I have an idea for a simple cipher that builds on old techniques that I thought I would share with this group. It has probably been done but I have not been able to find a reference to it.

Given two individuals who wish to encrypt their correspondence, brevity and a one time pad is pretty bullet proof. However there are weaknesses that can be exploited. Using a key word that can be deduced as well as common letter pairings, complexity and mistakes in the ciphering or deciphering and finally the pad or letter grid itself.

My idea is to remove as many of these failings as possible and make a quick and easy process. Starting with the letter grid or square. Blocks of jumbled alphabets have been used, paragraphs from books but what ever the the source the letters must be put into a table format, at least 2 copies, that are correct. Unless one just uses a book cipher then you just need two identical books. I do crossword puzzles and realized every day there is a solution to yesterday's puzzle in the paper. A ready made pad accurate across the distribution of the paper. All that is needed is the index line and column. There are holes in the puzzle, black zones, and those could reveal the structure of the pad given enough encrypted words. But could those be embraced. That took me the actual enciphering technique. Instead of complex routines what if I just use a frame. Knowing that I need an E, I pick one in the grid and then use a piece of paper to find the two index letters, one for the column one for the row.

To summarize,

Look in the paper on a specific day and retrieve the crossword solution. Clear text, "buy in my name" Find a letter b in the grid or table Put the corner of the piece of paper on the grid so that it points diagonally to the b Follow the paper edges to the two indexes and read the characters that identify the square with b in it. Repeat. Spaces could be indexed to blacked out squares.

To decipher,

Get the crossword solution for the proper date and time Put the index row and column in place Use a piece of paper to read the clear text from the grid

There are obvious problems, beyond my general unfamiliarity with the process, but my question would be a historic pointer to this process if there is one. Opinion on the security of the process, breaking the cipher. And finally given the process how many times could one use that grid? Obviously the New York Times crossword would be a far better source than a kids crossword .. but.

  • $\begingroup$ Why have you posted this text here? What is your question? $\endgroup$
    – mentallurg
    Jan 31 at 22:07
  • 1
    $\begingroup$ Also "Using a key word that can be deduced as well as common letter pairings" means you have a fundamental misunderstanding of what a one-time pad is. There's no such thing as a key word in a one-time pad, letter pairings don't matter, etc. A one time pad has a key exactly as long as the message. The key is completely random. The key is added to the message character by character modulo the number of characters in the alphabet (note: not the English letters, but any characters allowed. Could be a few billion if you allow all of Unicode). Anything else and it's not a one-time pad. $\endgroup$ Jan 31 at 22:44

What you seem to be describing is similar to a single-letter book cipher, but using the solution to a crossword puzzle instead of a book as the reference source.

While this might have some merits as an ad hoc hand cipher — after all, book ciphers have also been successfully used as such in the past — it also shares most of the drawbacks of book ciphers. In particular:

  • It has no secret key, other than the choice of which crossword puzzles to use. Anyone who can guess that, and has access to those puzzles, can decrypt the messages. This is a violation of Kerckhoff's principle.

  • It cannot encrypt letters that don't occur in the reference crossword. No Q or Z in today's puzzle? I guess you'd better not use those letters in your message.

  • Its security is strongly dependent on the encryptor's skill at choosing letters uniformly at random from the reference source — something that people tend to be bad at, especially if they're tired or in a hurry or just feeling lazy. In particular, if the encryptor keeps reusing the same coordinates for repeated letters, or tends to choose nearby coordinates for successive plaintext letters that happen to occur close to each other in the puzzle (e.g. in the same word!), or even just generally favors a particular corner of the puzzle, the security of the scheme may be substantially weakened.

Some of those problems could be remedied; for example, superencrypting the ciphertext with a simple substitution cipher with a secret key (or, perhaps more simply, just using the secret key to somehow scramble the row and column labels) could make this scheme significantly harder to crack. But other weaknesses, like the reliance on operator diligence, are harder to patch.


Not the answer you're looking for? Browse other questions tagged or ask your own question.