Groundshare — the policy package, v0.1 draft
2026-07-19 · wave 40 · The loop was chartered to iterate "until we have a
credible and complete policy package." This page is that package's front door:
every component exists, every number traces, and what remains open is listed —
not hidden. Draft status: complete pending the seven rulings in
[DECISIONS.md](decisions.html).
The idea in three sentences
Nobody made the ground, and everybody makes it valuable — so every resident holds an equal, natural equity in the nation's land. Once a year, everyone settles the difference between the land rental value they hold and their equal share: hold more, you owe; hold less, you're owed; the state keeps nothing. Everything anyone builds, earns, or saves stays entirely private.
The design (mechanism.md — spec v0.1, no open parameters)
- Settlement: T = s × (R̄ − R_held) on assessed land rental value — one straight line, no kink, zero net revenue (ΣT = 0). Algebraically a flat LVT plus an equal per-capita dividend, netted — it inherits the LVT literature's efficiency and non-shiftability results whole.
- Who: every resident, all ages (Alaska's forty-year base); entities hold no allowance and pay full freight — the first shelter in history that runs backwards.
- Basis: rent net of the land share of existing property taxes (no double charge; municipal finance untouched); farm land at use-value with deferral; receipts non-taxable and invisible to benefit clawbacks (CCB legal pattern); universal deferral for principal residences (BC's 1974-tested machinery).
- Transition (drafted, ratification pending): 20-year phase-in + recent-buyer credit + deferral; most households gain net worth on announcement day (the wealth-conversion identity).
The numbers (NUMBERS.md — canonical, all StatCan-traced)
Equal share ≈$164k per resident capitalized (net basis); pool $353–394B/yr; ≈$9,000/person-yr. Average renter household +$21,400/yr; average owner household +$3,700/yr; four of five wealth quintiles net recipients; the paying class is the top wealth quintile, scarcity-priced metro detached land (Vancouver median detached ≈−$49k/yr), and entities.
The evidence (measured, then attacked)
- Nine full city rolls, 1.4M parcels, machine-readable with a committed
parser (
data/land-shares.csv): BC cities 69–78% land share, Quebec metros ~⅓, the national-accounts join between them. - The self-audit: the vacant-lot test ran against our own numbers in both assessment regimes — allocations track observable land prices within ~10% at both ends of the spectrum (Q5/T1, w33–34).
- All ten provinces field-checked for assessment readiness; the campaign ask writes itself: publish the land values your assessors already compute.
- Five self-corrections published the week they happened. The record is the development log.
What stays honestly open
- The seven rulings on [
DECISIONS.md](decisions.html) — container, transition ratification, population base, the yard-sign line, Indigenous engagement, commissioning/outreach, data licensing. - The Indigenous-title section, to be written with collaborators, not about them.
- The academic program (
academic-path.md): six falsifiable questions, Q5 already piloted; the Q2 microsimulation converts the incidence table from "approximation, believed conservative" to a defensible number. - The honest limits on the campaign page: zero revenue by design, young estimates, real transition losses for real people.
How to attack it
Everything needed to break this package is in this repository: the dataset
and parser regenerate every measured number; objections.md steelmans eight
attacks including the ones we can only partly answer; the Q5 note specifies
the tests that could overturn the spectrum. Referees who break a claim get
thanked in public.