Economics – comparing architectures rather than assuming savings
Зміст статті / On this page (14)
- The investment question
- Three alternatives
- Five-year cost model
- Conservative, Base and Optimistic
- Metrics the pilot must collect
- Public benefit is not trading revenue
- Decision rule
- Why the cost case matters
- One use case under three architectures
- Five-year total cost of ownership
- Trading revenue is not the societal benefit of a data platform
- Conservative, Base and Optimistic – without invented numbers
- Pilot evidence and the decision rule
- Related chapters
The investment question
Does a federation of technical access nodes materially improve interoperability, availability and market entry relative to a centrally operated, documented, redundant API? If coordination and security costs exceed demonstrated gains, a simpler architecture may be the responsible choice.
Three alternatives
A – One statutory administrator plus open central API
Standard machine-readable contracts, documented access/permission procedures, synthetic tests, auditable neutrality and geographic redundancy. Potential advantages include fewer operational boundaries and simpler accountability. Risks include concentration of workload and correlated failures.
B – One statutory administrator plus interoperable technical nodes
One authoritative settlement framework remains. Independent technical access providers are joined through a directory and an agreed trust model, to the extent legally authorised. Potential advantages: local integration, reusable adapters and competing service implementations. Added expenses: a directory, permission coordination, independent PKI, certification, cross-operator incident handling and migration.
C – Several statutory metering administrators
A fundamentally different regulatory model, not just more servers. Official authority, disputes and settlement validity would require substantial legal reform. It is not EDDIE UA's current recommendation.
Five-year cost model
Capital costs include API development, geo-redundancy, directories, permission systems, node installation, integration adapters, security assessment, migration and legal work. Operating costs include compute, storage, audit logs, customer support, monitoring, certificates, schema lifecycle, incident handling and continuing assurance.
For each scenario, calculate discounted total cost of ownership: upfront CAPEX plus the present value of yearly OPEX and measured expected outage impacts. Node count and eligible-party count drive different parts of the cost structure. The discount rate, assumptions and unit prices must be sourced.
Conservative, Base and Optimistic
Conservative assumes slower adoption, expensive certification and modest reuse of common connectors. Base uses measured integration and operational costs from a representative pilot. Optimistic assumes high reuse, lower integration expense and a verified reliability benefit. No numeric savings percentage is published because these inputs have not been collected.
Metrics the pilot must collect
Integration engineering hours per eligible party, average and p95 onboarding duration, volume and reason categories for denied requests, cost per thousand API operations, permission-revocation latency, uptime/RTO/RPO, independent audit costs, certificate lifecycle expense, correction integrity and the cost of maintaining backward compatibility.
The same workload, metering-point mix and legal scope must be used for both A and B.
Public benefit is not trading revenue
Additional gross trading income for a specific aggregator does not automatically constitute net social welfare. A valid impact assessment needs a counterfactual: would the service or cost reduction have existed under a central open API? Consider privacy incidents, opportunity costs, distributional effects and whether any public support creates preferential advantage.
Decision rule
Choose B only if it is lawful, independently securable and delivers sufficient cost, access or innovation benefit compared with A. If the existing infrastructure with open standards performs better at lower cost, adopt A. Either result should be publishable without favouring Wesfin.
Why the cost case matters
The long-term cost of an energy-data platform is not the price of a prototype. It includes continuous operations, legal verification, security, backup, audit logs, support, interface changes and incident response. Federation adds independent operators, trust infrastructure, certificates and authority handover. These expenses need to be paid year after year.
We therefore do not claim that decentralisation guarantees a particular percentage saving. Such claims require measured integration costs, workloads, operator counts and a credible comparator.
One use case under three architectures
Imagine ten independent energy-service companies requiring lawful access to industrial interval readings. Under A, they integrate with one documented central API with redundancy and equal rules. Under B, they work through interoperable technical nodes joined by a trust directory, as far as legally permitted. Under C, a more fundamental legal reorganisation of statutory metering authority would be required.
A needs central API, redundancy and onboarding investment. B adds directory, cross-node security, PKI, operator audits and complex recovery but may reuse independent adapters efficiently. C adds legal and institutional complexity. None of the alternatives should be assigned zero cost simply because we prefer its diagram.
Five-year total cost of ownership
Start with up-front design, API, hardware, integration, security assessment, legal work, migration and testing. Add annual support, storage, compute, certificates, audits, personnel, incident handling and schema maintenance. Expected costs of service interruption should be included without double counting.
Discount future spending using an explicitly sourced rate. Document assumptions and sensitivity. Choosing optimistic parameters merely to favour federation is not an impact assessment.
Trading revenue is not the societal benefit of a data platform
Better data access may help a flexibility aggregator identify opportunities. But gross market revenue is not net customer value. Electricity costs, storage degradation, imbalance exposure, platform fees, taxes and asset limitations matter. Moreover, the same flexibility service might have been enabled under a simpler open interface.
The relevant comparison concerns the incremental cost and benefit of information access: onboarding time, engineering effort, availability, data quality, ability to serve smaller customers and user choice, with an appropriate counterfactual.
Conservative, Base and Optimistic – without invented numbers
Conservative assumes weak adoption and expensive audits. Base uses measured costs and realistic onboarding time from a lawful pilot. Optimistic assumes high adapter reuse, lower integration expense and verified resilience benefits. Test sensitivity to node count, security assurance, request volume, operator fees and failure frequency.
Our current model does not contain sufficiently verified inputs to publish a savings estimate. The absence of a number is a methodological safeguard, not a missing marketing claim.
Pilot evidence and the decision rule
Record engineering hours per integration, median and p95 onboarding, request and rejection categories, cost per 1,000 requests, certificate renewal, revocation delay, correction integrity, incident frequency and recovery time. Compare A and B under equivalent legal scope and workload. If A is lawful, safer and cheaper while meeting the same requirements, choosing A would be the correct evidence-based outcome.

