How these calculators are built
There is no shortage of calculators online. What we try to do differently is treat two things seriously: whether the arithmetic is actually right, and where the numbers came from. Four practices follow from that.
Figures carry their source
Any calculator that depends on a tax rate, a contribution rate or a statutory threshold states the date it was checked and what it was checked against, at the bottom of the page. The US Social Security wage base, the UK National Insurance bands and income tax thresholds, the pressure and torque conversion constants — each was taken from a primary source and gets updated when the rule changes. A calculator that doesn't say where its numbers came from gives you no way to tell whether you're looking at this year's rate or last year's.
Where we couldn't confirm something, we say so instead of filling the gap. The sleep calculator doesn't ask for your sex, because we could not find primary evidence that required sleep duration differs by sex.
When something turns out to be wrong it gets fixed and written down, not changed quietly. The random picker page said the winner was drawn with the browser's secure random numbers while the code was using ordinary Math.random — the code was changed to match what the page claimed, and the correction is listed with its date on the About page. The remaining three practices — how the arithmetic is tested, what happens when a widely repeated claim doesn't survive checking, and what we decline to build at all — continue below the list of calculators.
The arithmetic is tested automatically
The calculation logic in each page is kept separate from the display code, so thousands of assertions run against it without anyone checking by hand — boundary values that sit exactly on a threshold, round-trip conversions, and monotonicity (does a larger input actually produce a larger result?).
This has caught real errors. An hourly wage calculator was treating US FICA as a flat 7.65% with no wage cap and was missing the UK's 2% National Insurance band and the personal allowance taper — which produced a £545 monthly error at £80,000. A compound interest calculator was reporting simple interest as larger than compound. A currency converter held only 28 currencies, so it couldn't answer questions people were actually searching for. All three were found this way and fixed.
We check received wisdom instead of repeating it
Widely repeated claims sometimes don't survive contact with the source. When they don't, we say so on the page.
- "Wake up on a 90-minute cycle boundary and you'll feel refreshed" — that precision isn't achievable. Measured cycle length has a standard deviation of about 29 minutes, so across five cycles the uncertainty exceeds an hour.
- "Longer naps are better" — in a controlled comparison a 10-minute nap beat a 30-minute one, and the 30-minute nap produced a stretch of impaired performance right after waking.
- Ring and clothing size charts — national systems don't map onto one another cleanly, so we show the conversion formula rather than a single "your size is X" answer.
If it isn't our place, we don't build it
We don't build tools where a government service or a dedicated app already does the job more accurately, and we don't build anything that amounts to a medical or legal judgment. Tools that exist purely for fun carry an explicit note that they have no scientific basis, so they're never mistaken for the serious ones.
What you type stays in your browser
Every calculation finishes inside the page. Salary figures, savings balances, body measurements — none of it is transmitted to a server, because the code that does the arithmetic is already in the page you loaded.
Calculators that need outside data, such as exchange rates and metal prices, are the one exception, and even there the data flows one way: the values are fetched once a day when the page is built and baked into it. Nothing is requested while you use the tool, so your inputs still don't leave. The page shows the timestamp of the data it's using.
Korean and Japanese editions exist too
The same calculator often needs different rules in a different country, so the other editions aren't translations — they're rewritten against local standards. Recommended sleep duration is a clear example: Korea says 7–8 hours, the US 7 or more, Japan 6 or more, and each edition uses the guidance issued in that country. The Japanese gold calculator covers capital gains treatment under Japanese tax law; this English edition of the wage calculator covers US and UK tax bands.
See About for who runs this and how to get in touch, and each calculator page for the basis and check date behind its numbers.