social Psychological and Personality Science, 8(4), 355–362. https://doi.org/10.1177/1948550617697177.
    
14. Box, G. E. P. (1976). Science and statistics. Journal of the American Statistical Association, 71(356), 791–799. https://doi.org/10.1080/01621459.1976.10480949.
    
15. Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on Automatic Control, 19(6), 716–723. https://doi.org/10.1109/TAC.1974.1100705.
    
16. Popper, K. R. (1959). The Logic of Scientific Discovery. London: Hutchinson.