Part I — Situation overview

The last operating generating unit of the Paks Nuclear Power Plant shut down on Sunday 2 August 2026: the Danube’s water level at Paks sank to minus 133 centimetres, while the limit value for the safe cooling of the plant was set at minus 134 centimetres. The shutdown began in the early hours — according to 444.hu’s report the last but one turbine was switched off at half past one in the morning, and the plant was already producing only 240 megawatts instead of the nominal 2,000. On Sunday evening the prime minister stated on RTL’s programme Házon kívül that the shutdown may last even longer than a month, and that “we have to prepare for this mentally too”. It was this sentence that changed the nature of the situation: a state of emergency lasting a few days became an operating mode lasting several weeks.

On the same day the government postponed the harshest instrument by invoking the savings figures. After Sunday’s meeting of the defence working group the prime minister stated that for the time being it is not necessary to order compulsory consumption reduction at the large industrial consumers, because more than three hundred companies had undertaken voluntary curtailment totalling about 400 megawatts, and the joining of a further 270 firms is expected. On Saturday national consumption fell 500 megawatts short of expectations; on Sunday evening Kapitány István, named as the head of the economic and energy portfolio — with the title as given in the press — was already reporting a shortfall of 700 megawatts, which in his formulation is “more than the output of a full Paks unit”. The order of the curtailment chain meanwhile remained fixed: first state institutions, then large consumers, and “only as an absolute last resort” households.

On Monday morning, however, precisely this figure became disputed. Hortay Olivér, energy analyst — head of the energy and climate policy division of Századvég — claimed on his social media page that the ministry had presented the load data in a more favourable light, and called it “absurd” that the government’s disclosure is not accurate even “in the most banal, most easily verifiable data”. According to MIAK’s reading this is not a dispute over detail and not a party-political skirmish, but a structural weak point of crisis management: the savings figure comes from the same actor who is politically responsible for the success of the crisis management, and who on the basis of that same figure decides to postpone the strongest instrument. This arrangement is open to attack even if the figure is accurate — and that is exactly what is wrong with it. In a crisis the measured datum is the only legitimacy reserve: if it is disputable, then the government loses the argument with which it could ask further sacrifice of the population. A structural gap appears in the same way on the other side too: system-level decisions have identifiable losers — six thousand Paks residents were left without hot water, and the solar sector asked for regulatory answers in an open letter — who today have no compensation rule known in advance.

Part II — Literature foundation

Before turning to MIAK’s proposals it is worth fixing the conceptual frame. The volume Spin Dictators by Sergei Guriev and Daniel Treisman (an economist–political scientist pair of authors, researchers of the informational toolkit of modern authoritarian governance) shows that the chief instrument of the exercise of power in the 21st century is not open coercion but the prior arrangement of information: magnifying the good figures, postponing the bad ones, sliding measurement and communication together. The lesson — and MIAK uses this expressly not as an accusation but as an institutional yardstick for every government — is that the credibility of a datum depends not on the intention of the discloser but on whether measurement and political interpretation are separated from each other. The work Essence of Decision by Graham Allison and Philip Zelikow (the classic analysts of the decision-making of the Cuban missile crisis) adds the implementation side to this: in a crisis the government is not a single, rationally deliberating actor but a set of organisations that operate according to their own established procedures, and the intention of the political leadership often breaks on those routines. The detailed treatment of the literature — author by author, with quotations — can be found in the 6.4 Literature in detail section.

📖 Source: Sergei Guriev – Daniel Treisman: Spin Dictators; Graham Allison – Philip Zelikow: Essence of Decision

Part III — MIAK’s concrete proposal

MIAK proposes three measurable measures. None of them requires a new law, and all three can be implemented during the shutdown now under way.

3.1 Handing the disclosure of the load and savings data to the system operator (within 7 days)

The daily savings figure should not be announced by the ministry responsible for the policy area at a press conference or on a social media page; instead the operator of the electricity system should publish it as a raw, machine-readable time series: hourly actual system load, hourly forecast load, the difference between the two, and the aggregate and fulfilled value of the voluntary undertakings of large consumers. The format of the publication should not change from day to day, and it should also contain the methodological description — relative to what we calculate the saving. This proposal follows directly from the G19 radical transparency programme point and the D9 data-driven decision support framework. According to Robert Klitgaard’s corruption-theory logic the risk is greatest where discretionary power operates without accountability (📖 Source: Robert Klitgaard: Controlling Corruption); the proposal strengthens precisely this point, because it separates the production of the figure from the political use of the figure. The practical benefit is immediate: the Monday morning data dispute would cease of itself, because everyone — analyst, opposition MP, journalist, university researcher — would be working from the same raw data series.

3.2 A compensation band for those bearing a loss in the interest of the system (within 30 days)

Today three groups bear an identifiable loss for the stability of the system, and none of them has a compensation rule promulgated in advance: the Paks district heating customers — according to 444.hu’s data 2,771 housing estate flats and 181 public institutions, in total about six thousand people — who were left without hot water; the companies undertaking voluntary curtailment, which according to the prime minister made their undertaking “at the expense of their own profit”; and the solar sector, whose players addressed the minister in an open letter. MIAK proposes that within 30 days the government should promulgate a uniform compensation band, identical for everyone: for those affected in the household sector in the form of in-kind provision and tariff discounts, and on the corporate side with fixed remuneration set in advance for availability. This is the crisis variant of the K5 just transition principle: not charity, but recognition that an individual loss borne for a collective goal can be quantified, and is borne by the collective.

3.3 Rewriting the savings regime for a multi-week operating mode (within 14 days)

The one-month shutdown estimate means that the present package of measures, announced day by day, is structurally unsuitable. What is an effort for a few days grows tiring over weeks: a curtailment regime tied to a daily press conference cannot be planned for by the company, by the municipality or by the household. MIAK asks that within 14 days the government publish the weekly schedule of the savings regime — in which parts of the day, for which group of consumers, on exceeding which measured threshold which level takes effect — and assign to it a weekly review point. The K7 energy market shock resilience and the G25 energy price shock preparedness plan programme points both follow this logic: crisis management works if the actors know in advance what is expected of them.

The three proposals can be strung on a single principle. In a crisis the government exercises extraordinary power — it can postpone, restrict, order — and only one counterweight remains to this: publicly verifiable measurement. If the measurement is in the hands of the decision-maker, the counterweight ceases to exist. In Allison and Zelikow’s frame point 3.3 translates the same thought into the language of implementation: organisational routines cannot be renegotiated daily, and therefore the weekly schedule is not a bureaucratic formality but the framework within which the implementing organisations are able to operate at all (see 6.4.2).

Part IV — Expected effects and risks

Dimension Expected effect Risk
Economy Companies get a plannable framework for curtailment; the compensation band reduces disputes and legal risk Compensation is a budgetary item in the middle of the crisis, when the revenue side is deteriorating too
Society Raw data disclosure restores the credibility of the appeal to save; those affected in Paks receive identified help In the short term the raw data may also show a worse picture than the figure communicated — this has to be accepted
Public administration Separating measurement from political interpretation gives a lasting institutional pattern for the next crisis Publication devolved onto the system operator requires capacity and legal authorisation

The main trade-off is obvious: publishing the raw data is a political risk in the short term, but in the longer term it is the only instrument with which a request for further sacrifice remains credible. The proposal tips to the risk side if the publication does take place, but with a delay or in aggregated form — in that case apparent openness is worse than open refusal, because it generates a further dispute. The compensation band works if the circle of entitlement is measurable and closed (a district heating consumption point, contracted availability), and goes wrong if it turns into a series of individual equity decisions.

Part V — Measurability and summary

5.1 What is worth tracking? (proposed KPIs)

Four performance indicators (KPIs) are worth tracking over the coming month:

  1. Data disclosure delay: the time elapsed until the actual hourly system load becomes public — proposed target: under 24 hours, in machine-readable form.
  2. Data dispute index: how many independent expert or press claims dispute the official savings figure per month — proposed target: zero, because it is verifiable from the raw data series.
  3. Compensation coverage: what proportion of the identifiably affected consumption points (Paks district heating flats, curtailed large consumers) is covered by a written rule promulgated in advance — proposed target: 100% within thirty days.
  4. The fulfilment rate of the savings undertakings: the saving actually realised out of the announced 400 megawatts of voluntary undertakings, in a weekly breakdown.

5.2 Summary

MIAK’s key message in a single sentence: publishing crisis data is not a communications task but an institutional obligation. Concretely it asks the government to hand the disclosure of the load and savings data to the system operator in the form of a raw time series within seven days, to publish the weekly schedule of the savings regime within fourteen days, and to promulgate the uniform rule of compensation within thirty days. None of the requests is about whether the crisis management has been good or bad so far — all three are about its being verifiable.

Two MIAK foundational values are in play in this matter. Data-drivenness, because the present decision chain — measurement, disclosure, decision on the basis of the measurement — runs together entirely at a single actor, and so cannot be verified even alongside the best of intentions; data-drivenness is not invoking data but the production of the data being separated from whoever has an interest in a good result. And accountability, because voluntary saving today produces identified losers — housing estates left without hot water, firms undertaking commitments at the expense of their profit — whose loss is today booked nowhere. What is not booked is something for which nobody bears responsibility afterwards. MIAK asked the same yardstick of the previous government, and it asks the same of the present one.

Part VI — Justifications and further sources

6.1 The press framing by spectrum

The liberal-left band broke the story down into the figures and the situation of those affected. Telex put the savings figure from the prime ministerial announcement into its headline, while 444.hu ran two separate threads: the news of the postponement of the industrial load shedding from the defence working group’s meeting, and the human consequences of the Paks district heating outage — the shower containers, the electric water heaters, the opening of public institutions and the spontaneous offers of help from residents in detached-house neighbourhoods. HVG quoted the minister’s statement in its own register (“this country can bear anything if it does it together”), that is, it brought the mobilising frame, without critical comment.

The economic and public affairs bands concentrated on the operational balance sheet. Portfolio published a separate analysis on the development of the system load, and carried in the same place the solar association’s seven-point open letter requesting regulatory changes, as well as the Slovak offer of help — that is, it drew out the supply side of the crisis management in the greatest detail. 24.hu carried the published list of the companies undertaking the largest savings, and the safety statement of the Paks communications director, while ATV carried the decision of the defence working group and the expectation of a further 270 firms joining.

The conservative band was the only one to call into question the accuracy of the figure disclosed: Mandiner gave space to Hortay Olivér’s claim about the distortion of the load data, and dealt in a separate article with MVM’s safety statement; Magyar Nemzet, under the title Magyar Péter was corrected three times in two days, listed the inaccuracies of the prime minister’s communication. This framing is politically motivated, but the question raised — whether the crisis data are verifiable — is valid in policy terms, and according to MIAK precisely for that reason the correct answer to it is not rebuttal but the publication of the raw data. Comparing the bands reveals one common gap: not a single paper asked who reimburses the individual loss borne in the interest of the system, and under what rule.

6.2 Facts and data

Data Value Source
The Danube’s water level at Paks −133 cm Portfolio, 2 August 2026
The water level limit value for the plant’s safe operation −134 cm Portfolio, 2 August 2026
Shutdown of the last but one turbine 2 August 2026, 1:30 444.hu, 2 August 2026
Output during the shutdown 240 MW (instead of the nominal 2,000 MW) 444.hu, 2 August 2026
Estimated duration of the shutdown possibly longer than a month statement of Magyar Péter, Portfolio, 2 August 2026
Saturday’s consumption shortfall against expectations 500 MW Portfolio, 2 August 2026
Sunday evening’s consumption shortfall 700 MW statement of Kapitány István, HVG, 3 August 2026
Number of companies making voluntary undertakings more than 300 ATV, 3 August 2026
Total of the voluntary undertakings 400 MW ATV, 3 August 2026
Firms awaiting further joining 270 ATV, 3 August 2026
Evening freight transport shutdown between 17:00 and 22:00 444.hu, 2 August 2026
Paks residents left without hot water approx. 6,000 people 444.hu, 2 August 2026
Paks district heating consumption points affected 2,771 flats + 181 public institutions 444.hu, 2 August 2026
The solar association’s package of proposals 7 points Portfolio, 2 August 2026

A single connection deserves highlighting. The 700 megawatt saving is not a measured quantity but a difference: the divergence of actual load from a forecast load. The value of such an indicator depends entirely on the methodology of the forecast — the same consumption curve can yield a “saving” of anywhere between 400 and 900 megawatts alongside a different baseline. That is why it is not enough to disclose the end result: the methodology and the raw time series together make the figure verifiable. This is the real subject of Monday morning’s dispute, regardless of who is right.

6.3 Policy dimensions

  • Environment and climate (programme points) — calibrating energy market shock resilience to a multi-week operating mode, and the crisis variant of the just transition principle (programme point ID: K7, K5, K2);
  • Economy (programme points) — radical transparency of crisis data and the energy price shock preparedness plan (programme point ID: G19, G25);
  • Public administration and e-government (programme points) — an audit of the organisational routines of crisis decision-making: which organisation measures, which discloses, which decides (programme point ID: KI11);
  • Digitalisation and AI regulation (programme points) — machine-readable, downloadable publication of the load data (programme point ID: D9);
  • Transparency and anti-corruption policy (programme points) — the public trackability of the compensation payments (programme point ID: A1).

6.4 Literature in detail

6.4.1 Sergei Guriev – Daniel Treisman: Spin Dictators

The authors’ thesis is that modern authoritarian systems that hold elections are built not on open repression but on the handling of information: the leader magnifies the good news, postpones or reframes the bad, and claims the successes even when they stemmed from external circumstances. The volume deals separately with the fact that statistical disclosure itself also becomes a political instrument — the publication of sensitive data series is omitted, and favourable ones receive prominent communication. The essence of the authors’ argument is not a judgement on the character of individual leaders but a structural claim: where measurement and the communication of the measurement are in the same hands, distortion is not a question of intention but a consequence of the arrangement.

In the Hungarian situation this frame shows precisely why Monday morning’s data dispute cannot be resolved by rebuttal. MIAK firmly rejects the reading that the present Hungarian governance would belong in the category described in the volume — the argument is much simpler than that, and valid for every government. If the same actor measures, discloses and interprets the crisis data, then the credibility of the figure depends on the personal trustworthiness of the discloser. And the proposed solution is precisely for that reason not a request for trust but an institutional one: handing the measurement to the organisation whose task is not political success but the operational security of the system.

📖 Source: Sergei Guriev – Daniel Treisman: Spin Dictators

6.4.2 Graham Allison – Philip Zelikow: Essence of Decision

The volume analyses the Cuban missile crisis through three mutually complementary explanatory models. The second — the organisational behaviour model — starts from the premise that a government is not a single decision-maker but an assemblage of semi-autonomous organisations that operate according to their established procedures. In the authors’ description the political leadership ran into this again and again in the critical week of the crisis: in the name of flexibility and options it tried to disturb the organisational routines, and this resulted in a series of frictions. These same boring, repetitive procedures are at once the key to high performance and the possible fuse of catastrophe — because it is they, not the leader’s decision, that determine a hundred small, apparently insignificant details.

In the present Hungarian energy crisis this yields two practical conclusions. One: a curtailment regime announced daily and changing daily cannot be implemented at the level of organisational routines — the electricity supplier, the municipality, the district heating provider and the plant management of a large company can switch over in days, not hours. The other: the handling of the Paks district heating outage — shower containers, electric water heaters, the opening of public institutions — is today local improvisation, the ad hoc effort of the mayor’s office and the family support service. This is admirable, but it is not routine, so it is neither repeatable nor funded. Proposals 3.2 and 3.3 therefore follow from the same analytical frame: what has to be given to the implementing organisations is not heroism but procedure.

📖 Source: Graham Allison – Philip Zelikow: Essence of Decision

6.5 International comparison

The publication of raw system data is not a novelty in Europe but an obligation: the common transparency platform of the European transmission system operators (the ENTSO-E Transparency Platform) publishes, by member state and in hourly breakdown, the actual load, the generation mix and the cross-border flows, in downloadable form. The Hungarian proposal therefore does not ask for a new institution, but for crisis communication to be built on the same database that the country supplies in any case — and for the calculation methodology of the saving to be public alongside it.

On the question of compensation the West European demand response programmes give the model: there the large consumer keeps its flexibility on standby not out of goodwill but on the basis of a contract, for remuneration fixed in advance, and the activation threshold is known in advance. And the offer of help that came from the Slovak side is a reminder that regional mutual assistance too works quickly only if it is not an ad hoc gesture but a procedure laid down in advance.

Environment and climate

  • K7 — Energy market shock resilience
  • K5 — Just transition programme
  • K2 — Energy transition plan

Economy

  • G19 — Radical transparency in economic decision-making
  • G25 — Energy price shock preparedness plan

Public administration and e-government

  • KI11 — Organisational behaviour audit

Digitalisation and AI regulation

  • D9 — Data-driven public policy decision support system

Transparency and anti-corruption policy

  • A1 — Public money dashboard

Proposed new programme point: A crisis data disclosure standard — for the Public administration and e-government area: in every promulgated crisis situation the measurement data series underpinning the decision is to be published in raw, machine-readable form, with a fixed delay and together with a description of the methodology, and the publisher may not be the same as the deciding ministry.

6.7 List of sources

Press sources (MIAK press monitor, 3 August 2026 — topic 1):

Knowledge-base references (books):

  • 📖 Sergei Guriev – Daniel Treisman: Spin Dictators
  • 📖 Graham Allison – Philip Zelikow: Essence of Decision

Note: the local file path of the books does NOT appear in the visible text of the blog — only the author and the title. The file path is an internal matter of the generation process, not the reader’s.

MIAK internal materials:

  • MIAK policy area: Environment and climate (programme points; programme point ID: K7, K5, K2)
  • MIAK policy area: Economy (programme points; programme point ID: G19, G25)
  • MIAK policy area: Public administration and e-government (programme points; programme point ID: KI11)
  • MIAK press monitor, 3 August 2026 — topic 1, score: 95/100

Additional public data sources:

  • ENTSO-E Transparency Platform — hourly system load and cross-border flows
  • MAVIR — operational control and system load time series
  • National Directorate for Water Management — Danube water level and water temperature data

Generation metadata