Substitution cipher solver

Simple substitution and cryptograms

In a simple substitution cipher every letter stands for one other letter, consistently, all the way through. Newspaper cryptograms and cryptoquotes work this way, and so do the keyword cipher, Atbash and the Caesar shift. They're all special cases with a pattern in the key. Here the key can be any of 403 septillion arrangements of the alphabet. Far too many to try, but a solver doesn't need to try them all.

How the solver cracks it

It starts with a sensible guess: the commonest cipher letter is E, the next T, then A, and so on. Then it keeps swapping two letters of the key and scoring the result against English. A swap that makes the text read better stays; one that makes it worse is undone. That's called hill-climbing. A single climb can stall on a near-miss, so the solver starts again from fresh random keys and keeps the best answer.

The score comes from four-letter sequences counted across millions of letters of English fiction. Spaces aren't needed, which means it also cracks ciphertext that's been run together or written in blocks of five.

Helping it along

• Give it enough text. Around 80 letters solves most cryptograms outright. Shorter puzzles have several answers that look nearly as good.
• Lock in what you know. If you can see that XLI is THE, type T, H and E under X, L and I in the key, and solve again. Locked letters are shaded and the solver works around them.
• Watch for red letters. Two cipher letters pointing at the same plain letter is a contradiction, and it's highlighted.

If the solver can't find anything that reads, the cipher probably isn't a single substitution. Check the index of coincidence with frequency analysis. A low one points to Vigenère.