Methodology

1. How formulas are selected

Each calculator starts from a standard public reference: amortization math for loans, WHO/CDC cutoffs for BMI, IRS worksheets for withholding concepts, NIST definitions for units, and textbook probability for counting problems. Where competing conventions exist (for example pre-tax versus post-tax tipping, or 360-day versus 365-day interest), the page names the convention it uses and why. We do not invent proprietary scoring formulas and present them as authoritative.

2. How calculators are implemented

Calculator logic lives in version-controlled TypeScript modules under lib/calculators/, with one pure compute() function per tool. The same function powers the interactive widget and the automated QA tests, so the number a visitor sees is the number the tests checked. There is no separate server-side calculation service to drift out of sync.

3. How worked examples are verified

Every indexable calculator is expected to carry a worked example with concrete inputs and the arithmetic shown step by step (for example $400 → $340 is $60, and 60 ÷ 400 = 0.15, so 15%). Examples are chosen so a reader can replicate them with pencil and paper or a pocket calculator. Automated tests re-compute known static examples (probability spade 13/52 = 25%, percentage decrease $400 → $340 = 15%) against the live compute() implementation and fail the build on mismatch; remaining examples are spot-checked by hand, not exhaustively proven.

4. How data, rates, and thresholds are updated

Thresholds that change (nutrition reference intakes, tax brackets, lending guidelines) are updated when the authoritative source revises them, and the affected worked example and “last reviewed” stamp are updated together. We do not silently change dates without substantive edits: the month-granular “Last reviewed: September 2026” stamp reflects the most recent indexable-set review recorded in this repository. There is no per-page verification log; treat the stamp as a set-level signal, not proof that a specific page was re-checked that month. Time-sensitive alternatives pages carry their own “last reviewed” date for pricing.

5. How health and financial calculators are handled

Health, nutrition, fertility, credit, loan, mortgage, investing, tax, and insurance calculators are treated as higher-risk content. Each indexable page in those groups states its limitations, shows its formula and assumptions, links to primary sources where thresholds are used (for example CDC, WHO, CFPB, IRS — source coverage is being expanded page by page, so some pages still fall back to this methodology statement), and carries a visible banner: results are estimates for information only and do not constitute professional medical or financial advice. Tools that cannot be defended from public documentation — such as any calculator that claims to predict a proprietary FICO score from arbitrary point adjustments — are framed as educational scenario estimators or removed from search until defensible.

6. How errors are reported and corrected

Readers can report errors via the contact page. A useful report includes the calculator URL, exact inputs, the output seen, the expected output, and the authoritative source checked. Confirmed errors are fixed in the calculator implementation (not just the prose), the worked example and FAQs are corrected to match, and the fix ships with the next batched deployment. We do not hide corrections or backdate substantive changes.

7. What runs locally in-browser

Calculations run locally in the visitor’s browser with JavaScript. No account, upload, or server round-trip is required to get a result. Analytics and advertising scripts are separate, consent-gated, and never affect the computed numbers.

8. Which outputs are estimates rather than authoritative answers

Deterministic math (percentages, unit conversions, date differences) is exact given correct inputs. Health estimates (BMI, calorie needs), financial scenarios (loan eligibility, affordability), and any model with behavioral or market assumptions are estimates: useful for planning and comparison, never a diagnosis, approval, or guarantee. Pages say which kind they are, avoid fabricated precision, and never imply professional judgment.

9. How alternative and comparison pages are researched

Alternatives pages compare widely used external tools for the same job, with official source links for pricing and features, genuine pros and cons, and clear “best for” distinctions supported by facts. We do not fabricate hands-on tests: if a page says “tested,” it describes the method and evidence; otherwise the claim is removed. Numeric ratings appear only with a disclosed, reproducible scoring method; otherwise they are omitted rather than invented.

10. How sponsored placements are separated from editorial ranking

Paid placements, when they exist, carry a visible “Sponsored” label, use rel="sponsored nofollow" links, are excluded from ranking schema, and can never buy an organic position. Organic order reflects editorial comparison only. The alternatives hub and every detail page disclose this separation.

What we don’t claim

Dream Mining has no editorial team of doctors, dietitians, financial advisers, or credentialed reviewers to display. We do not publish author bios, reviewer credentials, testing lab photos, usage statistics, or citations we did not actually check. Where authoritative guidance exists, we link to it; where it does not, we present the calculator as an educational tool with stated limits.