Make the transfer
of authority visible.

Let A be the fraction of a fixed budget already delegated, and Δ the proposed additional fraction of that same total. Both lie between 0 and 1. Let each condition F, P, I, L, C be 1 only when evidenced for this expansion, and 0 otherwise. Unknown or disputed evidence is not treated as a pass.

G = F × P × I × L × C
A_next = A + G × min(Δ, 1 − A)

If every condition passes, the proposed addition is counted, capped at the remaining budget. Otherwise A_next = A. The values express a proposed decision rule, not a scientific score or a probability of democratic transition.

The formula records an actual authorised step. Repeating it does not create new authority by itself. Each new step requires a new assessment and authorisation. If authority is later lawfully changed or revoked, record that change explicitly; progress is not assumed to be irreversible.

For a fixed budget B, let d be the fraction of the delegated allocation whose agreed decisions have been implemented by the assessment date.

Delegated value = B × A
Implemented value = B × A × d

Define completion and evidence before measuring. Track each allocation once; split partially completed work into agreed milestones rather than assigning an unexplained score. Pending, overdue, disputed, and lawfully revised decisions should be reported separately.

This measures the budget value governed through implemented decisions. It does not measure money actually spent, outcomes achieved, or the legitimacy of those decisions. A high delivery rate cannot compensate for violations of the foundation.

If d is unknown, report it as unknown. If evidence only establishes a range d_low ≤ d ≤ d_high, the implemented value lies between BA d_low and BA d_high. Do not replace missing evidence with zero or a confident estimate.

For a simple yes/no example, let N be the eligible population, n the number of unique eligible people casting a valid yes/no ballot, and y the yes ballots. Require 0 ≤ y ≤ n ≤ N and N > 0.

Valid-ballot participation = n / N
Support among valid ballots = y / n
Recorded support among all eligible people = y / N

With 10,000 eligible people, 4,000 valid ballots, and 2,400 yes votes: participation is 40%, support among valid ballots is 60%, and recorded yes votes cover 24% of the eligible population. Non-voters’ preferences are unknown. Neither 60% nor 24% estimates society-wide support from this self-selected vote.

With no valid ballots, support among ballots is undefined. Report abstentions and invalid ballots separately. More complex voting systems need their own published rules. Any majority threshold, turnout rule, delegation system, electorate, and rights review must be decided before the vote. AI never adds a participant or a ballot.

Even a valid majority does not create legal authority outside the process’s mandate. Conversely, a legitimate vote can be binding under its agreed rules without every eligible person voting.

For distinct, non-overlapping budget allocations B_j, each with delegated fraction a_j and implementation fraction d_j:

A = Σ(B_j a_j) / ΣB_j
E = Σ(B_j a_j d_j) / ΣB_j

A is the delegated share of the combined budget. E is the implemented share. The denominator must be positive. Keep the budget scope, period, currency basis, and boundaries fixed for comparisons; document any change. Do not count the same allocation again at municipal, regional, and national levels.

For example, delegating all of one €100,000 allocation and none of a separate €900,000 allocation gives 10% overall, not 50%. The unweighted average would hide their different sizes.

Some powers—courts, rights, appointments, constitutional amendments—cannot sensibly be weighted in euros. Report their mandates separately. Any broader weighting scheme is a political choice that needs justification; it is not a universal measure of democracy.

For each responsibility, identify who controls its legal authority, budget, staff, infrastructure, and execution. Document the enforceable commitments that connect these actors to the agreed decision. Identify unresolved dependencies and who could prevent implementation.

This is the basis of conditions L and C. Historical delivery is evidence, but cannot prove future compliance. A government’s acceptance of proposals it likes is insufficient evidence that it will honour a delegated result it opposes.

The model does not turn military strength, wealth, or platform membership into extra votes. A legitimate rule for jurisdictions, representation, and power-sharing still needs to be agreed by the affected public and institutions. This is an unresolved design question, not a missing number a formula can invent.

Before starting, publish the scope, electorate, decision rule, baseline, measures, review dates, and expansion criteria. Compare the process with the existing way of making similar decisions where a fair comparison is possible.

Measure access and exclusion, understanding of alternatives, correction of AI errors, cost and time, privacy and security incidents, institutional responses, and implementation. Let independent evaluators and participants challenge the findings.

Success permits consideration of a defined next step. It does not require expansion, establish a universal rollout speed, or demonstrate that another country will follow the same path. No real-world values or success probabilities have yet been established for this model.