Part IX, Separation of Powers & Checks

Derived from Axiom 4 (every power has a proportionate check, including the checkers) and Axiom 12 (subsidiarity). This arranges every institution in the model so that no branch, including the new ones (experts, integrity, sortition), can dominate, and so the whole forms a closed web of mutual accountability.

IX.1 The branches of the state

The classic three-branch model is insufficient here because this design adds expert execution, an integrity function, and selection-by-lot. The model has five loci of power, each checking the others:

No branch holds power in its own right; each is delegated and checked (Axiom 1, 4).

Minimal sufficient institutions (Axiom 14). Five branches, two legislative chambers, and a small set of single-purpose independent authorities (fiscal, monetary, boundary, appointments, migration) is the complete institutional set, there is nothing further to learn. The independent authorities are deliberately narrow: each does one technical job under the same transparency and oversight as everything else, and any that could be merged without loss is merged (the existing patchwork of watchdogs is consolidated, §XV.5). Completeness is bought with bodies; complexity is contained by giving each one a single published job and a citizen-facing summary (§II.0). The marginal test is not left as an assertion: every standing body is justified item-by-item, and each consolidation recorded, in the complexity ledger (docs/COMPLEXITY.md) — which also fixes the citizen's comprehension load at a counted sixteen concepts (the 1·5·4·4·3 map of §0.8), so "the machinery is minimal and the citizen's load is bounded" is a proof, not a claim.

The Axiom 11 vs Axiom 14 trade rule. Anti-capture (Axiom 11) tends to add safeguards; minimal-complexity (Axiom 14) warns each is cost, attack surface, and opacity. They are reconciled by an explicit marginal test: a safeguard is justified only if the capture-resistance (and outcome quality) it adds exceeds the attack-surface, cost, and citizen-comprehension burden it creates. Each Part's failure-mode table is the evidence its safeguards pass this test, and any proposed new body must pass it too. This is why the design claims to be the minimum complexity that satisfies every axiom, not merely a large one (docs/COSTING.md §1).

IX.2 The bicameral legislature, election and lot

The mandate (Part III) is expressed through two chambers that check each other:

Representative Assembly (elected by STV+, Part III)

• Proposes and makes law; sets national priorities and budgets; confirms appointments (§IV.4).
• Accountable to the electorate; proportional and locally linked.

Citizens' Sortition Chamber (selected by lot, Part II.3)

• Demographically representative, paid, expert-briefed, deliberating over time.
• Scrutinises legislation; can require reconsideration / delay (a suspensory check, not a permanent veto); deliberates the value trade-offs flagged under §V.4.
• Immune to campaign money, lobbying, and re-election incentives, the antidote to the elected chamber's characteristic weaknesses.

Interaction: the Assembly governs and legislates; the Sortition Chamber reviews, delays, and decides ordinary value trade-offs in the first instance (§V.4.2 Tier V-a). Its value ruling binds the experts' execution, but is suspensory before the Assembly and the people: the Assembly may substitute its own value judgement only by a transparent supermajority, and either may be put to the people by referendum on petition — while constitutional-magnitude or rights-touching value questions go to the people directly (Tier V-b). "Decide" and "suspensory" are thus reconciled: the Chamber decides first; the electorate has the last word. Neither chamber can dominate, and the Chamber cannot legislate alone. Election supplies accountability; lot supplies incorruptibility; together they supply both.

Scrutiny cannot be bypassed — but the protected class is defined narrowly, to avoid gridlock. The un-bypassable classes are constitutional and franchise changes, rights-limiting measures, and structural tax changes — not every appropriation line ("anything touching public money" would freeze the entire budget). For those protected classes, no procedural shortcut, guillotine, or emergency framing may skip full committee and deliberative scrutiny. Everything else — including the ordinary annual budget and emergency spending — runs a fast-track deliberative mode on fixed statutory deadlines, so scrutiny is guaranteed where it matters without a throughput collapse (reconciling with the §I.6 emergency-spending path). Collapsing scrutiny of the protected classes is a clear early marker of backsliding and is tracked on the outcomes ledger (§VI.7).

IX.3 The executive and the head of state

• The executive function is the Expert Execution Layer (Part IV) delivering the mandate, coordinated by a head of government accountable to the legislature, term-limited, and removable. The head of government sets priorities and directs coordination across the expert domains, and answers to the legislature for delivery, but cannot override a Board's technical judgement within its scope by fiat, nor direct it outside the published mandate (an unlawful direction, §IV.6). Where the two disagree on whether a question is one of priorities (political) or method (technical), the value-flagging route (§V.4) settles it. This keeps "ends from the people, means from the competent" intact at the very top of the executive.
• The executive executes; it does not set values or make law, those come from the people and the legislature (§0.5).
• Head of state: holds no policy, veto, command, or discretionary power (Axiom 1: no hereditary or perpetual governing power). The office is ceremonial and unifying, carrying only the non-discretionary guardian-of-the-constitution duties of §XVIII D2 — convening the legislature, referring a suspected breach to the Court, formally affirming due-process outcomes — which are ministerial duties triggered by defined events, never a discretion to withhold (a power to convene is not a power to delay). Whether it is a reformed strictly-ceremonial monarchy or an elected non-executive presidency is a value choice for the people (§V.4), recommended as elected (§XVIII D2); either way governing authority never attaches to it.

IX.4 Judicial independence and review

• The Rights & Constitutional Court enforces the Charter (§I.8): it may strike down any act, legislation, expert decision, emergency measure, that breaches the Charter.
• Independence, structurally entrenched. The Court's existence, core Charter-enforcement jurisdiction, fixed size (an odd number, set in the Charter), staggered single-term schedule (long, non-renewable terms so no single mandate reshapes the bench), secured and uncuttable funding, the composition of the independent authority that appoints it, and the exhaustively-listed grounds for removal (adjudicated by an independent judicial tribunal, never by the political branches) are all super-entrenched (§I.9.2): they cannot be altered by ordinary legislation or a single amendment. This closes the court-packing / court-restructuring route — expanding, shrinking, or re-staffing the Court — by which most modern "lawful" captures actually run. Altering the Court's composition is barred outright while its independence is being litigated.
• Limits on the Court (Axiom 4): it interprets, it does not legislate or amend; it cannot make policy. On ordinary Charter questions it is final and can be overruled only by the people via §I.9. But it is not the judge of the boundary of its own power: the contents of the super-entrenched core, and whether an amendment illegitimately touched it, are settled by the constitutional-convention procedure (§I.9.4a) — a convened joint sitting, not a standing body — not the Court alone, and the Court must publish and keep current the enumerated core contents (§I.8.4). It is monitored for integrity by the Integrity bodies (§VI.1).
• An independent ordinary justice system handles civil and criminal matters, with the usual due-process guarantees (§I.3).

IX.5 The web of mutual checks

Every branch is checked by at least two others, closing Axiom 4 system-wide:

There is no institutional apex. On ordinary Charter questions the Court is final, but it is not the judge of the boundary of its own power: the contents of the super-entrenched core, and whether an amendment breached it, are settled by the constitutional-convention procedure (§I.9.4a) — a convened joint sitting of the Sortition Chamber (a majority of members), the Assembly, and a Court panel, ending in a referendum — not by the Court alone. The chain of checks therefore terminates only in the people (through that procedure and the amendment routes) and the super-entrenched core, so that no single institution, the Court included, sits above the rest.

IX.6 Subsidiarity and local government

Power defaults to the most local effective level (Axiom 12):

• Local and regional government mirrors the national model at appropriate scale, elected representation (STV+), expert delivery within mandate, and integrity oversight.
• The centre must justify centralisation, the burden is on the national level to show why a decision cannot be made locally, not the reverse. This is enforced as a procedural review — did the centre publish a genuine reasoned justification against the defined criteria? — not a substantive re-decision by the Court (which would breach the interpret-not-legislate limit, §IX.4); the substantive subsidiarity judgement sits with the deliberative bodies, the Court checking only the adequacy of the reasoning.
• Fiscal subsidiarity, local bodies have the resources and revenue powers to match their responsibilities, with transparency and anti-corruption applying identically (Part VI).

IX.7 The free press and information ecosystem

A free press is a check the constitution cannot fully formalise but must protect:

• Independence guaranteed, no state control or capture of the press; plurality of ownership protected against monopoly (a capture vector) — enforced by an independent media authority (not the executive) against published ownership-concentration thresholds, with structural firewalls (funding, editorial independence, labelling) separating the state Public Information function from government, so "complements, never crowds out" is mechanised, not merely asserted.
• Protection for journalists and sources, including whistleblowers (§VI.9).
• Disinformation is met with transparency, provenance, and verifiable truth, never censorship (§II.5, §III.8, §VIII), because a state that decides what may be said has already failed §I.3.
• The Public Information function (§II.5) complements, but never replaces or crowds out, an independent press.

IX.8 Intergenerational representation

Future citizens cannot vote, yet today's decisions bind them (Axiom 13). The model gives them a voice:

• A Future Generations mandate (a commissioner role and a standing duty within the Sortition Chamber) must assess and publicly report the long-run and intergenerational impact of major decisions (debt, environment, irreversible commitments), modelled on, and stronger than, Wales' Well-being of Future Generations approach.
• For irreversible or long-horizon decisions, a negative Future Generations assessment raises the bar: proceeding requires a published, reasoned supermajority override of the Assembly. To keep this a brake rather than an unelected veto, it is bounded: the triggering thresholds are numeric and published (debt beyond a stated share of GDP; genuinely irreversible environmental commitments; multi-decade lock-in above a stated size), not elastic — related or sequenced commitments are summed for the test (an anti-salami rule, so a lock-in cannot be sliced into sub-threshold tranches) and the GDP denominator is methodology-locked (§X.4). One brake, not four: the Future-Generations trigger and the IFA's intergenerational-fiscal red flag (§X.5) are the same §V.4 override on a single ladder, not parallel supermajority gates on one decision; and the escape clause (§X.4, the rule-deviation limb) and §X.9 crisis fiscal powers (the emergency-spend limb) are the two limbs of one crisis-fiscal régime with a defined precedence, not independently-invocable tracks (Axiom 14); whether a decision is in-scope is a published, appealable determination (to the Sortition Chamber on the classification, to the Court on process), so a commissioner cannot expand the brake by calling everything long-horizon, nor a government dodge it by calling everything reversible; the supermajority level is fixed in the Charter; and the brake is suspensory-plus-supermajority, never an absolute veto. An unelected assessor may force the present to consciously release the brake; it may never stop an elected supermajority (Axiom 13).
• The wellbeing objective (§0.2) and outcomes ledger (§VI.7) explicitly weight the long run, so short-termism is measured and visible.

IX.9 Failure modes and safeguards

Part IX ends. The core institutional design is complete: sovereign people, an honest mandate, competent bounded execution, integrity and self-review, crisis resilience, a verifiable tech substrate, and a closed web of checks. Next: the wider constitution, Part X, Public Finance & the Fiscal Constitution.