---
title: "Mechanism Authority"
subtitle: "A persistent owner of the state's mechanism-correction chain"
author: "Elias Kunnas"
description: "Mechanism Authority 1.0 is the first installable correction layer for the unowned cross-institutional function: broad mandate, limited compulsion. Parliament retains final closure. 2.0 is a separately authorized module."
canonical: "https://mekanismirealismi.fi/mechanism-authority"
url: "https://mekanismirealismi.fi/mechanism-authority.md"
date_published: "2026-02-24"
date_modified: "2026-09-14"
llms_txt: "https://mekanismirealismi.fi/llms.txt"
---
# Mechanism Authority

A persistent owner of the state's mechanism-correction chain

*Elias Kunnas*

**Mechanism Authority 1.0 opens cross-sector mechanism questions, designs alternatives, couples findings to a public reply duty, and tracks accepted corrections through to outcomes.**

Parliament retains legislative power and political closure. Mechanism Authority's assessments are public, challengeable, and under independent supervision.

[Published analyses](/mev/index.md) · [Summary (FI)](mev-tiivistelma.md) · [Correction power: 2.0](mechanism-authority-2.md) · [Finnish 1.0](mekanismivirasto.md)

## The problem

Existing bodies measure, model, audit, or review fragments of state action. No actor owns the complete cross-system mechanism lifecycle. Findings from the National Audit Office, the Ombudsman, and the Chancellor of Justice can remain without a named implementation path: decrees stay in Finlex, the same illegalities are documented again, an accepted correction becomes a working group. The gap is a persistent owner that can open an ownerless problem or a better alternative, force a reply, and keep an accepted correction traceable.

**Knowledge can be produced in many places. The correction path must not be ownerless.** Universities, ministries, VATT, the National Audit Office, citizens, and competing models can produce analysis. Mechanism Authority concentrates ownership of the case's progress, not production of the truth. One procedural coupling point; many independent analysis points.

## Proposed capacity

Mechanism Authority 1.0 gives a persistent owner to the function that falls between existing bodies: opening cross-sector mechanism questions, designing alternatives, bringing a finding to a decision, and tracking an accepted correction. It can examine problems for which nobody is preparing a bill, and the structures that block correction.

Mechanism Authority 1.0 builds the analysis-and-correction chain. Political decision-owners select the correction, competent authorities deliver it, and Mechanism Authority keeps the chain traceable. A wider governing architecture sets a common direction and can change the state's own machinery.

Mechanism Authority can be installed at different scales, depending on what authority, capability, funding, and cooperation can actually be assembled. Statutory preparation of 1.0 is the [legislative assignment](mev-perustaminen.md). A larger reform opportunity can be pursued directly. Published [mechanism analyses](/mev/index.md) develop the protocol and form the first case stock. A statutory Mechanism Authority adds information access, a reply duty, persistent ownership, and delivery tracking.

## Core tasks

Mechanism Authority's institutional identity is these tasks:

- a persistent owner of cross-system evaluation, alternative design, and correction follow-through;
- own-motion work even when no bill is in motion;
- independent analysis and the information access the task requires;
- a specified finding coupled to a time-limited public reply;
- an accepted correction with named delivery and tracking;
- an effective challenge to Mechanism Authority's own assessments.

**In Mechanism Authority 1.0, Parliament enacts and repeals statutes.** Mechanism Authority opens, analyses, returns, and tracks mechanism cases. Pay, endowment, headcount, and board composition are implementation choices.

## Proposed operating contract

What follows is the selected 1.0 design in draft-section form. Evidence on existing Finnish capabilities is in the [evaluation-and-correction chain](mev-analyysi.md).

### 1 § Purpose and Mission

**Purpose of operations:** The primary purpose of the Mechanism Authority's operations is to maximize the long-term success and vitality of Finnish society.

**Means of implementation (Mandate):** To implement this purpose, the agency oversees, develops, and designs the steering mechanisms used by public authorities – such as legislation, funding models, and incentive structures – to ensure their coherence and functionality.

**Scope:** The agency's task is also to identify and report on such structural obstacles in society or new phenomena that jeopardize the realization of the operational purpose, regardless of whether they are currently subject to public regulation.

The agency's activities are guided by the goal of **maximizing** society's ability to maintain well-being and functional capacity across electoral terms. The full-accounting metric is remaining safety margin: how much capacity society has left to respond to threats not yet known. The metric is a trajectory, not an isolated drop in one stock. Spending a current buffer to build future capacity is not a mechanism failure merely because one present stock falls.

**Classification of a finding.** A mechanism assessment distinguishes: an effect contrary to the target; a missed target; an implementation gap; insufficient evidence; material harm to other societal capacities; and the existence of a better feasible alternative. The assessment names the comparison, the horizon, the material uncertainties, and the decision the finding supports. Missing a target does not by itself show that repeal or suspension is better than continuation or repair. Mechanism Authority may still argue that an adopted target itself weakens the safety-margin trajectory; that is a different claim from “the chosen mechanism cannot deliver the chosen target.”

### 2 § Proactive Design and Public Analysis

Mechanism Authority operates as a center of excellence for mechanism design:

- **Proactive consultation:** Ministries must consult Mechanism Authority during preparation of significant initiatives (over 100 million euro impact, structural reforms, or system-level incentive changes), before inter-ministerial consultation. Consultation is confidential. For minor proposals, consultation is voluntary. The goal is to identify problems at a stage where fixing them is still easy.
- **Mechanism design:** Mechanism Authority doesn't just evaluate proposals – it designs alternatives. "You want to incentivize X and avoid Y? Here are three mechanisms that achieve this. Option A works like this, costs this much. B works differently, costs that. We recommend B because..."
- **Modeling:** Mechanism Authority models the likely effects of mechanisms before deployment. Simulations, game-theoretic analysis, international comparisons. "If you implement this model, we predict these behavioral effects."
- **Mechanism library:** A public database of mechanisms that work and don't work. International comparisons: how have other countries solved similar problems? Documented failure patterns: this is how you fail.
- **State causal map:** Maintains and publishes a comprehensive causal map: an integrated model of all significant factors affecting society's capital stocks — including legislation, funding models, incentive structures, the tax system, demographic trends, technological development, international dependencies, informal institutional architecture, and implementation outcomes relative to legislative intent
- **Sector analyses:** Regular analyses of how mechanisms in different sectors perform – and how they should be designed. "How does healthcare financing work? How does public procurement affect quality? Where are incentives broken?"
- **Post-mortems:** When a major reform fails, Mechanism Authority produces a public analysis of causes and an alternative design. "Why did the healthcare reform fail? How should it have been done?" The goal is learning, not blame.
- **Structural analyses:** Analysis of decision-making architecture: how laws are made, how incentives affect decision-makers, and which structures prevent or enable course corrections. All structures are mechanisms and thus analyzable.
- **Election platform auditing (Dutch model):** Before parliamentary elections, parties can voluntarily submit their platforms for Mechanism Authority's calculation. Mechanism Authority evaluates the economic and dynamic effects of platforms using a unified model – "what does this promise cost and what does it produce?". This ends the culture of "unfunded promises" and forces parties toward honesty. (In the Netherlands, CPB's "doorrekenen" has been in use for decades and has transformed political culture.)

### 3 § Pre-legislative Review

All government bills must be submitted to Mechanism Authority's mechanism review before parliamentary consideration. The review evaluates:

- **Incentive structures:** Does the law create a situation where acting against its objectives is profitable?
- **Game-theoretic robustness:** How will rational actors (companies, citizens, authorities) respond to the rule?
- **Metric distortion:** Will the metric become a target that distorts behavior? (Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure.")
- **System effects:** How does the law interact with other mechanisms and the overall system?
- **Future-proofing:** Will the mechanism still work when circumstances change? (Demographic shifts, technological development, economic shocks.)

Review focuses especially on significant initiatives: over 100 million euro impact, structural reforms, or system-level incentive changes. Mechanism Authority is not obligated to comment on minor regulatory changes but may always do so at its discretion.

Mechanism Authority issues a public assessment of each significant bill's mechanism effectiveness and risks. If Mechanism Authority's assessment is positive or conditional, the law proceeds normally. If Mechanism Authority issues a **negative opinion** — finding the mechanism likely to produce outcomes contrary to its stated goals — a two-track procedural rule is triggered:

**First track (executive internal):** An amendment to the Government Rules of Procedure (*Valtioneuvoston ohjesääntö*) blocks a red-rated bill from advancing to the cabinet plenary session until the ministry has either fixed the identified mechanism failures or the Prime Minister has explicitly overridden the return. This is an internal executive procedure and requires no constitutional amendment.

**Second track (parliamentary transparency):** An ordinary law mandates that Mechanism Authority's negative opinion is attached as a mandatory annex to the government bill. If the government proceeds despite the negative opinion, the responsible minister must deliver a public response to parliament explaining why the bill advances despite Mechanism Authority's findings. The response and its justification become a permanent part of the law's preparatory materials.

Parliament retains final decision-making authority at all times. Effect comes from two mechanisms: the executive track forces delay and corrections; the transparency track makes override politically costly.

**What the opinion attaches to.** An opinion names the reviewed version of the proposal and the mechanism assessed. A later material change in incentives, funding, responsibility, or implementation is a new review need in the scope of that change; every editorial edit does not reopen a full review. The conditions of a conditional opinion, and whether they were met, are recorded. Review has a deadline, a reasoned extension rule, and an urgent route: a named decision-maker may proceed before the opinion is finished, but the departure is recorded and followed by subsequent review. Without version binding, scrutiny can be bypassed by changing the text. Without a time rule, scrutiny becomes delay power.

### 4 § Continuous Monitoring

Mechanism Authority monitors mechanism effectiveness throughout their lifecycle:

- **Real-time monitoring:** Monitors the implementation phase of mechanisms. Identifies problems during rollout – not years later in retrospective auditing.
- **Outcome measurement:** Compares actual results to intended societal effects (not process metrics)
- **Mechanism failure notice:** If a law produces the opposite result of what was intended, Mechanism Authority issues a public notice
- **Citizen feedback:** Citizens can report "broken mechanisms" (e.g., "I lose money if I accept work"). Mechanism Authority investigates and publishes verified findings.
- **Ministry response obligation:** A mechanism failure notice must be responded to within a deadline. If no response comes, Mechanism Authority publishes a "no response" notice and the matter moves to the named recipient via the §4a default route.
- **Automatic ex-post review:** For significant legislation (over 100 million euro impact), Mechanism Authority sets a three (3) year review trigger. Did the predicted effects materialize? If actual outcomes diverge significantly from forecasts, Mechanism Authority publishes a deviation report and the matter moves to the named recipient via the §4a default route. (Cf. Germany's Normenkontrollrat mandatory ex-post evaluation.)

### 4a § Own-Initiative Mechanism Opening

Mechanism Authority's task is not limited to reviewing government bills, existing laws, or analyses commissioned by ministries. The agency may, on its own initiative, publish a **mechanism opening** (*mekanismiavaus*) when it detects that a significant incentive, information, coordination, funding, responsibility-allocation, or oversight mechanism is producing public harm – or that a missing mechanism architecture in a domain where state capacity should exist is producing such harm – and when no active political or administrative process is addressing the issue adequately. A mechanism opening may also be published when a material improvement opportunity exists, without first establishing harm against an existing specification. That a mechanism is not broken must not silently become that it is not eligible for comparison.

Unlike the §4 mechanism failure notice, which concerns observed effects of an existing statute, a mechanism opening may concern a problem that has no active legislative owner at all. A mechanism opening is a public problem-opening, not a policy recommendation: it does not choose political values for parliament; it makes the mechanism question into an institutional object.

**Required structure of a mechanism opening.** Each opening must specify at minimum:

1.  **Phenomenon** – what is observably happening
2.  **Mechanism** – what rule, incentive, information asymmetry, veto structure, coordination failure, funding model, or responsibility allocation produces it
3.  **Actors** – who the players are
4.  **Incentives** – what each actor rationally does under current rules
5.  **Present state and comparison** – what result, operating pattern, or trajectory the current arrangement produces, and the feasible alternative it is compared with
6.  **Handling status and grounds for Mechanism Authority** – who already handles the matter, what current handling covers, and why independent review or alternative design is needed. Ownerlessness and a telos break are possible findings, not required properties of every opening
7.  **Evidence** – what data, cases, models, or institutional facts support the diagnosis
8.  **Uncertainty** – what would refute or weaken the claim
9.  **Owner** – which ministry, agency, committee, or political process is responsible for responding; if no owner exists, how ownerlessness is handled
10. **Mechanism paths** – what coherent solution families exist, without Mechanism Authority choosing political values

**Preliminary inquiry and formal opening.** Mechanism Authority may open a preliminary inquiry when there are credible grounds to suspect a significant mechanism problem or an unused improvement opportunity. A preliminary inquiry does not trigger all of the response duties of a formal opening. §2 sector analyses and mechanism design remain available without a formal opening.

A formal mechanism opening requires a specified and material case. The agency must explain why independent review or alternative design is needed. That explanation may contain explicit assumptions and open questions. A complete causal theory of why previous handling failed is not a prerequisite. An improvement opportunity need not be described as a failure or as ownerless.

**Public opening and technical appendix.** The agency publishes each mechanism opening in two forms: a 2–4 page public opening readable without specialized background, and a technical appendix with the full modeling, assumptions, data, and uncertainty analysis. The public opening contains a summary that names the stated goal, the rational behavior the incentive structure produces, the predictable result, and the hidden choice space.

**Handling ownerless mechanism failures.** If no ministry, agency, or committee owns the mechanism failure, Mechanism Authority records the finding as an **ownerless mechanism failure**. Ownerlessness does not remove the response duty. The installation package's default route is the cabinet collectively: it names a recipient within a deadline or records a reasoned decision not to act. A parliamentary committee handles the matter after a recipient is named or the refusal to name one is recorded. “The relevant committee” is not the default route when relevance itself is the dispute.

**Case-record ownership.** Mechanism Authority remains responsible for continuity of the case record until the finding is reasoned as rejected, a competent decision-owner has expressly chosen the next action, or handling is stopped on a specified ground. Sending the paper to another actor does not by itself transfer follow-up responsibility. Transfer of delivery responsibility requires a recorded recipient and a recorded scope of that responsibility.

**Mechanism-case register.** Each mechanism opening is entered in a public register. An open-failure view is one cut; an improvement opportunity is marked separately from an adverse finding. Status is not one scale. The register separates four dimensions that can be true at once:

- **Finding status:** open · upheld · contested · refuted · not yet assessable
- **Response status:** unanswered · answered · public override recorded
- **Implementation status:** not applicable · accepted · named owner · overdue · verified
- **Outcome status:** not yet assessable · met · missed · mixed

Example: *finding upheld; correction accepted; implementation overdue; outcome not yet assessable* is not the same as *finding upheld; political override recorded; mechanism unchanged; consequences under observation*. The second is not “resolved” in the same sense as a successful repair. The register also keeps these cases distinct: finding rejected; finding accepted but the decision-maker chose not to correct; correction accepted but unimplemented; correction implemented and its effect assessed.

**Accepted-correction ledger.** When a competent decision-owner accepts a correction, the record names the delivery owner, the resource decision or the unresolved resource requirement, the next deadline, the implementation state, the verification method, and a pre-authorized return to a decision forum if delivery fails. Recording a resource need does not itself appropriate money; a missing appropriation is an open decision, not funded capacity. A reply, a working group, or a plan does not mark the correction as implemented. Without this, the chain can be: Mechanism Authority finds a defect → ministry agrees → working group established → case marked answered → operating mechanism unchanged. This ledger does not import 2.0 repeal or suspension powers into 1.0. It makes an already chosen correction administratively real.

The register also shows given responses, predicted effects, and later outturn. It is Mechanism Authority's public calibration ledger: predictions do not disappear.

**Conditional calibration.** Each forecast names the mechanism, the decision, and the implementation assumptions it conditions on. Later evaluation separates forecast error, a changed decision, an implementation deviation, and changed conditions. Harm avoided because a warning changed the policy is not automatically scored as a false forecast. Uptake of a recommendation is not itself a successful effect.

**Mechanism alert.** If the agency assesses a mechanism failure as urgent, broad-impact, or hard to reverse, it may classify the opening as a **mechanism alert** (*mekanismihälytys*). An alert shortens the response time. It does not import a veto or automatic parliamentary consideration into 1.0. Important structural point: *every* formal mechanism opening triggers a response duty – the alert classification only changes urgency. This prevents classification from becoming a battlefield where parties bypass the response duty by classifying down.

### § 5 System-level monitoring

Mechanism Authority maintains a public dashboard of factors critical to society's long-term success. Crossing a threshold is a published finding and may open a § 4a case. In 1.0 it does not trigger automatic parliamentary consideration. Automatic effects belong only to separately legislated formulaic thresholds.

Thresholds are based on public analysis. Mechanism Authority can change indicators and thresholds but must justify changes publicly.

**Goal coherence report:** Mechanism Authority publishes an annual report evaluating the coherence of ministry actions with the government's stated strategic objectives. The report identifies where silo boundaries prevent goal achievement and where ministry actions cancel each other out.

### 6 § Independence and Expertise

Mechanism Authority's independence and expertise are secured through:

- **Independence:** Mechanism Authority 1.0 is established by special law, like the Bank of Finland and Sitra, with status “in connection with parliament” (cf. Constitution Section 90). A ministry cannot give instructions. The founding statute protects operational independence. Constitutional permanence is enacted separately.
- **Capital funding:** Mechanism Authority is funded by capital (Sitra model), not annual budget. This protects long-term work from budget cuts and removes salaries from political negotiation.
- **Board (executive body):** The board is responsible for Mechanism Authority's day-to-day operations, producing analyses, and publications. The first board is selected by an independent selection committee consisting of international academics and representatives of Nordic peer institutions. At least one member from mechanism design or algorithmic game theory, at least one with non-academic background. Parliament confirms the board by the procedure set in the statute. Thereafter the board appoints its successors, confirmed by the supervisory council.
- **Terms:** 7-10 year term, non-renewable. Removal only by decision of a three-judge panel.
- **International experience:** The director general must have at least 5 years of international experience. This breaks small-country network dependency.
- **Compensation:** Competitive, index-linked base salary benchmarked against international institutions and the [Singapore PSD model](https://www.psd.gov.sg/) (top quartile of private sector). No discretionary benefits, no performance bonuses. Official residence and graduated supplementary pension. Details in rationale.
- **Cooling-off period:** 24-month prohibition on working for supervised entities, at full pay.
- **Supervisory council (external audit body):** 5 members, of which 3 non-Finnish. Supervises board operations, approves methodology changes, handles removal proposals, and disposes of §7 challenges. It also reviews whether Mechanism Authority's selection of questions and allocation of effort serve its mission better than feasible alternatives. Mandate compliance is not evidence of adequacy. The first supervisory council is selected by the same independent committee as the first board; parliament confirms. Members serve staggered, non-renewable terms. The board does not appoint or renew its supervisor. International composition protects against local pressure and prevents network capture.
- **Diverse expertise:** Economists, game theorists, systems architects, behavioral scientists. Also mechanism designers who have built incentive systems in practice – algorithmic game theory, distributed systems. No single school hegemony.
- **International advisors:** Mechanism Authority can invite world-class experts to part-time advisory roles (senior fellow). This enables leveraging top expertise without full-time commitment.
- **Public methodology:** All analyses, data, and models open and challengeable.

### 7 § Powers

**Access to information:** Mechanism Authority has the right, notwithstanding confidentiality provisions, to receive free of charge from state authorities, institutions, and other entities performing public functions all information necessary for its tasks. Mechanism Authority has the right to direct technical access to information systems – raw data, bypassing ministry reporting layers. This means access to, for example, healthcare queue databases and social security benefit data directly, so Mechanism Authority can run its own analyses rather than relying on ministry-produced summaries. Confidential information received remains confidential in Mechanism Authority's possession.

**Modeling capacity:** Mechanism Authority has independent access to state microsimulation models (especially the SISU model) and the right to run alternative scenarios. This enables independent counter-modeling – Mechanism Authority can test ministry claims rather than relying solely on ministry-produced calculations.

**Publication right:** Mechanism Authority has the right to publish analyses without prior approval. The government cannot prevent or delay publication.

**Limits of authority:** Mechanism Authority cannot repeal laws or enact new ones. A negative pre-legislative opinion triggers mandatory return and a public response obligation via a two-track procedure (§3), but parliament retains final decision-making authority. Effect comes from the binding nature of pre-legislative review, publicity, automatic alerts, and making political responsibility visible.

**Public challenge against Mechanism Authority.** An outsider's correct criticism must not depend solely on Mechanism Authority's willingness to recognize it. Anyone may file a challenge that identifies an official finding, the alleged error, and the evidence. The supervisory council (§6) decides admissibility and disposes of the challenge; a refusal is independently contestable. The original author does not alone decide the appeal of their own finding. A material dispute names, when warranted, a reviewer independent of the original assessment. Confidential source material remains confidential; the challenged claim is stated at the level that can be answered. Possible dispositions: independent re-examination, correction of the official finding, reconsideration of a classification, publication of dissent, or review of the method. The disposition changes the official status of the assessment. The challenge, the reply, and the disposition are published automatically, subject to confidentiality limits.

**Capacity protection.** Own-motion openings and constructive search are protected from routine bill review. Mechanism Authority samples cases that screening missed for structural risk. Workload is measured across the whole chain — follow-up, challenges, model maintenance, openings — not only counted bill opinions. A capacity shortfall is published and resolved by a service-level or resource decision.

## Rationale

### 1. Nobody measures whether it works

The state operates without knowing whether its actions work:

- **Parliament** sets a goal ("temperature 21°C")
- **Government** implements ("turn on the heating")
- **Nobody measures** whether the temperature is actually 21°C, whether the heater is broken, or whether an open window negates the effect

Mechanism Authority 1.0 couples measurement to existing decision-owners: it measures, names the defect, and requires a reply.

### 2. Proactive beats reactive

When a bill arrives for review, political capital is already committed. The ministry has committed to a direction, and correction is expensive. Review becomes confrontation rather than collaboration.

Proactive consultation reverses this: Mechanism Authority is a resource, not an inspector. "We're considering this mechanism – what could go wrong?" Problems are found at a stage where fixing them is still easy and there's no loss of face.

Public sector analyses make a defect visible without a bill. Visibility does not itself produce a correction; it lowers the work required for an existing decision-owner to act.

### 2a. Mechanism Authority also opens new questions

Reactive review power is necessary but insufficient. A verifier is always downstream of politics: it answers questions others have chosen to ask. The largest mechanism failures, however, are not in active proposals – they lie outside, in matters that have no owner, whose losers are diffuse, whose winners are organized, or whose costs fall on future governments.

Three layers help clarify what Mechanism Authority decides and what it does not:

| Layer | Who decides | Mechanism Authority's role |
|----|----|----|
| **Values** (what Finland should pursue) | Parliament · government · voters | no authority |
| **Direction** (what the chosen goal requires as a mechanism) | Politics, but forced into legibility | strong agenda- and option-mapping role |
| **Mechanism fit** (does the chosen structure work) | Mechanism Authority is authoritative on the procedural assessment | strong analytical role |

**Authoritative on mechanism fit** means authority to issue the official procedural assessment and trigger its response path. It does not mean exclusive ownership of causal truth. Material countermodels, minority findings, challenges, and unresolved residuals remain part of the official record.

The mechanism opening power (§4a) operates on the middle layer. It does not choose values, but it also does not wait for politics to open the question. It forces *politics to name the choice* when the choice is being disguised as technical necessity, or when it is not being made at all.

**Five-category taxonomy.** The failures that mechanism openings address fall empirically into five categories that normal political processes do not handle:

1.  **Ownerless cross-ministry failures** – each ministry controls one part of the mechanism, but no one owns the resulting equilibrium. Classic case: work incentives across taxation, benefits, childcare fees, housing support, and debt enforcement.
2.  **Diffuse-loser, concentrated-winner failures** – everyone pays a little, one group benefits a lot, so the mechanism survives.
3.  **Future-person failures** – demography, pensions, infrastructure decay, defense-industrial capacity, fertility, education quality, debt sustainability. The injured party is not at the table.
4.  **Hidden-adjustment failures** – politics claims all goals are achievable, but the system adjusts through queues, quality decline, debt, rationing, burnout, discretion, or under-enforcement.
5.  **Missing-architecture failures** – not "this proposal is bad," but "an entire policy domain lacks designed mechanism architecture even though state capacity should require it." The Pekka Kuusi gap: since the 1960s, Finland lacks systematic capacity-building mechanism design.

The categories are not a static list in the statute – they are an analytical structure showing where current systems systematically fail. Freezing the list into legislation would exclude a sixth category we have not yet identified.

**Audience binding.** A mechanism opening fails if it becomes a PDF no one reads. The fate of the National Audit Office is instructive: good reports, weak binding force. Unlike central banks, Mechanism Authority cannot rely on real-time market reaction as its audience-binding mechanism. Binding must be built from three components: **procedure** (response duty, deadline, committee handling), **public memory** (open mechanism-failure register, prediction calibration over time), and **quotable conflict** (public summary that makes the hidden choices in politics visible). Without these three, problem-opening becomes Sitra-style futurism – interesting, but institutionally inert.

**Relation to the broader framework.** The mechanism opening power is the concrete institutional answer to the broader problem named in [The Unpopulated Meta](https://kunnas.com/articles/the-unpopulated-meta): meta-level institutional architecture design is a near-empty layer. Mechanism Authority's value is not only in detecting failures – it is in building the public object and procedural loop through which a failure becomes governable.

### 3. The missing function

Every actor in the state apparatus has a role:

- **Ministers** make policy
- **Civil servants** draft and implement
- **Lawyers** check legal form
- **Finance ministry economists** prepare impact assessments
- **National Audit Office** audits use of funds
- **Constitutional Law Committee** evaluates constitutionality
- **Council for Regulatory Impact Analysis** comments on impact assessment quality

**Why are existing bodies insufficient?**

- **National Audit Office (VTV)** can assess results and impacts, but audits are selective, largely retrospective, and tied to public finance. A finding can lead to a recommendation and follow-up; it does not automatically produce an accepted-correction ledger. See the Finnish [evaluation-and-correction chain](mev-analyysi.md).
- **Finnish Council of Regulatory Impact Analysis** is advisory and selective (about 30–50 opinions a year). Opinions can make gaps visible; they do not create a duty to repair, implement, or reopen. The [2024 annual review](https://julkaisut.valtioneuvosto.fi/items/f8ffe18a-1e71-4f8a-a08f-1718f3830c7d) reported an average of 3.4/5; the 2025 review also reported improvement. The Economic Policy Council warned about SOTE's soft budget constraint — that is a different body.
- **Finance ministry economists** produce forecasts AND prepare government decisions. [The ministry defends this](https://vm.fi/-/valtiovarainministerion-ennusteen-riippumattomuudesta), but the structural conflict of interest remains.

These bodies have substance. The gap is a persistent owner of the cross-system chain: opening, alternative, decision, implementation, and learning. Mechanism Authority uses existing expertise; it does not replace it with the claim that nobody asks whether a mechanism works.

Ministries optimize their own silos, but nobody looks at the whole – another missing function. Result: incentive traps where working doesn't pay. Each part is optimized, but the whole is broken.

**Shadow government:** Currently the missing function is filled by management consultants. [McKinsey](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/seizing-finlands-growth-opportunity), Boston Consulting Group, and Deloitte have been deeply involved in preparing major reforms (healthcare, regional government). Problems: (1) international standard models don't fit Finnish context, (2) consultants have financial incentives to please the client, (3) calculation models are secret as "trade secrets." This "shadow government" wields significant power in shaping reform parameters, but bears zero responsibility for implementation success. When consultants leave with their PowerPoints, civil servants are left to solve practical problems. Mechanism Authority replaces this with public, accountable capacity.

### 4. Law is code that runs on society

If software development deployed code to production without testing, systems would crash immediately. In legislation, this is normal practice: pass a law and "see what happens."

Mechanism Authority is **quality assurance** for mechanisms. It tests mechanism logic before deployment and measures effectiveness afterward.

### 5. Case study: Healthcare reform

Finland's 2023 healthcare reform (SOTE) funding model creates a situation where:

- Regions receive more funding the sicker their residents are
- The state covers deficits
- Efficiency is not rewarded

The Audit Committee identified during preparation that the reform's two main objectives – reducing health disparities and containing costs – are mutually contradictory and cannot both be maximized. The mechanism design flaw was known. But the committee couldn't stop the "legislative train" – it has no emergency brake.

**Four predictable mechanism failures:**

1.  **Soft budget constraint:** Regions have no taxing power, but they know the state will bail them out due to constitutional service obligations. Classic moral hazard – identified in economic literature, but its magnitude was underestimated.
2.  **Wage harmonization:** When staff transferred from hundreds of employers to one, wages rose toward the upper end. Hundreds of millions in additional costs that weren't anticipated.
3.  **ICT integration transaction costs:** Consultant calculations assumed IT system consolidation would bring savings. In reality, maintaining parallel systems during transition increased costs.
4.  **Outsourcing price rally:** Large, rigid regions couldn't negotiate effectively. Staffing agencies priced their services high, exploiting labor shortages.

2024 deficit forecasts grew to over 1.4 billion €. The original promise was 3 billion € in savings. Each of these mechanism failures was predictable – if anyone had asked: "How will rational actors respond to these incentives?"

Mechanism Authority's pre-legislative review would have made these findings public, mandatory to address, and automatically subject to parliamentary decision: should we pass a law whose mechanism is documented as broken?

Now one can say "we didn't know." Mechanism Authority removes this defense.

### 6. International models

Long-lived civilizations have developed similar organs: Roman censors, the Chinese imperial censorate, Venice's Council of Ten. Common feature: a constitutional body insulated from immediate political pressure, tasked with maintaining system functionality over a longer horizon than any election cycle.

**Modern comparisons:**

- **CPB (Netherlands):** [Centraal Planbureau](https://www.cpb.nl/en) independently analyzes the economic effects of bills. Its calculations have strong institutional and public authority. [Election-program scoring](https://www.cpb.nl/en/charted-choices-2025-2028) is voluntary; results do not legally bind parties, but the political cost of ignoring them can be high. Mechanism Authority extends the model: not just economic effects, but mechanism effects and own-initiative openings.
- **RSB (EU):** The Regulatory Scrutiny Board reviews the intervention logic of Commission impact assessments. A negative opinion requires revision; after two negative opinions a specified political route remains. This is a precedent for a procedural gate with an override.
- **RPC (UK):** Regulatory Policy Committee rates impact assessments. A weak assessment requires explicit ministerial justification to parliament.
- **NKR (Germany):** The Normenkontrollrat reviews all bills (~200/year) for compliance costs. NKR's distinctive feature is mandatory ex-post evaluation: legislation exceeding one million euros in compliance costs must be re-evaluated after 3-5 years. Did the predicted effects materialize? This closes the feedback loop that other models leave open.

Mechanism Authority combines these: CPB's analytical capacity, RSB's procedural gate with an override, RPC's public rating, and NKR's mandatory ex-post evaluation. And adds: own-motion opening, an accepted-correction ledger, and full lifecycle monitoring, which none of these do.

### 7. Constitutional implementation roadmap

Mechanism Authority 1.0 mechanisms are designed to be implementable without constitutional amendment. Each mechanism maps to an existing legal instrument:

| Mechanism | Legal instrument | Constitutional assessment |
|----|----|----|
| Mandatory return of red-rated bill (§3, first track) | Amendment to Government Rules of Procedure (*Valtioneuvoston ohjesääntö*) | Internal executive procedure. No PeV issue – the government already constrains its own agenda through procedural rules. |
| Public response obligation (§3, second track) | Ordinary law | Comparable to existing comply-or-explain mechanisms. Likely passes PeV review. |
| Ex-post review trigger (§4) | Ordinary law | Standard legislative mandate. No constitutional issue. |
| Published threshold crossing (§ 5) | Finding and possible § 4a opening | Does not by itself place the item on parliament's agenda. Automatic effects belong only to legislated formulaic thresholds. |
| Accepted-correction ledger and challenge (§4a, §7) | Ordinary law | Record-keeping and appeal, not new legislative power. |
| Mechanism Authority independence and mandate (§6) | Special law (*erillislaki*) | Requires normal legislative process. Precedent: Intelligence Oversight Ombudsman 2019 (*Tiedusteluvalvontavaltuutettu*). The founding statute protects operational independence; constitutional permanence is enacted separately. |
| Sunset and emergency suspension (Mechanism Authority 2.0 §§8–12) | Constitutional amendment (Section 73) | Ordinary route: a majority leaves the proposal over an election; the next parliament adopts by two-thirds. Urgent route: five-sixths, then two-thirds in the same parliament. Not two successive two-thirds votes. |

Mechanism Authority 1.0 is designed to *seek* installation under the present constitution. The clauses still need article-by-article legal validation. “Installable” is not already demonstrated. Drafting 2.0, enacting 2.0, and activating those powers are three different events. Supporting 1.0 does not authorize 2.0, and a published crisis protocol does not activate itself.

**The same authority contract on every surface.** 1.0 does not prevent enactment. A negative opinion returns a bill inside the cabinet until the ministry repairs it or the Prime Minister overrides; parliament receives a public ministerial reply. A mechanism opening requires a reply, not a correction. An accepted correction requires a delivery ledger, not a working group. 2.0 expiry and suspension are not part of 1.0.

### 8. International pressure

Finland has fallen behind international recommendations:

- **GRECO (Council of Europe anti-corruption body):** Finland has fulfilled only 4/14 recommendations from the fifth evaluation round (2024). Finland is under "enhanced monitoring."
- **EU rule of law reports:** The Commission has repeatedly noted Finland's deficiencies – especially regarding "trading in influence" and criminalization of foreign bribery.

This is not just reputational risk. The EU increasingly ties rule of law requirements to funding mechanisms. Finland's credit rating (AA+/AA) and EU funding require that oversight mechanisms are in order.

**Mechanism Authority's role in anti-corruption:** Mechanism Authority is not a criminal investigation body – investigation of individual bribery and abuse cases belongs to police and prosecutors. Mechanism Authority's role is *structural*: it analyzes whether transparency mechanisms (like lobbying registers) work, whether rules create conflicts of interest, and whether procurement laws favor certain actors. This is analysis of "corruption-enabling architecture" – not chasing individual cases.

### 9. Compensation design

Mechanism Authority's compensation structure is designed using mechanism design principles. Goal: attract world-class expertise without political suicide.

**Lesson from the VTV scandal:** A Finnish National Audit Office scandal showed that the problem wasn't high salary but *opaque discretionary benefits* – airline points, beauty treatments, unclear travel expenses. Small lifestyle perks destroy legitimacy more effectively than high transparent salary. Mechanism Authority's solution: high base salary, zero discretionary benefits, nothing that ends up in tabloid headlines.

**Salary level and comparisons:** Specific figures are defined in a separate implementation plan. The reference frame is not domestic ministries but international institutions (IMF, ECB) and the **Singapore Public Service Division (PSD) model**, where key positions are compensated at the private sector top quartile. Finnish public sector economist salaries do not compete for world-class talent — Mechanism Authority's compensation must close this gap to attract the required expertise.

**Graduated commitment structure:** Official residence, graduated supplementary pension, and academic affiliation (adjunct professorship, research sabbaticals) are designed to bind top talent to long-term work and make early departure to a supervised entity financially unattractive. The structure compensates for the purchasing power differential that expatriate Finns face when returning from international institutions.

**Why no performance bonuses?** Deliberate choice. Performance bonuses create distorted incentives and short-termism. Mechanism Authority's impact is measured in decades, not quarters.

**Political framing:** Mechanism Authority's supervised mechanisms control billions. Preventing one policy error pays back the director general's entire career salary a hundredfold. Right framing: "cheap insurance" and "patriotic return," not "high salary."

### 10. Small country network risks

In Finland, the relevant talent pool is small and networks are dense. This creates two capture mechanisms:

**"Old boys' network":** Officials, politicians, and business representatives have attended the same schools, served in the same organizations, and meet at the same events. Formal independence doesn't help if supervisor and supervised are decades-long acquaintances.

**Future employer constraint:** A regulator has practically only a few potential employers in Finland (major banks, pension companies, ministries). This creates an implicit incentive to regulate gently.

**Mechanism Authority's responses:**

*Cooling-off period at full pay:* 24-month prohibition on working for supervised entities, 100% salary during cooling-off period. Paid transition time enables international return or academic career.

*International experience requirement:* Director general must have at least 5 years of international experience. A person whose career is at the IMF or Fed is not dependent on Finnish banks' or ministries' approval.

*Diaspora recruitment:* Mechanism Authority's target group is Finns already abroad. Culturally competent, detached from networks.

**First board recruitment:**

The first board determines the agency's culture. Selection errors at this stage become permanent. Selection is based on work samples, demonstrated capability, and independence from the entities Mechanism Authority evaluates. Affiliation is neither a substitute for capability nor a bar to it.

Independence means a member is not bound to a supervised decision-owner or its preparation machinery in a way that would block an adverse finding. It does not mean insulation against all effective correction. Control by one domestic network is blocked by international selection, work samples, and the supervisory council — not by categorically banning every central-bank, research-institute, or ministry career path.

The following are starting points for scrutiny, not automatic bars: a current post in a supervised ministry; current lobbying in a supervised field; a current leadership post in a supervised research institute. The bar is a verifiable conflict, not the name of a career path.

### 11. Addressing counterarguments

**"Who watches the watchmen?"** — Public methodology is not enough. §7 specifies a public challenge: standing, admissibility, named respondent, deadline, independent disposition by the supervisory council, remedies that change the official assessment, and automatic publication. The original author does not alone preserve the finding. An outsider's correct criticism must not depend solely on Mechanism Authority's willingness to recognize it.

**"Who decides what's a good mechanism?"** – Mechanism Authority's assessments are based on public methodology: incentive analysis, game theory, system dynamics. A finding names the decision it supports. Methodology is challengeable on the §7 route. Predictions are testable against the mechanism and implementation they conditioned on: a warning that changes the policy is not automatically a false forecast. Unlike political promises, Mechanism Authority's claims don't disappear.

**"This adds bureaucracy"** – On the contrary: Mechanism Authority consolidates fragmented functions. Currently mechanism oversight is scattered: Council for Regulatory Impact Analysis (scarce resources), National Audit Office's transparency register (disconnected from other analysis), Strategic Research Council studies (which nobody reads), and ministries' own impact assessments (written after the political decision). Mechanism Authority unifies these functions in one institution: the council's analytical work, transparency register data, and integration of research into the legislative process. Result: one agency instead of fragmented committees.

**"Ministries won't listen"** – Hence publicity. If a ministry ignores Mechanism Authority's warning and the law fails, responsibility is documented. A public prediction creates political cost for ignoring it.

**"Experts aren't neutral"** – Hence diverse expertise (§6): economists, game theorists, systems architects – no single school hegemony. Hence public methodology: anyone can challenge the analysis. And hence long, non-renewable terms: no need to please political masters.

**"We already have the Council for Regulatory Impact Analysis"** – Mechanism Authority extends the council's narrow base function – pre-legislative review – into an entirely different kind of institution: a governor of governance. The council checks impact assessments; Mechanism Authority covers the full lifecycle, designs mechanisms proactively, and analyzes systems at system level. The council is Mechanism Authority's prototype, not its copy – but politically, the upgrade framing is correct. Internationally, every effective oversight body (EU RSB, UK RPC, German NKR) emerged from strengthening existing advisory bodies, not from creating new institutions from scratch. In the Netherlands, CPB started as advisory (1945), built credibility over decades, and became politically binding before legislation confirmed its status (2012–2013). The council is important but insufficient: (1) It reviews only draft bills – authority ends when the law is passed. (2) It's purely advisory – 32% of its recommendations are completely ignored. (3) Total budget approximately €280,000/year, 2.5 FTE secretariat. They can read a proposal and check logical consistency but cannot re-run ministry econometric models. (4) Results aren't improving: the average score in 2024 opinions was 3.4/5, lower than the previous year. In the SOTE reform, the council warned about the soft budget constraint – the government ignored the warning, and the exact predicted mechanism failure materialized at a cost of billions. The council reacts to bills; Mechanism Authority can also generate them.

**"The Yli-Viikari case shows independent agencies get corrupted"** – The National Audit Office scandal is a warning about bad architecture, not a reason to abandon independence. Mechanism Authority's structure is designed with this lesson in mind (see section 9): high transparent salary, zero discretionary benefits, mandatory international evaluation, three-judge removal procedure. Ethics is an engineering problem: build constraints, don't rely on virtue.

**"Procedural rules limit democracy"** – Amending the constitution requires a 2/3 majority. The central bank's mandate cannot be changed by simple majority. The life expectancy coefficient cuts pensions automatically without any parliamentary vote. Democracy routinely constrains its future self when it recognizes its susceptibility to short-termism. Mechanism Authority's mandatory return and public response obligation follow the same principle: parliament binds its future self because it knows political pressure will override mechanism analysis when the analysis is politically inconvenient. Mechanism Authority's override is always possible – but never invisible. This is not limiting democracy but protecting it from its own weaknesses.

**"The mandate is too broad"** – The mandate must be broad because the problems are systemic. A narrow mandate ("review HE impact assessments") already exists – that's the Council for Regulatory Impact Analysis, and 32% of its recommendations are ignored. System-level mechanism failures arise precisely from cross-interactions between laws, funding models, demographic trends, and informal institutions. If the mandate is scoped down to be politically comfortable, the function dies inside the scope. In practice, breadth is managed through prioritization: Mechanism Authority focuses on significant initiatives (over €100M, structural reforms) and selects its own analysis targets. Breadth doesn't mean doing everything simultaneously – it means nothing is excluded from scrutiny a priori.

**"This is too experimental"** – The status quo is what's experimental: legislation is the only complex system that isn't tested before deployment. Cybersecurity, military, finance, AI – all have solved this. The risk isn't establishing Mechanism Authority; the risk is not establishing it.

### 12. Mechanisms of mechanisms

Laws are mechanisms. But the legislative process is also a mechanism. The electoral system is a mechanism. The Constitutional Law Committee's role is a mechanism.

Mechanism Authority's authority covers all levels: individual laws, their interactions, and structures that produce laws. If some structure systematically prevents the emergence of functional mechanisms, Mechanism Authority must analyze it – even if Mechanism Authority cannot change it.

Mechanism Authority's task is to improve mechanisms. The constitution is a mechanism. If it doesn't work, Mechanism Authority says so – and says how it should work. What parliament does with this information is parliament's business.

### 13. Fix it yourself or it gets fixed for you

States that don't fix their own mechanisms eventually become subject to external correction. Greece in the 2010s, Argentina repeatedly, Italy under EU pressure – when internal correction capacity fails, the IMF, ECB, or EU set the terms.

Finland isn't there yet. But the direction is clear:

- Public debt is growing structurally
- GRECO and the EU have already noted deficiencies
- The healthcare region funding model is documented as broken
- Pension system sustainability requires corrections that aren't being made

Without internal correction capacity, Finland will drift into a situation where outsiders – credit rating agencies, the IMF, the European Commission – define what must be done. Then the choice is no longer *what* to fix but *how quickly* to comply.

**Mechanism Authority is a sovereignty investment.** It builds the capacity to identify and fix mechanism failures before they accumulate into crisis. The alternative isn't "no correction" – the alternative is "correction on external terms, without discretion."

External correctors optimize for their own objectives: the IMF for creditors, the European Commission for eurozone stability, credit rating agencies for investor risk management. These are not the same as the long-term interests of Finnish citizens. Greece's austerity served German and French banks, not Greek pensioners. Mechanism Authority is the only institution whose explicit mission is to optimize for Finnish society's success – not creditors, not the eurozone, but Finland.

Countries that build their own analytical capacity (Netherlands, Singapore) stay in control. Countries that don't (Greece, Italy) lose control. Finland can choose which group to join.

### Background research

**[Finland's evaluation and correction chain (FI):](mev-analyysi.md)** Strong but fragmented partial capabilities; five separable breaks. Mechanism Authority 1.0 is the proposed persistent owner of the unowned cross-institutional function.

### Academic background

Academic **mechanism design** is an established field of economics. Nobel Prizes in 2007 (Hurwicz, Maskin, Myerson) and 2012 (Roth, Shapley) recognized the field's importance. Mechanism design is applied in practice – auctions, school choice, organ donation – but applications are narrow and bounded.

Adjacent fields include: game theory (mathematical foundation), public choice theory, institutional economics, behavioral economics, and Regulatory Impact Assessment (RIA). However, none of these systematically asks of legislation: "How will rational actors respond to these incentives? What is the equilibrium when everyone has adapted?"

Comparison with other fields:

| Field         | Stress testing                      | Maturity    |
|---------------|-------------------------------------|-------------|
| Cybersecurity | Red teaming, penetration testing    | High        |
| Military      | War games, red team                 | High        |
| AI safety     | Jailbreaking, specification testing | Medium-high |
| Finance       | Stress tests (CCAR, EBA)            | Medium      |
| Legislation   | —                                   | Undeveloped |

The most developed work happens in fields where "code is law": in AI safety, researchers study "specification gaming" (achieving goals while circumventing rules), and in distributed systems (DAOs), tools like cadCAD are used to simulate governance systems before deployment. The lesson from these fields: you cannot secure a system without trying to break it.

Mechanism Authority would institutionalize this approach for legislation. In software development, a similar shift occurred when testing moved from ad hoc activity to its own function – Quality Assurance (QA). Mechanism Authority would do the same for legislation: systematic testing before deployment, monitoring during operation.

[![Mechanism Realism – A User Manual for Civilization](assets/cover_front_600.jpg)](/index.md)

MEKANISMIREALISMI: A USER MANUAL FOR CIVILIZATION

Why doesn't Finland work, despite everyone doing their best? Because nobody looks at incentives, feedback loops, or time horizons – only people, parties, and promises. The book presents the diagnosis and the fix. Mechanism Authority is one part of it. (Finnish only)

[Buy the book](osta.md) [Read more](/index.md)

## The Framework Behind This Proposal

Mechanism Authority is not a standalone policy idea. It is an institutional implementation of a broader framework developed across [100+ essays](https://kunnas.com/articles/) asking: what does physics require for complex systems to persist? The institutional case for lifecycle ownership does not require accepting Telic Systems, syntropy, or a physics-derived political telos; the broader corpus is one motivation and source of analytical primitives.

The argument runs: governance without an explicit purpose optimizes for whoever captures it ([Telocracy](https://kunnas.com/articles/telocracy)). Every surviving civilization developed a guardian function to audit whether institutions serve their stated purpose; modern democracies eliminated it ([The Fourth Branch](https://kunnas.com/articles/fourth-branch)). Character dissolves in bad incentives – only architectural constraints persist under optimization pressure ([Ethics Is an Engineering Problem](https://kunnas.com/articles/ethics-is-an-engineering-problem)). Mechanism Authority is the Finnish instantiation of this architecture: the missing feedback loop between stated intentions and actual outcomes.