ROT13 / Caesar Cipher

Suetonius records that Caesar protected his correspondence by moving each letter three places along the alphabet. Two thousand years later the same trick survives as ROT13, the thirteen-place shift that turns up in spoiler tags, README jokes and the opening round of nearly every capture-the-flag competition. This tool applies the whole rotation family — Caesar at any shift, ROT13, ROT5, ROT18, ROT47, Atbash and Vigenère — and, when the key is missing, works it out for you.

That last part is the point. A Caesar has only twenty-six possibilities, so the honest way to attack one is to try them all, and the honest way to present the result is to say which one reads like English. Paste a ciphertext and every shift appears at once, each scored by how far its letter distribution sits from the distribution of written English; the closest match goes to the top with a star on it, and one click sets the direction to Decode and the shift to that number. What used to be twenty-five drags of a slider and a squint is now a table you read.

When nothing in that table reads, the Analyse tab explains why rather than leaving you to keep guessing. The index of coincidence measures how often two letters picked out of the text happen to match — around 0.067 for English, around 0.038 for random letters — and crucially it is untouched by any single-alphabet substitution, so a Caesar of English has exactly the index of the English. A low value therefore means more than one alphabet is in play and no shift will ever recover it, which is the signal to reach for Vigenère instead of trying harder.

A Caesar shift has a key space of twenty-six, of which one is the identity. That is not a weakness to be worked around; it is small enough that the complete search is the correct method, and any cleverer approach is wasted effort. What makes the search practical to read is scoring rather than eyeballing, and the score used here is Pearson's chi-squared: for each of the twenty-six letters, take how many times it appears in the candidate, subtract how many times English would predict, square the difference and divide by the prediction, then add up the results. A candidate close to English produces a small number; a candidate that is still gibberish produces a large one.

The reason this works at all is that a shift does not change the shape of a letter distribution, it only slides it sideways. English has a large spike at E, a smaller one at T, and a long tail through J, Q, X and Z. Enciphered at shift 7 that spike sits at L instead, but it is still one spike of the same height, so exactly one of the twenty-six rotations puts the spike back on E and every other rotation leaves it somewhere improbable.

Short texts are where this reasoning breaks, and the tool says so instead of pretending. Below twenty letters no row is marked as the answer: English predicts 0.015 occurrences of Z in a twenty-letter sample, so a single Z is enough to swamp the arithmetic and hand the crown to a wrong shift. The candidates are still listed, because on a short message a human reading five rows will beat the statistic — the tool just does not claim otherwise.

ROT13's relatives all answer the complaint that it leaves half the message alone. ROT5 rotates the ten digits by five, ROT18 does both at once, and ROT47 rotates every printable ASCII character — letters, digits and punctuation alike — by forty-seven, which is why it turns "Hello, World!" into "w6==@[ (@C=5P" and why it, like ROT13, is its own inverse. Here they are not separate code paths but names for two numbers: letters rotate by one amount and digits by another, and ROT18 is simply thirteen and five. Atbash, which reverses the alphabet so that A becomes Z, is older than all of them and needs no key at all.

Vigenère is the one that changes the game. Instead of one shift for the whole message, a repeating keyword supplies a different shift for each letter, so the same plaintext letter encrypts differently depending on where it falls. That flattens the letter distribution the chi-squared score depends on, which is exactly why the frequency panel and the index of coincidence are worth consulting before you spend time on shifts. This tool enciphers and deciphers Vigenère given the keyword; recovering an unknown keyword is a different piece of analysis and is not built in.

And then the family ends, deliberately. None of these ciphers is security, and the clearest demonstration of that is on the page itself: the panel that breaks a Caesar sits directly beneath the panel that creates one, and it finishes before you have let go of the mouse. They are the right tools for a spoiler tag, a puzzle hunt, a CTF warm-up or a lesson about why frequency analysis matters. For anything that genuinely needs to stay private, the answer is a modern cipher with a real key, not a shift.