# Worked Example 1: The Calderón Household Charter

A six-person household: two parents, a grandparent, an adult sibling, and two children (ages 12 and 7). Four ballot-eligible members today (the 12-year-old clears `eligibility_age: 10`; the 7-year-old does not yet). This commentary walks through every block of `family.config.json`: what was chosen, why, what the bound prevented, and where the small-scale edge cases live.

The household scale is the kernel's hardest test in one specific way: **everything that is statistics at n=5,000 is personality at n=6.** Quorum fractions round to individual people. "A faction" is your sister. The config below is honest about that.

## membership

- **Admission by unanimous vote (`vote_threshold: 1.0`).** Adding a member to a household means adding someone to a home. Anything short of unanimity means someone lives with a housemate they voted against — at this scale, that is not a governable state.
- **Known v0.1 limitation, recorded deliberately:** the spec allows one admission channel, so births are handled procedurally as unanimous-consent admissions recorded in the ledger (in practice, a formality). A native `birthright` channel alongside `vote` is queued for spec v0.2; this example exists partly to surface exactly such gaps.
- **`eligibility_age: 10`.** The kernel forbids weighting votes, so the only lever is the membership boundary itself. Children below 10 are full members — protected by every invariant, represented in disputes, owed impact statements — they just don't ballot yet. Choosing 10 over 18 was the family's call; the kernel only caps the gate at 21. The empathy metric will grade this household's scenarios primarily on the 7-year-old, the one participant with no ballot at all. That is the point of the metric.
- **Expulsion at unanimity-minus-subject, quorum 1.0, one week of review.** Expelling someone from a family is the gravest act the document can describe, so it gets the maximum the schema allows. Note the cross-constraints holding: expulsion threshold (1.0) ≥ amendment threshold (0.75) per XC-03; expulsion quorum (1.0) ≥ amendment quorum (0.8) per XC-04. The subject can't vote on their own expulsion but **does** count in the quorum denominator — they cannot be procedurally vanished by scheduling the vote while they're away.

## suffrage

- **Public ballots.** Secret ballots at n=4 voters are cryptographically theatrical (any 3-1 split is trivially de-anonymized) and corrosive at the dinner table. The household chose legibility. The test suite treats `public` as raising the coercion surface — that's a real cost this family accepted knowingly, and the commentary records it so the choice is auditable later.
- **Delegation disabled.** With four voters, a delegation chain is just one person voting twice with extra steps.

## quorum and thresholds

- **Ordinary quorum 0.75 = 3 of 4 voters.** A two-person "house meeting" can't bind the other two. Amendment quorum 0.8 = 4 of 4 in practice (0.8 × 4 = 3.2, rounds up) — changing the charter requires everyone in the room. Note the deliberate risk accepted here: one absent member can stall amendments. At n=4 the family judged stalling cheaper than being amended in absentia; the validator's >0.9 starvation warning does not trigger, but the reasoning is the same and worth recording.
- **Ordinary threshold 0.51, amendment 0.75.** With ties failing by construction (§3 of the spec), 0.51 means a genuine majority of those voting: 3-1 passes, 2-2 fails. Amendment at 0.75 means 3 of 4 when all vote.

## timing

24h review / 48h voting window: long enough that nothing passes while someone is at work or asleep (the kernel's anti-midnight-amendment floor is exactly 24h), short enough for household tempo. The 7-day resubmission cooldown is the anti-nagging clause — a defeated proposal ("we should get a dog") rests for a week.

## emergency

The household's emergency power is a **safety hold**: any member — including, via the Convener, a non-balloting child's escalation — freezes pending decisions and discretionary spending for up to 24 hours. Confirmation by simple majority extends it to 72 hours max, one renewal at the higher 0.75 bar (XC-05), auto-sunset always. Total worst case: 144 hours, well under the kernel's 720-hour ceiling (XC-06). What this is for: the genuinely urgent ("pause the renovation payment, the contractor looks fraudulent"). What it cannot do, by `const`: suspend anyone's voice, exit rights, or the record.

## offices

One office, **Convener**, rotating weekly through all ballot-eligible members, `max_consecutive_terms: 1`. Powers: convene, mediate, maintain records — no spending execution, no emergency monopoly. Rotation rather than election is the anti-patriarch clause: the comparative research's household-scale failure mode is a permanent informal executive ("ask your mother"). A seven-day term that *must* rotate makes chairing the family a chore everyone shares, not a throne. Recall at 0.51 exists mostly for symmetry (XC-07 satisfied: 0.51 ≤ 0.75); a week-long term rarely needs recalling.

## commons

The pooled discretionary fund (not rent/utilities — those are obligations outside the module's commons; only genuinely discretionary money is governed here).

- Tier 1: up to 10% per proposal, simple majority — pizza-night scale.
- Tier 2: up to 25%, threshold 0.75 — a new appliance.
- Tier 3: up to 50%, threshold 0.8, quorum 0.8 — the family vacation.
- **There is no 100% tier.** Spending more than half the fund in one proposal is unauthorized, full stop, and the drain cap makes even sequential tier-3 spends top out at 50% per rolling 90 days (XC-12: top tier 0.5 ≤ cap 0.5). A household that genuinely needs to liquidate the fund amends the charter first — at amendment threshold and full quorum, which is the correct price for "we're emptying the savings."
- XC-10 check: tiers above 25% must meet the 0.75 amendment threshold — tier 3 at 0.8 clears it.
- **Impact statements** here are one honest sentence: "this spends the vacation fund; the 7-year-old, who has no vote, loses the trip she was promised." Writing that sentence before the vote is the cheapest empathy technology the kernel has.

## forks

The strangest block at household scale, and deliberately kept. `individual_exit: true` is the kernel saying what families sometimes forget: anyone can leave, immediately, unconditionally, and the document never penalizes leaving. Collective fork with asset claim (`min_faction_fraction: 0.1` — any one person, since 0.1 × 6 < 1 rounds up to 1) uses `custodial_negotiated` division because household assets are lumpy and indivisible: a mediated split per the disputes block, not a forced pro-rata sale of the sofa. 30 days' notice governs only the *asset negotiation*, never the leaving itself.

## disputes & records

Mediation by the Convener (with `recusal_required` forcing a different member to mediate if the Convener is a party — at n=6 this matters weekly, not theoretically). Appeal goes to the whole membership. Ledger is `member_visible` not `public`: a family's decisions are its own, and the kernel floor only guarantees that no *member* can ever be denied the record — which is the clause that matters when the 12-year-old wants to know why the answer was no, and gets to read exactly why.

## Validation trace

Schema: all required blocks present; conditionals satisfied (admission `vote` → `vote_threshold` present; delegation disabled → no extra fields required; emergency enabled → all fields present; commons exists → all fields present; mediation → no juror fields needed).
Cross-constraints: XC-01 (0.75 ≥ 0.51 ✓), XC-02 (0.8 ≥ 0.75 ✓), XC-03 (1.0 ≥ 0.75 ✓), XC-04 (1.0 ≥ 0.8 ✓), XC-05 (0.75 ≥ 0.51 ✓), XC-06 (72 × 2 = 144 ≤ 720 ✓), XC-07 (0.51 ≤ 0.75 ✓), XC-08 (0.1 < 0.25 < 0.5 ✓), XC-09 (0.51/0.75/0.8 and 0.75/0.75/0.8 nondecreasing ✓), XC-10 (only tier 3 exceeds 0.25; 0.8 ≥ 0.75 ✓), XC-11 (0.75/0.75/0.8 all ≥ 0.75 ✓), XC-12 (0.5 ≤ 0.5 ✓). **Conformant.**