A governance proposal — Draft 1.6

A democracy that keeps its promises.

Competent Democracy is a proposal for a political system built around binding, expert-approved manifestos, open-source measurement, and a tribunal that enforces what was promised.

"Many forms of Government have been tried, and will be tried in this world of sin and woe. No one pretends that democracy is perfect or all-wise. Indeed it has been said that democracy is the worst form of Government except for all those other forms that have been tried from time to time…" — Winston Churchill
Read the case
Diagnosis

Three problems modern democracies share.

The democratic settlement of the twentieth century delivered freedom and prosperity at unprecedented scale. It is also showing three persistent failures.

Short-termism

The electoral cycle rewards quick wins. Long-term challenges — climate, infrastructure, demographic change — fall outside the window in which a politician's career is judged.

Gridlock

Fragmented parliaments, opaque coalitions, and slow procedures mean that even agreed decisions take years to reach delivery. Citizens watch decisions evaporate between vote and outcome.

Accountability deficit

Politicians can promise what cannot be delivered, govern through manifestos no one tracks, and face no consequence beyond an election four or five years later. The cost of broken promises falls on the public, not the promiser.

The idea, in one breath

Stand for election on a manifesto of measurable goals; have those goals reviewed by independent expert committees and locked 90 days before polling; have progress scored by an open-source AI run by a separate non-profit; let citizens watch every score in real time; and if you don't deliver — or if you promised what could not be delivered — lose office, and lose the right to hold it again.

What it does not change

People differ — by upbringing, by life experience — and settle into political philosophies: conservative, liberal, socialist, libertarian, environmentalist. Competent Democracy keeps that exactly as it is. It takes no view on which goals a country should pursue. Parties still propose from their own convictions, and voters still choose between them. All that changes is that a goal must be stated clearly enough that anyone can tell, afterwards, whether it happened.

Mechanics

Five moving parts.

Each part is conventional in some other domain — corporate governance, professional regulation, scientific peer review. The novelty is the assembly.

01

Manifesto-lock

To stand for election, a party publishes a manifesto of measurable, time-bound, budgeted goals — each reviewed and approved by the relevant expert committee. The manifesto is locked 90 days before election day. After that, goals can only change through a defined Change Management process.

02

Expert pre-approval

Standing committees of experts — drawn from a single public roster maintained by an independent commission — review every goal before it appears on a ballot. Experts publish their political philosophical views and declarations of interest; they cannot belong to a political party. The design does not pretend expertise is politically neutral — it is partisan-aware but not partisan-controlled.

03

Open-source scoring AI

Progress is scored by purpose-built open-source software. An international non-profit (D2DEV) develops the framework; a national non-profit (D2OPS) operates it on the country's data. Code, parameters, and inputs are public, and scores are appealable.

04

Two tracks of accountability

Track A is collective and political: governments missing their milestones lose office automatically. Track B is individual and adjudicative: an independent prosecutor brings false-promise cases before a specialised Manifesto Tribunal.

05

Citizen App

Every locked goal, every score, every parameter change, every appeal, every override, every sanctions outcome is published on a single public application. The App is also the channel for electronic voting alongside paper balloting.

Architecture

How the institutions fit together.

Competent Democracy is a small number of named institutions, each with a single role, each holding a check on at least one other. The separations are deliberate: no single coalition should control both the rules of measurement and its operation.

Institutional map: parties select Expert Committee members from the public Expert Roster maintained by the Roster Commission. Committees review goals before manifesto-lock. The Manifesto feeds D2OPS, which runs the open-source D2DEV scoring framework and routes appeals to the Appeals Committee. The Academic Supervisory Committee audits the framework. Milestone-misses trigger Track A political dismissal; false-promise referrals go to the Manifesto Prosecutor and then the Manifesto Tribunal. The Citizen App publishes every goal, score, change, appeal, and outcome to citizens, who vote and petition through it.
Expert Roster Commission
9 members. Sole role: admit and remove experts from the public roster against published conditions.
Expert Committees
21 standing national committees by field. Members selected by parties from the roster; minimum diversity floors apply.
Academic Supervisory Committee
7 academics. Annual normative audit of the framework; standing on appeals; the system's intellectual conscience.
D2DEV / D2OPS
D2DEV owns the open-source framework internationally. D2OPS operates it nationally and holds per-goal specs.
Appeals Committee
5-person sub-committee of the D2OPS board. 70% supermajority to override a score.
Office of the Manifesto Prosecutor
Statutorily independent prosecutor's office. Chief Prosecutor appointed for a single 7-year term.
Manifesto Tribunal
3 sitting senior judges (plus 2 reserves) hearing false-promise sanctions cases. Appeals on law to the supreme court.
Citizen App
Single public surface for every goal, score, change, appeal, and outcome. Also the e-voting channel.
Measurement

An open-source scoring system, not a black box.

Measurement is the most consequential design choice in the system — and therefore the most heavily checked.

  • Public code. The full D2DEV codebase is open-source and developed in public. New releases are reviewed by independent academics before deployment.
  • Public parameters. Every per-goal spec — metric type, data sources, thresholds, milestones, baseline — is published at manifesto-lock. Every change is logged with a written justification.
  • Whitelisted inputs. Score inputs come only from a published data-source whitelist. App opt-in polls are advisory; they are never an input to a score.
  • Contestable. Government, opposition, expert committees, and the Academic Supervisory Committee all have formal standing to appeal a score. A 70% supermajority of the Appeals Committee is required to overturn it.
  • Non-decisional. AI suggests goals and measures progress. Every consequential decision — admitting a goal, approving a score, dismissing a government, sanctioning a politician — is made by named institutions against published rules.

Illustrative mock-up. The Citizen App is described in §8 of the spec.

Sanctions

Two tracks of accountability.

Political consequences for the government as a whole; adjudicative consequences for the individual politician who made a promise that should not have been made. They run in parallel.

Track A

Collective & political

Miss the milestones, lose the office. No individual fault required, no court involved — a clean ex-ante rule applied automatically.

  • Auto-dismissal. If fewer than 75% of due milestones score as "met" at the annual review, the government is dismissed without a vote.
  • Forward-ineligibility. Executive politicians who delivered less than 60% of milestones across the term cannot stand for executive office in the next term.
  • Long-term carry-over. A 70% supermajority is needed to drop a long-term goal inherited from a previous government.
Track B

Individual & adjudicative

Promises made outside the locked, expert-approved manifesto remain the principal source of cases. Brought by an independent prosecutor; tried as a civil-regulatory proceeding with full procedural protections. There is no custodial sanction — nobody goes to prison for a broken promise.

  • Fine. Up to two years' net ministerial salary, on proven negligence. A personal liability: it cannot be reimbursed by the party or the state.
  • Ineligibility. Bar from executive office for one electoral term.
  • Permanent ineligibility. For recklessness or intent, or a repeat finding within eight years — at the tribunal's discretion.
Realism

When the world changes: Change Management.

The system is designed to make promises stick. But a pandemic, a war, a natural disaster, a major economic shock can render previously-feasible goals materially unachievable. Change Management is the controlled, supermajority-gated exception — deliberately narrow, time-limited, and built so that "crisis" cannot become an indefinite excuse for suspending the rules.

01

Invoking it is hard

Three concurrent conditions: a formal motion (by the government, 25% of parliament, or two or more affected expert committees); certification by each relevant expert committee that the event materially affects the named goals; and a 60% parliamentary supermajority. The motion must name the triggering event, the goals affected, the adjustment, and a duration — and is published in full on the Citizen App within 24 hours.

02

A scalpel, not a curtain

Only the goals enumerated in the motion are affected; everything else keeps scoring and carrying sanction risk as normal. For listed goals, scoring is frozen at "under §6 review" and the false-promise defence covers only promises rendered unachievable by the cited event — not unrelated misconduct. Crisis-response goals can be added quickly, but they are themselves scored and sanctionable from day one.

03

It cannot outlive the crisis

Initial duration is capped at six months. Every renewal requires a fresh motion at the full invocation threshold. If no renewal passes, the override lapses automatically and the goals revert to live scoring and full sanction risk. Promises made during the override stay protected by it; promises made after the lapse do not.

04

It ends, then it is judged

Three independent early-termination paths: a government motion (simple majority), a force-end petition by 40% of parliament (simple majority to pass — ending an override is deliberately easier than starting one), and a public recommendation by the Academic Supervisory Committee. Within 90 days of any end, the committee publishes a review of the invocation — including whether anyone's conduct warrants referral to the Manifesto Prosecutor.

Anti-capture

Safeguards against capture of the measurement system.

The scoring system is the system's most powerful instrument and would be its most attractive target. Four cross-cutting safeguards apply at all times.

  • Opposition review at manifesto-lock. A per-goal spec cannot be approved by a majority that excludes the committee's opposition members. At least one opposition expert must vote in favour.
  • Independent academic review of framework changes. No D2DEV release reaches deployment without a published academic review of its diff — explicitly searching for normative choices embedded as technical defaults.
  • Annual normative audit. The Academic Supervisory Committee publishes a yearly report identifying framework or parameter choices that, in its judgement, encode contested values rather than neutral measurement. Findings are debated in parliament.
  • Whistleblower protection. Any D2DEV or D2OPS employee, contractor, or expert may file a confidential concern with the Academic Supervisory Committee. Retaliation is a criminal offence.
Academic, legal, political and public review

Read the full specification.

This site is a précis. The proposal exists at full length because the questions it raises deserve full-length answers. If you intend to engage substantively, read the spec.