CalculatorQueenCalculate Clearly. Decide Confidently.

From source to result, with a visible chain of evidence

A polished interface is not proof of a correct answer. Our release system treats formula definition and reference testing as product requirements, while clearly distinguishing engineering-tested site-owner authorization from independent review.

The six-stage release gate

  1. Define the intended decisionConfirm what a person is trying to solve, which modes are required, and which superficially similar searches belong on one canonical page.
  2. Approve the formula authorityRecord the primary source, jurisdiction or scope, effective date, formula version, inputs, units, defaults, valid range, precision, and rounding.
  3. Implement a pure calculationThe shared engine validates typed inputs and runs deterministic formula modules. Page copy and layout cannot silently alter calculation logic.
  4. Run independent vectorsNormal, boundary, invalid, reverse, round-trip, and authoritative examples run in automated tests. Expected values are reviewed separately from code.
  5. Measure competitor parityNamed competitors, captured inputs, settings, dates, and expected outputs form explicit parity profiles. Differences are documented instead of hidden.
  6. Release and re-reviewOnly a current, explicitly production-authorized version enters the sitemap. Formula or source changes invalidate its content-addressed authorization, rerun dependent tests, and require a new publication decision.

What “verified” means

“Engineering tested” means the current formula version has named sources, validation rules, and passing mandatory reference vectors. “Independently reviewed” is used only when the required reviewer records also exist. A site-owner production authorization is not independent expert, QA, editorial, or legal approval. No status means a tool is universally suitable for every jurisdiction, professional context, or unusual input.

Competitor agreement is not used as a substitute for a primary source. Two competitors may disagree because of hidden defaults, differing dates, jurisdiction rules, or rounding. Parity is therefore recorded per competitor profile rather than claimed globally.

Numerical behavior

Units and normalization

Inputs are normalized before calculation and results retain their unit. A conversion factor is versioned alongside the formula when its definition can change; exact mathematical conversions are explicitly marked exact.

Precision and rounding

Intermediate values are not rounded merely for display. Each definition names its output precision and rounding mode. Display values and machine values are kept separate so a formatted answer does not contaminate later steps.

Invalid and ambiguous inputs

The engine rejects missing, non-finite, out-of-range, or logically impossible inputs with stable error codes. It does not silently repair a value or invent a default that the specification has not approved.

High-stakes calculations

Health, tax, lending, investment, legal-threshold, and other high-impact tools show source effective dates, limitations, and re-review dates. Until a qualified domain review is recorded, their results remain engineering-tested estimates rather than independently reviewed professional guidance and must be verified before consequential use.