WeaknessesandRisks
Critical analysis of limitations and potential risks of digital planning
This page lists what can break the open planning model and says what is done about each item. Some risks are closed by a design decision; some are only moved from hidden to visible; and some are not solved at all at the current level of engineering. The last are marked as unsolved, not as “directions for development”.
Capture of goal-setting
The largest risk here is not technical. Talk of “optimisation through technology” by itself justifies handing power to those who deploy the technology: the system designers stay at the top and citizens become users. The word “open” in the name is no protection — open data coexists comfortably with closed decisions.
The problem: technology does not create democracy; it amplifies whatever configuration of power it is placed into.
What we do: goal-setting is carried in a separate loop on a one-person-one-vote basis, and automation is deliberately kept out of it. Voting by weight of assets or tokens is excluded — it reproduces a shareholders’ meeting. The technical loop delivers the goals and has no standing to set them.
Oracles: living labour behind every “automatic confirmation”
A smart contract does not know what happened in the physical world. It knows what an oracle told it — a sensor, an enterprise system, an auditor, a data labeller. Every one of those sources can be faked, and the work of the people who keep them running is usually invisible and badly paid. An “objective indicator” is somebody’s shift.
The problem: trust does not disappear, it relocates — from the official to the data supplier.
What we do: treat oracles as labour rather than as a source of objectivity. Cross-checking of independent sources, liability backed by stake and reputation, disclosure of who exactly confirmed a fact, and payment for that work as work. This does not remove the risk; it makes it visible and attributable.
Throughput and the cost of confirmation
A national economy is billions of operations a day. Public distributed ledgers are not built for those volumes, and the permissioned networks that can carry them lose the very property the blockchain was chosen for — independence from the operator. The cost of confirmation, meanwhile, is not zero, and somebody pays it.
The problem: a “digital Gosplan” as end-to-end on-chain accounting is not feasible at the current level of engineering.
What we do: we do not design an end-to-end ledger. Only what must be indisputable and changes rarely goes on chain: the registry of goals and limits, the acceptance of a stage, the movement of quotas. Bulk operations stay in ordinary systems and reach the ledger as a summary. That removes the load — and returns the question of trusting the summary to the point above about oracles.
A measure that becomes a target stops being a measure
Goodhart’s law is not repealed by digitisation. “Gross output at any cost” does not go away — it becomes “KPIs at any cost”. Worse: the more precisely payment for a met indicator is automated, the faster and cheaper it becomes to fix that indicator, because what now has to be fixed is not a report but a single input.
The problem: the cleaner the automation of payment against a metric, the stronger the incentive to falsify the metric.
What we do: limit the reach of quantitative mechanisms in advance — they do not enter care, education, medicine, culture or science. Inside the permitted area: several indicators rather than one, a deferred share of payment for quality, rotation of metrics, human audit of a sample. None of these guarantees anything; they raise the cost of gaming.
Legal and organisational unreadiness
Working engineering does not give a working institution. Four questions have to be answered: who operates the registry and on what authority; what happens to obligations in the event of a failure or a chain split; how on-chain acceptance relates to existing procurement and budget accounting; who is liable if an oracle confirmed something untrue. Not one of them is settled by writing a contract.
The problem: without a legal frame a smart contract is not an obligation but a script.
What we do: treat this as a condition of entry rather than a detail of rollout. A pilot starts with answers to the four questions above, and in a sector where existing regulation already permits the arrangement — otherwise what you get is a demonstration, not an operation.
Simplified history and politicised terms
The comparison “rigid Gosplan versus flexible digital model” is convenient and wrong. The planning system was not uniform: there were attempts at cost accounting, elements of market incentives, the 1965 reform, periods of decentralisation. The opposite simplification is equally wrong — “the sums were right, the computers were missing”. Both versions put an argument where history should be. Separately: any discussion of planning is instantly labelled, and the argument about the label crowds out the argument about the substance.
The problem: distorted history yields wrong conclusions, and a dispute over the name displaces the question of whether the mechanism works.
What we do: the historical material is examined separately, in the section on the Soviet Gosplan, and is not used as an argument from authority. The model has to stand on its own — without leaning on “the USSR nearly managed it”. We dropped the word “technosocialism” for the same reason, but renaming changes no mechanism and answers no criticism.
Conclusion
Of the six risks listed, two are closed by a design decision — the goal-setting loop and the limit on where metrics apply. Two more, oracles and the legal frame, are moved from hidden to visible but not removed: they stay the responsibility of people and institutions. Throughput and the cost of confirmation are unsolved at the current level of engineering, and we do not believe that is merely a matter of time and money.
Digital transformation of public administration must rest on a combination of technology, common sense and institutional readiness — not on a change of slogans or on simple analogies with history.