Part III — The public power that serves it

22. Changing the Rules

The record holds a proposal that the Assembly placed on its docket and the electorate approved, and the proposal declares that it changes the article protecting the floor. The rules conclude no change in the law from those entries: the entries record a decision about a proposal, and they hold no replacement text that could be reviewed, published or put in force.

Changing a constitution requires agreement on both the decision and the text being changed. Approval of a proposal extends only to its own exact replacement, and a technically compatible replacement still needs that approval. The design therefore distinguishes political authority, certification of an exact candidate, publication and the version in force.

The political route requires a current record for the exact proposal: a full-membership Assembly result, a national referendum result, any required Regions Council and affected-region consent, separate completeness assurance, constitutional review, challenge, correction and a current certificate. The thresholds are two-thirds of the full Assembly membership and more affirmative than negative valid referendum votes, with no turnout quorum; a tie fails. Additional regional approvals apply where the federal settlement is directly affected.

Those are legal requirements over recorded results. Counting the electorate, authenticating each submission and finding any voter left out belong to the offices whose certificates the rules read. The resulting authority is bounded to the recorded proposal and current source; identifying, publishing and installing a replacement are separate stages.

Certification, publication and effect

The exact-change route connects the political result to the existing text, the proposed replacement and the transition between them. Independent compatibility review must concern the same candidate, and its review of effects must match that replacement, including any vocabulary change. Consent and compatibility are separate premises, and each is required.

A candidate needs consent that matches it exactly. A different candidate inserted into an upstream record matches nothing downstream and makes that record ambiguous. When two selections compete, both are marked as a conflict and neither candidate comes into force. A transition binds the base it was made for, so replayed against a later base it certifies nothing. Returning to earlier wording needs fresh authority and a new transition while preserving the intervening record.

Anyone may request independent review without the operator's permission or a valid candidate certificate. The request binds the office authorised to hear such disputes, whenever someone other than that office makes it. A current reviewed source defect withholds certification and publication, takes away the status of version in force even from a text that already has it, and obliges the operator to preserve the amendment history and correct the affected use; every person's floor remains owed throughout. Only a positively certified failure to respond moves the review duty to an independent alternate, and silence counts as neither consent nor a finding against the requester. Correcting a source leaves each individual judgment and remedy to its own process.

Whether publication occurred and an institution adopted the replacement belongs to the second book. The method's section Comparing and selecting an amendment distinguishes the formal evidence requirements from the separate comparison of exact text and version changes.

The protected core

The protected core is what every amendment must preserve, and the rules name its provisions in a closed list: universal standing, the floor, equal protection, due process, core liberties, the constraints protecting the commons, the categorical refusals on the use of force, the bans on sending anyone into persecution or torture and on collective expulsion, prompt independent review of detention, a remedy that works, and the capacity of the Assembly and the Constitutional Court to sit. Three protections for animals complete it: a credibly sentient animal's direct protected-subject status, the ban on severe avoidable suffering, and the ban on killing solely for one of the dispensable purposes the rules list. The food rule and the detailed standards are constitutional law that an amendment may change inside those three, and a food purpose authorises no avoidable suffering.

The economic constitution entrenches no single ownership form, provider mix, market share, profit, bailout, immunity or monopoly. Democratic law may change economic instruments above the floor, and every such change must preserve the core.

Exact-change certification requires positive review of the whole core, and a separate positive review of each of the three animal protections. A finding that a candidate removes an animal's direct status or weakens either ban must be independently reviewed and must compare the actual base, candidate and effects, tied to the proposal and transition; a text label or an unreviewed allegation is no such finding. The finding holds against any majority, emergency, profit, custom, claimed necessity or generic statement of compatibility. It blocks certification, authorised publication and selection as the version in force, and it obliges the responding office to refuse the candidate and secure an independent court challenge.

A reviewer's reading of the core refuses a candidate when it gives its grounds: which listed provision is at stake, the change in the candidate that reaches it, and the reasons. A contrary reading without those grounds withholds certification exactly as a missing positive review would, and the record stays open to a decision. Anyone may ask the independent final compatibility review for that decision, and the request obliges it to decide. The final reviewer stands apart from the record's attesters, its operator, the political bodies and whoever administered the result. Its finding that the candidate is compatible stands in for the reviewer's positive reading and answers a reasoned refusal, while every other field of the record still needs the reviewer who binds the candidate and the independent effect reviewer alike. An independently found animal-core breach still blocks certification whatever the final review finds. A reading of the core, however wide, creates no right; it decides only whether this candidate proceeds.

Refusal reaches this candidate alone. Discussion of it, challenge, everyone's existing human rights, animals' care and the lawful history of amendment all continue, and the breach finding makes their preservation a duty. The refusal is a conclusion over the recorded evidence of the candidate's effects, so the rules can miss a change concealed from that evidence.

What a proposal entry establishes

A name on a docket records a proposal. A declared target records what it says it changes. An approval entry records an outcome. None supplies the exact candidate, current political certificate, independent effect review, publication or selection as the version in force required above.

Whatever target an approved proposal declares, a protected article, an ordinary one or none at all, its entries confer no status as law. The register of protected names lists the floor article, prisoner personhood and the register itself, and it reads neither a candidate's text nor its effects. A copy of the source edited outside this route has the consequences of its edited text and none of the route's authority: a replacement comes into force only after its actual text and vocabulary have been reviewed.

People and proposal names

A proposal decision concerns the proposal and its candidate text. A personal finding needs its own legal grounds and process. Suppose a person's name is placed on the docket as a proposal, with a protected target declared and an approval recorded for it, and suppose the same person has a recorded contribution. Those entries produce no credibility finding against the person, and the recorded contribution identifies the act and confers no general recognition status. A proposal that declares the protected floor as its target likewise receives no personal credibility finding.

Authority to amend comes from the exact-change route alone, through certification, publication and selection as the version in force, and a bare proposal entry completes none of its stages.

What entrenchment requires

An unamendable core limits even a large majority and binds later participants who did not choose it. They bear its cost: a majority that wants to change what the core protects, however large and however lasting, has no lawful route to that change. Everything outside the core, from economic instruments to the food rule and the institutions themselves, stays open to the same demanding route, whose cost falls on reformers whose proposal has majority support but lacks two-thirds of the full Assembly.

The route holds only where each connection holds: the political result includes the entitled electorate, approval concerns the candidate actually reviewed, review protects the core, and the resulting text is the version in force. A register of target names answers none of those questions.

The integrity of the record those rules are written in also remains necessary. Anyone with access to the source file can change it outside the authorised process. The rules deny such an edit authority but leave it physically possible, and who made it and what it changes lie outside what they read.

The next chapters turn to what the design may do to a person, beginning with protection for someone who exposes authority.

Run it: the companion runs this chapter's cases on your own device, at dhilipsiva.dev/rights-nobody-has-to-earn/cases.

Argument: A fixed core behind an open route

The companion states these rules as Articles 23 and 30.

I put a small core beyond amendment because a majority able to withdraw standing, the floor, due process or a working remedy from its opponents could change who takes part in the next decision, and animals have no ballot with which to resist. Everything else stays amendable, because institutions are means and later generations must be able to correct them. The route asks for two-thirds of the full Assembly, so an ordinary governing majority cannot change the rules alone, and a national referendum, so representatives cannot change them without the voters. It has no turnout quorum, because a quorum lets opponents defeat a proposal by staying home. I review the exact text for its effects because approval attaches to words, while the core protects what the words do.

Constitutions have drawn this line before, in their text and by judicial implication, as Yaniv Roznai's comparative study of amendment limits sets out.^[1] Germany's Basic Law bars any amendment affecting the division into Länder, their participation in principle in legislation, or the principles of its Articles 1 and 20, among them human dignity, basic rights binding the state, and a democratic and social federal state under the rule of law.^[2] In Kesavananda Bharati India's Supreme Court held, by a narrow majority, that Parliament's power to amend does not include power to alter the Constitution's basic structure.^[3] In 2010 Colombia's Constitutional Court struck down a law calling a referendum on a second presidential re-election, for procedural defects and because the change would replace structural axes of the Constitution, among them alternation in office and the separation of powers.^[4] David Landau reports that the decision was complied with and "probably did prevent a significant erosion of democracy".^[5]

The strongest alternative is a fully revisable constitution. Jeremy Waldron argues that where democratic and judicial institutions work and people committed to rights disagree about them, representatives should settle the disagreement, and judges should not strike down their laws.^[6] Its best evidence comes from Landau himself, who finds the doctrine of unconstitutional amendments often hard to deploy against would-be autocrats, or easy for them to avoid.

My choice costs a later majority, however large and lasting, that wants to change what the core protects. It gives reviewers power over whether a reform crosses the line, which is why a reading refuses only when it names the provision, the change and its reasons, and an independent final review can answer it. Reformers with majority support can be blocked by more than a third of the Assembly. Elkins, Ginsburg and Melton find constitutions imperilled by crises, with design able to lower their risk of death;^[7] a core too rigid for its society may be replaced rather than amended.

I would reconsider if readings of the core repeatedly blocked reforms that kept its protections whole, if the route repeatedly stalled rights-preserving reforms with majority support, or if constitutions without a fixed core proved as reliable at keeping the losers of a political contest able to live, dissent and seek correction.