The Math That Lets You Whisper to Strangers

For most of human history, encryption had a chicken-and-egg problem. To send a secret message, sender and receiver needed to share a key — a password, a codebook, a one-time pad. To share the key safely, they needed a secure channel. To create a secure channel, they needed... a key.

Whitfield Diffie, Martin Hellman, and Ralph Merkle cracked the puzzle in a 1976 paper out of Stanford. Their idea: each person publishes one half of a key pair to the world and keeps the other half secret. Anyone can lock a message to you using your public half; only your private half can unlock it. No prior meeting required.

The machinery underneath leans on math problems that are easy in one direction and brutally hard in reverse. RSA, published a year later by Rivest, Shamir, and Adleman at MIT, uses the fact that multiplying two large primes is trivial but factoring the result back into those primes is — for a 2048-bit number — beyond any computer humans have built. Other schemes lean on discrete logarithms or elliptic curves.

British intelligence had quietly invented the same idea four years earlier. James Ellis and Clifford Cocks at GCHQ worked it out between 1969 and 1973, but the work was classified until 1997. The civilians got the credit and the patents because they published.

The entire web runs on this. Every padlock icon in a browser is a stranger-to-stranger key exchange happening in milliseconds — an idea that was literally inconceivable before the Carter administration.