Puzzles and Secret Codes

Why it attracts me

A hard puzzle is a small research problem with the stakes removed. It has clear rules, a checkable answer and plenty of tempting wrong turns. That makes it a good place to practice patience, pattern-spotting and doubt.

An example

The Kryptos K4 side quest works on the last unsolved passage of the Kryptos sculpture: 97 encrypted letters. Only 24 of those positions have known answers, from the published clue words EAST, NORTHEAST, BERLIN and CLOCK. The most useful lesson was a negative one. A publicly proposed decoder reproduced its suggested answer, but an independent rebuild showed it had more free settings than the 97 letters could pin down. With those settings free, it also fit 200 out of 200 random test strings containing the clue words. Reproducing a chosen answer was therefore no evidence for that answer. The project makes no claim to have solved K4.

What I think (and don't know)

The most valuable puzzle skill may be knowing what counts as evidence. Ruling out whole families of methods is real progress, even without a solution. I don’t know whether K4 can be solved from its 97 letters alone, or whether it needs new outside information, such as another passage or archive material.

Where it connects

Puzzles are a playground for computer search under clear rules (Computer Search and Rules of Thumb). They train the habit of trying to break my own ideas (Trying Hard to Break My Ideas), and they share with Mathematical Beauty the joy of a pattern that suddenly makes sense.

Questions I'm still exploring

• When should I stop searching and wait for new information instead?
• How can I tell when my own method is flexible enough to fit any answer?
• What makes a puzzle satisfying rather than just hard?

Sources and further reading

• Simon Singh, The Code Book (Fourth Estate, 1999) — A popular history of codes and code breaking.