in reply to Re: Playfair cipher
in thread Playfair cipher

Thanks for your comment ww. I wanted to comment on the merit of using a group of "dummy" characters (as they are called) rather than a specific character. The reason for using a dummy character in the first place is two-fold. (1) to prevent the creation of a new rule for handling characters in the same square and (2) to prevent simpler cryptoanalysis (english is full of double letters "mm oo ee (ww? :) )" . The addition of a single dummy character satisfies both of these requirements.

What your approach does is further confuse the relationship between duped chars and their encoding. This might make the statistical attack easier harder if the playfair cipher did not suffer from a much more significant weakness.

This weakness is the fact that E(AB) = reverse(E(BA)) (where E = the encrypting function). This means that words like DEceivED and REceivER and DEpartED will maintain a AB(..)+BA pattern. This is how most statistical attacks of a playfair cipher begin.

It is interesting to note that the fact that a double letter will never appear in the encrypted message actually weakens the cipher for the mere fact that a message with no double letters is extremely suggestive that the encryption is a playfair cipher.

Thanks for your response. Your idea can be easily adapted to helping the weakness of playfair which is that anytime a AB(..)+BA pattern is found replace it with a AXB(..)+BA (or a AB(..)+BXA to make the statistical attack weaker.

Update: Mistyped above. Text is struck out.