Hidden Numbers Entry #0004 Classified Declassified

Benford’s Law: why 1 leads the ledger

In real-world data, about 30% of numbers start with the digit 1 and only 4.6% start with 9. Fraud investigators use that gap as a lie detector.

Play reel — loads from YouTube
Reel — entry #0004 1:19

Intuition test — answer before you read on

Roughly what share of leading digits in natural datasets is the digit 1?

In real-world data, about 30% of numbers start with the digit 1 and only 4.6% start with 9. Fraud investigators use that gap as a lie detector.

What everyone sees

A column of figures — populations, invoices, river lengths, stock prices — that should surely start with each digit about equally often.

What is actually happening

Data that spans several orders of magnitude is roughly uniform on a logarithmic scale, and on that scale the space between 1 and 2 is far wider than the space between 9 and 10. The leading digit therefore follows log10(1 + 1/d): 30.1% for 1, 17.6% for 2, down to 4.6% for 9. Humans inventing numbers spread their first digits far too evenly, so tax authorities and auditors run Benford tests over expense claims and election returns to find the fabricated batches.

Why it stays hidden

Because it only appears in aggregate. No single invoice looks wrong. The signature lives in the shape of ten thousand of them at once, which is exactly the view a forger never has.

Genuine numbers are unevenly distributed. Invented numbers are suspiciously even.

Genuine numbers are unevenly distributed. Invented numbers are suspiciously even.

The hidden part — entry #0004

Collect this card

Genuine numbers are unevenly distributed. Invented numbers are suspiciously even.

0 / 10,000 collected

Circulate this file