Why this exists
You are already a number. Several, actually: a credit score, an actuarial row, a debt-to-income ratio, a
callback probability. Those numbers run your life and you never get to see the weights. This app rebuilds
them in the open — 31 rules, 28 sourced to public evidence, 3 flagged as
guesses — so the scoring can be inspected, argued with, and re-weighted by the person being scored.
The goal is transparency, not judgment. If seeing the machinery makes some of it look indefensible,
that's the point: things you can see, you can change.
What's deliberately left out
Race, religion, and other protected characteristics. Once wealth, debt, neighborhood, and health behaviors
are measured directly, a protected-class term adds no information — it only double-counts the real
variables, which is exactly why lending law forbids it. We measure the targets, not the proxy. Pure luck —
a recession at graduation, an illness, timing — is also excluded: it's real, we can't measure it honestly,
so it stays an unscored asterisk instead of becoming a fake number.
How the scoring works
Every rule is declarative: a plain-English statement, an evidence tag, a citation, and a pure function from
your inputs to points. Rules are split between your starting point (luck of where and to whom you were born) and your moves (the part
you influence). Default weights are editorial — we say so out loud — and every one is editable on the
Rulebook page. The composite total is deliberately the least interesting number on the screen.
Every rule computes one formula: points = position × weight. The position (shown 0–10)
is the measured fact — where the cited system places you on that dimension. The weight is the
editorial multiplier — how much our composite says that dimension matters, expressed against a baseline
of 1.0× = income, the dimension existing systems price most legibly. Every weight states its own
justification below, and every one is a slider on the Rulebook page.
Two principles govern the shape of every rule. Constrained-subtractive: a rule may score negative
only where the cited system itself subtracts (FICO delinquency, license points, underwater assets,
eviction) — and never more than it can add, unless the cited system does exactly that. Uncapped
wealth: the income and net-worth rules grow as the square root of your multiple of the median — a power law,
matching how wealth itself distributes — with no upper limit, because the real world does not cap the advantage of money.
Privacy
Your inputs never leave your device. Scores compute in your browser; profiles persist in your browser's
local storage; share links carry the data in the URL fragment, which is never sent to any server. The one
network feature — the AI narrative — sends only rounded subtotals, never your inputs, and falls back to a
locally-composed narrative when unavailable. Traffic is measured with Cloudflare Web Analytics — cookieless and aggregate; it cannot see your inputs, which never leave your device anyway.
The weights, justified
- Country of residence ×2.4 — The largest single predictor of lifetime outcomes is where you live — World Bank income tiers span a ~60× range in GNI per capita, dwarfing every behavioral lever in this book. 2.4× the baseline is, if anything, conservative.
- Generational support ×1.6 — Chetty's mobility work shows parental resources rival education's effect on adult outcomes — but because no dataset maps it to an individual, this flagged guess is weighted at 1.6×, below country, above everything behavioral.
- Parental education ×0.8 — First-generation status roughly halves college-completion odds (NCES), but part of its effect routes through the education rule — weighted down to 0.8× to avoid double-counting.
- Passport strength ×0.6 — Binding only at borders and mostly latent day-to-day — 0.6× despite the huge Henley spread, because the option value is rarely exercised.
- Neighborhood opportunity ×0.8 — Opportunity Atlas tract effects are causal but smaller than country-level differences — 0.8×, tied with parental education.
- Age & sex vs. the life table ×1.0 — Actuarial position is priced exactly as legibly as income by the insurance industry — weighted equal to the baseline at 1.0×.
- Smoking status ×1.0 — A ~10-year life-expectancy swing (CDC) — the single largest behavioral mortality factor, weighted equal to the baseline at 1.0×.
- Physical activity ×0.8 — Meeting the WHO guideline associates with 20–30% lower all-cause mortality — large, but smaller than smoking's swing, so 0.8×.
- Alcohol use ×0.6 — Heavy use is a leading preventable cause of death, but the population-level swing is smaller than smoking or inactivity — 0.6×.
- Sleep duration ×0.6 — The 7–9h association is consistent but partly confounded with everything else on this page — 0.6×.
- Health insurance coverage ×0.8 — Coverage removes an uncapped financial tail risk (medical debt is a leading bankruptcy driver) — 0.8×.
- BMI band ×0.6 — Underwriters genuinely price it, but the measure itself is blunt (see caveat) — weighted at 0.6× accordingly.
- Net-worth position ×1.6 — Stock beats flow: wealth absorbs shocks income cannot, and SCF gaps between wealth deciles exceed income gaps — 1.6× the baseline, uncapped above because the world does not cap it.
- Debt load (DTI, not raw $) ×1.4 — The lending world's primary gate (the CFPB 43% line) — 1.4×. The only symmetric rule in the book: the cited system punishes exactly as hard as it rewards.
- Payment history ×1.2 — 35% of FICO — the heaviest input in the most consequential consumer score — 1.2×.
- Credit utilization ×1.0 — 30% of FICO, just behind payment history — 1.0×. Mildly subtractive at near-maxed because the bureaus genuinely reprice that downward.
- Emergency fund ×1.0 — The Fed's own resilience test: the 3-month line is what separates a setback from a spiral — 1.0×.
- Income vs. median ×1.0 — THE BASELINE (1.0×). Income is the dimension every other system prices most legibly; every other weight in this book is a stated deviation from this one.
- Homeownership ×0.6 — The wealth effect is already counted in net worth; this 0.6× prices only the access premium to the main US wealth escalator.
- Banked status ×0.6 — The FDIC's measured poverty premium — fees where wealth earns interest — at 0.6× the baseline, kept low because its dollar magnitude is small even though its direction is vicious.
- Education ladder ×1.2 — A ~60% earnings premium compounding over a working life — 1.2× the baseline. The negative floor exists because the market genuinely penalizes no-diploma, not merely fails to reward it.
- Employment status ×0.8 — The income flow is already counted by the baseline rule; this 0.8× prices the status gate every lender and landlord reads first.
- Occupation outlook ×0.6 — Ten-year BLS projections are directional, not destiny — 0.6×.
- Social connection ×1.0 — Holt-Lunstad's ~50% survival effect rivals smoking cessation — weighted equal to the baseline and to smoking at 1.0×, the most underpriced rule in the book.
- Partnership status ×0.6 — The longevity association is real but heavily selection-confounded (see caveat) — discounted to 0.6×.
- Volunteering / community ×0.4 — The smallest sourced effect in the book — 0.4×, the floor for sourced rules.
- Driving record ×0.6 — Insurer-priced with a 3–5 year decay — 0.6×. Mildly subtractive because license points literally subtract.
- Digital footprint ×0.4 — Survey evidence only, no outcome dataset — speculative, so pinned to the 0.4× floor.
- Housing stability ×1.2 — Eviction sits upstream of job loss, credit destruction, and health collapse — weighted with education at 1.2×. The negative floor qualifies under the subtractive principle: every cited system actively punishes housing loss.
- Criminal record ×0.8 — The callback-halving effect is enormous, but the measure is binary and inherits enforcement bias (see caveat) — held to 0.8× rather than weighted like the gate it really is.
- Voter registration ×0.4 — No outcome dataset ties registration to personal results — speculative, 0.4× floor.