Manifesto for Honest Computation

Phase VI (Final Synthesis)"

abstract: >

Five papers have traced an arc from diagnosis to prescription. The qubit is a

scaffold, not an invariant. Alternative paradigms exist. Physical limits are real.

A diversified portfolio is rational. Institutions must be reformed. This final

paper --- the Manifesto --- synthesizes these findings into principles for honest

computation and a call to action for investors, researchers, policymakers, and

the public. We propose five principles: (1) the substrate IS the algorithm; (2)

correlation over particle; (3) joules per solution as the universal metric; (4)

falsifiability as a condition of funding; and (5) institutional independence of

verification. We propose a specific resource reallocation --- from the ~60% currently

consumed by gate-model quantum computing toward thermodynamic, optical, neuromorphic,

and analog quantum simulation --- and a Falsification Pledge by which institutions

commit to pre-registering computational claims with time-bound success criteria.

Computation is too important to be left to narrative production. The future of

computing is honest, or it is nothing.

keywords:

• manifesto
• honest computation
• post-classical computing
• quantum computing reform
• computational principles
• research ethics

Preamble

Computation is the most important technology of the twenty-first century.

It determines what we can know, what we can build, what we can cure, and

what we can defend against. The future of computation is, in a very real

sense, the future of human capability.

For the past two decades, a significant fraction of the world's

computational R&D expenditure --- roughly $35 billion --- has been directed

toward one specific computational paradigm: the fault-tolerant universal

gate-model quantum computer. This investment was made on the basis of

extraordinary claims: exponential speedups for problems of commercial

importance, revolutionary new industries, a transformation of the

computational landscape comparable to the invention of the transistor.

Those claims have not been realized. Not one commercially viable quantum

computer exists. Not one problem of genuine economic value has been solved

faster, cheaper, or more accurately by a quantum computer than by a

classical alternative. The roadmaps have been revised outward, repeatedly,

from "five years" to "ten years" to "the 2030s" --- and those timelines

assume flawless execution of error-correction protocols that currently

consume thousands of physical qubits per logical qubit.

This is not a temporary setback. It is not a matter of insufficient

funding, insufficient talent, or insufficient time. The failure is

structural. It is embedded in the institutional incentive systems that

fund computational research, in the ontological confusion that treats

qubits as particles rather than relational properties, in the neglect of

fundamental thermodynamic limits, and in the absence of independent

verification with enforcement mechanisms.

Five papers have mapped this terrain. This final paper --- the Manifesto ---

synthesizes their findings into actionable principles and a call to

action. We do not merely diagnose. We prescribe. We do not merely

critique. We construct. The future of honest computation depends on it.

1. What We Have Learned

1.1 The Diagnosis (Phase I)

The qubit-gate-circuit model is an epistemic failure. It projects particle

ontology onto a relational, field-theoretic reality. The qubit is not a

thing; it is a relation. The gate is not an operation on an object; it is a

correlation between measurement contexts. Decoherence is not an enemy to be

suppressed; it is the physical process by which quantum systems become

classical --- and it can be harnessed as a computational resource.

The $35 billion quantum computing industry has optimized for narrative

production, not computational output, because the institutional incentive

structure rewards ambitious claims with prestigious publications, media

coverage, and capital inflows, while the correction --- independent

verification, qualification, and refutation --- operates on a timescale

too slow to affect capital allocation.

1.2 The Alternatives (Phase II)

Measurement-based quantum computing (MBQC), continuous-variable quantum

computing (CV-QC), topological quantum computing (TQC), field-theoretic

computation, analog quantum simulation, thermodynamic computing,

neuromorphic computing, and optical computing --- these are not minor

variations on the qubit model. They represent genuinely different

ontological commitments, with different scaling properties, different

manufacturability constraints, and different problem-class affinities.

The photonic platform (PsiQuantum, Xanadu) offers room-temperature

operation and leverages existing semiconductor manufacturing infrastructure,

making it the strongest commercial case in quantum computing. Neutral atom

arrays (QuEra, Atom Computing) offer the most promising path for near-term

scientific impact through analog quantum simulation. Ising machines and

thermodynamic solvers are already commercially deployed for optimization.

Neuromorphic processors already achieve ~10³× energy advantage over GPUs

for inference workloads.

1.3 The Limits (Phase III)

The Landauer bound does not privilege quantum over classical computation.

The Margolus-Levitin theorem imposes a fundamental quantum speed limit within

a factor of ~30-200 of current practice. The error-correction overhead ---

10² to 10³ physical operations per logical operation --- multiplies the

thermodynamic cost of quantum computation beyond what any polynomial

algorithmic speedup can overcome.

Only exponential speedups --- Shor's algorithm for factoring, and possibly

quantum simulation at sufficient scale --- can overcome the energy penalty.

For everything else, the joules-per-solution comparison favors specialized

classical hardware: Ising machines for optimization, optical processors for

linear algebra, neuromorphic processors for inference, reversible classical

processors for general-purpose computation.

1.4 The Portfolio (Phase IV)

The rational investment response to these findings is a diversified

portfolio across physical substrates, allocated in proportion to the

evidence for each substrate-problem pair. Approximately 45% thermodynamic

and analog (Ising machines, p-bit networks, analog quantum simulators),

25% optical (integrated photonic matrix processors), 20% neuromorphic,

and only 5-10% fault-tolerant quantum --- a near-inversion of the current

allocation where gate-model QC absorbs roughly 60% of advanced computing

investment.

1.5 The Institutions (Phase V)

The current institutional structure --- venture capital with 10-year fund

horizons, government agencies without post-award evaluation mechanisms,

academic publishing with prestige cascades, and a regulatory vacuum --- is

configured to produce narrative assets, not computational ones. The $35

billion outcome is the equilibrium of this game.

Five reforms are necessary: an Advanced Computing Projects Agency (ACPA)

modeled on DARPA; mandatory independent verification for computational

claims above a materiality threshold; structural separation of quantum

hardware and software; joules-per-solution procurement for all government

computational contracts; and a public registry of computational claims

with time-bound falsification criteria.

2. The Five Principles of Honest Computation

From this synthesis, we distill five principles. They are not merely

guidelines. They are conditions that any computational research program

must satisfy to be considered honest.

Principle 1: The Substrate IS the Algorithm

Computation is a physical process. The physical substrate --- its Hamiltonian,

its degrees of freedom, its natural dynamics, its coupling to the

environment --- determines what can be computed and at what cost. The choice

of computational scaffold (qubit vs. qumode, gate vs. measurement, digital

vs. analog) is never ontologically neutral. It encodes assumptions about

what computation IS, and those assumptions have consequences for what can

be computed efficiently.

The honest approach: match the problem class to the physical substrate

whose natural dynamics solve it most efficiently. Do not force all problems

into a single scaffold. Do not optimize a scaffold that is misaligned with

the physics. Let the physics do the computing.

Principle 2: Correlation Over Particle

The fundamental computational resource in quantum systems is not the state

of individual qubits but the correlation structure --- entanglement, phase

coherence, collective modes --- that connects them. A quantum computer is a

machine for steering correlations, not for manipulating particles.

The honest approach: design computational architectures where correlation

is the primitive --- not the qubit, not the gate, not the circuit depth. In

MBQC, the cluster state IS the computation. In CV-QC, Gaussian entanglement

IS the resource. In TQC, braiding correlations IS the algorithm. In

thermodynamic computing, the free energy landscape IS the solver.

Principle 3: Joules Per Solution, Not Quantum Volume

The universal metric for computational advantage is total energy consumed

to deliver a verified, correct answer to a specified problem at specified

scale. "Joules per solution" --- measured at the wall plug, including all

overhead --- is the only metric that is invariant under scaffold variation,

platform choice, and algorithmic approach.

"Quantum volume," "quantum utility," "quantum supremacy" --- these are

scaffold-specific metrics that privilege the qubit-gate-circuit model.

They do not measure what users care about: does this machine solve my

problem faster, cheaper, or more accurately than alternatives?

The honest approach: publish joules-per-solution numbers for commercially

relevant problems, compared against the best available classical

implementation. If the joules-per-solution number is not available, the

claim is not yet an investment proposition.

Principle 4: Falsifiability as a Condition of Funding

No computational research program should receive public funding without

pre-registered, time-bound, falsifiable milestones. "We will demonstrate

X joules-per-solution advantage on problem Y of commercially relevant

scale Z by date W." If the milestone is not met by the stated date, funding

is terminated. No extensions. No rebranding. No milestone creep.

Falsifiability is not an optional feature of honest science. It is the

defining feature. A research program that cannot state what would count as

failure is not a research program; it is a faith-based initiative. Faith-

based initiatives have their place --- in religion, in art, in speculative

philosophy --- but they should not be funded with public money or marketed

to investors as technology investments.

Principle 5: Institutional Independence of Verification

Evaluation of computational claims must be performed by institutions that

are structurally independent of the claimants. The academic community

cannot verify its own claims, because the prestige cascade ensures that

ambitious claims are rewarded before they are verified. Venture capitalists

cannot verify the claims of their portfolio companies, because the

information asymmetry is unbridgeable and the incentive is to mark up,

not to verify. Government agencies cannot verify claims from programs they

have funded, because program managers have career incentives to report

success.

Independent verification requires independent institutions: an ACPA with

DARPA's structure, a public registry with enforcement mechanisms, an

independent verification corps funded by a levy on computational research

expenditure. The principle is separation of powers: those who make claims

must not be the ones who verify them. Those who fund claims must not be

the ones who evaluate them. Those who benefit from claims must not be the

ones who adjudicate them.

3. The Honest Portfolio: A Concrete Reallocation

The principles above imply a specific reallocation of computational R&D

capital. We present it as a concrete recommendation, not as a theoretical

exercise.

3.1 Near-Term (1-3 Years): Deploy What Works

What to STOP funding at near-term scale: Universal fault-tolerant

gate-model QC on superconducting or trapped-ion platforms. The thermodynamic

arithmetic does not work for commercially relevant problem sizes within

VC-relevant timeframes. These platforms should be maintained as fundamental

research programs at ~5% of current funding levels, not as commercial

ventures at ~60%.

3.2 Medium-Term (3-7 Years): Scale What Shows Promise

3.3 Long-Term (7-15+ Years): Invest in the Physics

4. A Call to Action

The principles and portfolio above are not self-executing. They require

action by specific actors.

4.1 To Investors (Venture Capital, Institutional, Retail)

• Demand joules-per-solution numbers. If a quantum computing company

cannot state --- in writing --- the problem class, scale, joules-per-solution,

classical competitor, and falsifiable milestone with timeframe, do not

invest. Period.

• Diversify across substrates. The optimal portfolio is not a bet on

one quantum computing company. It is a diversified allocation across the

Problem-Substrate Mapping. If you must invest in quantum computing,

allocate no more than 10% of your advanced computing exposure to

gate-model QC.

• Shorten your evaluation cycle. The Narrative Production Function

operates on a timescale of months. Independent verification operates on

years. Close the gap: require independent verification BEFORE capital

deployment, not after. Your limited partners will thank you.

4.2 To Researchers (Academic and Industrial)

• Pre-register your claims. Before announcing a result, register the

claim, the falsification criteria, and the independent verification

protocol in the Public Registry of Computational Claims. If the registry

does not yet exist, publish these on your institutional repository.

• Celebrate negative results. A well-designed experiment that falsifies

a hypothesis is as valuable as one that confirms it. The institutional

incentive structure currently penalizes negative results by making them

harder to publish and less prestigious. Change this: submit negative

results to journals, cite them, and reward colleagues who publish them.

• Reject scaffold-specific benchmarks. Do not report "quantum volume"

or "quantum supremacy" without also reporting joules-per-solution on a

commercially relevant problem. If the commercially relevant problem is

not yet solvable, state when it will be --- with a specific date --- and

hold yourself to that date.

4.3 To Policymakers and Funding Agencies

• Establish an Advanced Computing Projects Agency. Model it on DARPA.

Fund it at $500 million per year. Staff it with active researchers on

3-5 year terms. Give it the authority to terminate programs that miss

milestones. Require it to publish all results --- positive and negative ---

within six months.

• Mandate independent verification. Any claim of computational

advantage above a $10 million funding threshold must be independently

verified within 12 months. Make independent verification a condition of

continued funding. Fund the verification infrastructure through a 0.5%

levy on computational research expenditure.

• Reform procurement. All government computational procurement above

$1 million must evaluate bids on a joules-per-solution basis. No more

buying "access" to quantum computers that solve no commercially relevant

problems.

• Establish a Public Registry of Computational Claims. Fund it. Staff

it. Give it the authority to flag unverified claims. Make compliance a

condition of government funding and a factor in government procurement.

4.4 To the Public

• Demand evidence. When a company announces "quantum supremacy" or

"quantum advantage," ask: on what problem? Compared to what classical

alternative? At what joules-per-solution? What is the falsification

criterion? What is the deadline?

• Follow the joules, not the narrative. The quantum computing industry

has produced an extraordinary volume of press releases, TED talks, and

Nature covers. It has produced zero commercially viable computational

output. These two facts are not unrelated. The narrative production

function consumes resources that could be directed toward actual

computation.

• Support institutional reform. Write to your representatives. Demand

that public funds for computational research be allocated on the basis of

evidence, not narrative. The $35 billion already spent came, in

significant part, from taxes. The next $35 billion should be spent

differently.

5. The Falsification Pledge

We, the authors of this Manifesto and the five papers that precede it,

commit to the following:

1. **Every claim we make about computational advantage will be accompanied

by a specific joules-per-solution number compared against the best

available classical implementation.**

1. **Every prediction we make about future computational capability will

be accompanied by a specific deadline and a specific falsification

criterion. If the deadline passes without the criterion being met, we

will publicly acknowledge the failure.**

1. **We will not report "quantum volume," "quantum utility," or any other

scaffold-specific metric without also reporting the joules-per-solution

on the most commercially relevant problem accessible at the reported

scale.**

1. **We will submit all our computational claims to independent

verification --- and we will publish the verification reports, whether

they confirm, qualify, or refute our claims.**

1. **We will celebrate negative results. A falsified hypothesis is a

contribution to knowledge. A perpetually unfalsifiable research program

is a drain on knowledge.**

We invite --- we challenge --- every institution, company, research group,

and individual working in post-classical computing to adopt this pledge.

Make it public. Link to it from your website. Include it in your grant

proposals and your investor pitch decks.

If you cannot adopt the Falsification Pledge, ask yourself why. What are

you protecting? Who benefits from the ambiguity? What would happen to your

funding, your valuation, your prestige, if your claims were subjected to

rigorous, independent, time-bound verification?

The answer to that question is the answer to the question this entire

series has been asking: is quantum computing a scientific enterprise or a

financial one? Is it seeking truth, or seeking capital?

6. Conclusion: Honest, or Nothing

The first five papers in this series have made the case. The qubit is a

scaffold, not an invariant. Alternative paradigms exist. Physical limits

are real and binding. A diversified portfolio is the rational investment

response. Institutional reform is necessary and possible.

This Manifesto adds the final element: a set of principles and a call to

action. Computation can be honest, or it can be nothing. A computational

paradigm that substitutes narrative for evidence, prestige for verification,

and fundraising for falsification is not a scientific enterprise. It is a

machine for converting capital into press releases.

The laws of physics do not negotiate. The Landauer bound applies equally to

classical and quantum computation. The Margolus-Levitin theorem limits the

speed of any quantum evolution. The error-correction overhead multiplies

the thermodynamic cost of fault-tolerant quantum computation beyond what

any polynomial algorithmic speedup can overcome. These are facts. They are

not subject to revision by press release, by TED talk, or by Series D

funding round.

The future of honest computation is not a single device. It is an ecosystem:

thermodynamic optimizers for combinatorial problems, optical processors for

linear algebra, neuromorphic processors for probabilistic inference, analog

quantum simulators for quantum many-body physics, reversible classical

processors for general-purpose computation, and --- if and when the physics

permits --- fault-tolerant quantum processors for cryptography and other

problems requiring exponential speedups at scale.

This ecosystem will not emerge from the incumbent quantum computing

industry, whose institutional incentives reward the opposite of honesty.

It will emerge from new institutions --- an ACPA, an independent verification

corps, a public registry of claims --- and from the courage of individual

researchers, investors, and policymakers who insist on evidence over

narrative.

The choice is ours. Honest computation, or nothing.

This Manifesto is the sixth and final paper in the "Qubit Delusion" series.

All papers are available under CC-BY-4.0 at Zenodo via the concept DOI

10.5281/zenodo.21254028. The Falsification Pledge can be adopted by any

institution, company, or individual. Attribution requested, not required.

Series References

• Phase I: "The Qubit Delusion" --- DOI: 10.5281/zenodo.21254143
• Phase II: "Beyond the Qubit" --- DOI: 10.5281/zenodo.21254901
• Phase III: "The Physics of Computation" --- DOI: 10.5281/zenodo.21255013
• Phase IV: "The Problem-Substrate Mapping" --- DOI: 10.5281/zenodo.21255346
• Phase V: "Institutional Reform" --- DOI: 10.5281/zenodo.21299211
• Phase VI: "Manifesto for Honest Computation" --- this paper