Compute Without Extraction: The Civic Allocation Regime Between Cloud Dependency and Community Value
Description
Compute-dependence is becoming infrastructural for public administration, education, health, welfare, energy, mobility and local resilience. Yet debate about the compute transition remains largely concerned with siting, efficiency, grid capacity, carbon intensity and competitiveness. These questions matter, but they do not answer a prior institutional question: when a community or public system becomes dependent on external compute, what ensures that value, capability, authority, contestability and resilience flow back rather than only outward?
This working paper defines extraction as a directional asymmetry in value. An arrangement becomes extractive when it serves or materially shapes a public function, creates an outward flow of data, capability, economic value, decision authority or resilience, and does not provide a comparable return relevant to the dependency created. The definition is deliberately separated from the proposed remedy so that the central thesis remains falsifiable.
The paper proposes a civic allocation regime based on five answering returns: lawful authorisation and restraint over local data use, retained capability, value return, contestability and active fallback. These rights would be vested in the served community and exercised through existing forums rather than administered by a new permanent compute authority.
The principal enforcement mechanism is burden-shifting. Once a plausible reciprocity gap is raised, the provider, procuring authority or system operator would be required to demonstrate reciprocity. The paper examines the legal difficulty of defining a public function, the risk that community rights remain merely decorative, and the possibility that disclosure duties produce compliance flooding rather than accountability.
Applications to Robodebt and SyRI illustrate evidential and contestability failures. A fictional municipal housing case demonstrates how a service may become faster while losing local knowledge and resilience, while a commercial automation boundary case shows why public consequence alone does not establish a public function.
This is a conceptual policy-design working paper released for adversarial review. It does not constitute legal advice.
Abstract
Compute-dependence is becoming infrastructural for public administration, education, health, welfare, energy, mobility and local resilience. Yet debate about the compute transition remains focused largely on siting, efficiency, grid capacity, carbon intensity and competitiveness. These questions matter, but they do not answer a prior institutional question: when a community or public system becomes dependent on external compute, what ensures that value, capability, authority, contestability and resilience flow back rather than only outward?
This working paper defines extraction as a directional asymmetry in value. An arrangement becomes extractive when it serves or materially shapes a public function, creates an outward flow of data, operational capability, economic value, decision authority or resilience, and does not provide a comparable return relevant to the dependency created. The definition is deliberately separated from the proposed remedy so that the central thesis remains falsifiable.
The paper proposes a civic allocation regime organised around five answering returns: lawful authorisation and restraint over local data use, retained capability, value return, contestability and active fallback. Rather than creating a new permanent compute authority, these would be vested as community rights and exercised through existing courts, regulators, councils, ombudsman structures and public-interest forums.
The principal enforcement proposal is burden-shifting. Once a plausible reciprocity gap is raised, the provider, procuring authority or system operator would be required to demonstrate reciprocity. The paper examines the legal difficulty of defining a public function, the risk that community-held rights remain merely decorative, and the possibility that disclosure duties produce compliance flooding rather than substantive accountability.
Applications to Robodebt and SyRI illustrate failures of evidential fairness and contestability. A fictional municipal housing case shows how a service may become faster while losing local knowledge and resilience, while a commercial automation boundary case demonstrates why public consequence alone does not establish a public function.
The paper’s central claim is that a compute transition without a civic allocation regime does not end extraction; it merely relocates it.
Notes
This working paper is released for adversarial review as a conceptual policy-design contribution. It does not constitute legal advice, regulatory guidance, or a statement of existing law in any single jurisdiction.
The legal examples discussed are drawn from different jurisdictions and are used for comparative and analytical purposes only. The paper does not claim that the cited doctrines already combine into an existing legal regime.
Any unnamed municipal, commercial, employment, procurement, or service-delivery scenarios are fictional or stylised composites created solely to illustrate the proposed diagnostic. They do not describe or allege misconduct by any particular person, authority, provider, employer, company, or institution.
The author has sought to distinguish established facts, interpretive claims, normative proposals, contingent conclusions, and conjecture. Readers should verify legal and empirical claims against the cited primary sources before relying on them.
Notes
Post-publication note — administrative velocity and reversible provisional deployment
Added 10 July 2026
A further implementation difficulty concerns the mismatch between the speed of modern compute deployment and the slower decision cycles of courts, regulators, councils and ombudsman structures. A civic allocation regime cannot require prior administrative approval for every software update without producing paralysis. Its practical design would therefore need to govern dependency upstream through standing procurement conditions, retained fallback, concise reciprocity disclosures, material-change triggers and reversible provisional deployment.
Routine technical changes could proceed within an agreed reciprocity envelope. Changes affecting data use, decision authority, retained capability, contestability or continuity would cross that envelope and trigger heightened obligations. This reuses the paper’s five-flow structure as a change-control test: the relevant question is not whether software has changed, but whether the change alters the direction or balance of civic dependency.
Where a plausible reciprocity gap is raised, interim protection should operate through preservative defaults rather than a rapid discretionary judgment on the merits. A properly triggered challenge could automatically preserve human review, prevent the removal of existing fallback, retain the previous deployable version and keep affected decisions reopenable while the provider or procuring authority responds. These measures would freeze potentially irreversible changes without determining that extraction has occurred. The responding party could seek to lift them by demonstrating that the deployment remains within the reciprocity envelope.
The governing principle is not that public institutions must match the speed of compute transaction by transaction. It is that public-function compute must remain legible, reversible and contestable at the speed institutions can responsibly govern.
This velocity problem is not resolved in Working Paper v0.5 and should be treated as an additional operational dependency of the proposed regime. The concept of reversible provisional deployment requires fuller development, including the contractual form of the reciprocity envelope, the threshold for triggering preservative defaults, the duration of interim protection and safeguards against tactical or abusive challenges.
Files
Compute_Without_Extraction_v0-5.pdf
Files
(360.4 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:26923a8a852f363592ecfe8f28cd1f74
|
360.4 kB | Preview Download |
Additional details
Dates
- Created
-
2026-07-10Date of public release on Zenodo