4
$\begingroup$

As a font designer, I was thinking that it would be very easy to design a special font that could either jumble characters of a language (or of a code) or simply (with the features offered by the OTF format) overlap into a complex design that only the other user (the reader) would be able to unscramble... Is this something that exists and could be plausible?

EDIT: To better explain myself, I just created this image. (A) represents a whole sentence. Each circle is a word. (B) details two different kinds of "words" that could exist: wavy, straight lines... but always with 26 peaks (for example an alphabet with 26 letters). When one single letter from a font is typed, just one wave or peak would be drawn until the whole word is typed. (C) is another detail of overlapping of two circles, each one is a word. (D) explains how a word could be read: each circle having 26 peaks, the distance of the peaks from the center tells the reader what letter it is. I made the distance very visible in (D) but it could be extremely subtle / only visible when analysed by microscope/etc.

Such a code could exist on the front page of a newspaper and no one would notice. And it's all encodable in a "simple" OTF font...

font-code-design

$\endgroup$
  • $\begingroup$ I don't see how the overlap of fonts can be reversed by the user and not by an attacker; can you indicate how this is supposed to work? $\endgroup$ – Maarten - reinstate Monica Jul 21 '15 at 8:35
  • $\begingroup$ Simply because the attacker sees the final glyph and has no idea of how it was obtained (think about a star made with 50 overlapping shapes that could be variations on a triangle and not knowing what these shapes mean let alone what they're supposed to look like... yet you can see the clear part of the star that pokes out of the general blur). Or the overlapped shape could simply be flattened (PNG instead of TXT) and this way it would be impossible to know what each character is supposed to look like when not superimposed with others... $\endgroup$ – MicroMachine Jul 21 '15 at 8:50
  • 2
    $\begingroup$ @fabriced See Kerckhoffs's principle. $\endgroup$ – yyyyyyy Jul 21 '15 at 8:54
  • $\begingroup$ @yyyyyyy I'm sorry if I wasn't clear enough. I'll edit it. $\endgroup$ – MicroMachine Jul 21 '15 at 9:11
  • 2
    $\begingroup$ This also reminds me of steganography, especially if this is placed in an image. $\endgroup$ – the_lotus Jul 21 '15 at 13:37
6
$\begingroup$

Is this something that exists and could be plausible?

Yes, things like that already exist and have even been used by well-known serial killers!
(So much for creating a dramatic intro – lol)

Monoalphabetic Substitution Cipher

What you are referring to, could be categorized as a classical “pigpen cipher”; a monoalphabetic substitution cipher where graphical symbols are used. You might change the symbols and maybe even the way you encode things, but replacing text with symbols will always boild down to a simple, monoalphabetic substitution cipher.

Nowadays, you won´t find too many examples of it in action, but things looked different hundreds of years ago when a simple substitution cipher was able to baffle the masses. Yet, those days have long gone and suchs ciphers can now mostly be found in children's books on ciphers or “secret writing” gimicks packed in breakfast cereal boxes.

The rare exceptions to this rule are limited to things like (for example) the notes from the “Zodiac Killer”. But – all in all – the enigmatic charisma of “danger” and “death” provided more mystery to those notes than the usage of symbols… something which can be broken/deciphered rather easily, assuming enough data or information is available.

example ciphertext: note of Zodiac killer

Modern Cryptography and Kerckhoffs' Principles

For better understanding: in modern times (and on this site), we tend to handle cryptography according to well-vetted rules… among them some which go back to Kerckhoffs' principles, which defines that a cipher algorithm should be made public (aka: regarded to be known by “the enemy”).

One of the reasons why some notes by the Zodiac Killer haven´t been deciphered yet, lies exactly in that area: the algorithm was never published and the data provided by the Zodiac killer did not provide enough information to deduct the actual contents of the ciphertext with absolute confidence.

Now, please don´t get me wrong. As someone who has professionally designed some fonts for advertising and other purposes, I am always intrigued by the idea of putting some visuals in communication. But I can´t help to share my two cents to spare you from walking the wrong path unwillingly:

  • Would it make sense from a modern-day cryptographical point of view to create such fonts?

    No. Not if you actually expect them to provide any kind of real cryptographic security.

  • Would it be a nice hobby and find some fans nevertheless?

    Sure… as long as you keep in mind you´ll most probably be spending more time creating the fonts than others (cryptanalysts, as well as “ye average” puzzle freaks) will need to break the ciphertexts created with them.

    Let´s take your font idea and put it on the front page of a newspaper as you implied. That might work once. But the more you publish, the more data will be available and since we´re practically talking about a monoalphabetic substitution cipher, it won´t take too long for the first people to decipher your stuff by “guessing” the keys (with merely a sprinkle of Cryptanalysis).

    Especially, since you´re even limiting the security down to…

    26 peaks

    … instead of the 256 “peaks” (= 8 bits = 1 unsigned byte) we work with in modern crypto.

    As a simple rule of thumb: The smaller the alphabet, the easier the cryptanalysis!

I surely agree with you that one can hide a secret in the open. But as soon as someone knows there´s a secret to look for, security starts to fade towards zero. That´s where you need a secure cipher, and – whether we like it or not – fonts can not replace cryptographically secure algorithms.

Nota Bene

If you want to take a look around at the different fonts and symbols, use your preferred search engine to check on “cryptography symbols“ and co.

screenshot google image search

You´ll also discover your idea is already used and published in newspapers as (cryptographic) puzzles. There are even a few pages diving into such graphical ciphers. One of many: http://www.phantomgallery.64g.ru/shiphen/cod6en.htm

screenshot of example webpage

$\endgroup$
3
$\begingroup$

Substitution ciphers of all kinds are easily broken by frequency analysis. And since you are only replacing symbols from one alphabet with those from another, that fits the description of a substitution cipher.

Substitution ciphers exist for over 2000 years, and Al Kindi broke it in the 9th century already.

In order to create something secure today, a good understanding of current cryptography is mandatory.

$\endgroup$
3
$\begingroup$

No, specially designed fonts do not play a (major) part in modern cryptography.

With a specially designed font you can maybe achieve a single level of transposition and a single level of substitution. Transposition is the swapping of data, in this case characters. This is probably what you call "jumbling characters". Substitution is the replacing of data, in this case representations of characters.

A single level of transposition and substitution is not considered enough to create a modern cipher. You may be able to obfuscate the plaintext, but that's not considered cryptography. You may however be interested in the Playfair and Hill classical ciphers that perform polygraphic substitution. Those are reasonably strong ciphers which you may not want to try to break without a computer.

EDIT 1: You are also talking about stacking characters together if I'm not mistaken. It is possible to do this, but the scheme in itself should not provide security; it should be assumed that you need a key in addition to the scheme. Kerckhoffs's principle has already been mentioned by yyyyyyy. It is unclear if such a scheme can be created without more information, but it seems unlikely that it will handle keys with 128 bits of security.

EDIT 2: If you are just talking about hiding the message in another medium in such a way that it cannot be detected then you are talking about steganography. However in your example you make very strange patterns; many persons will wonder if the pattern encodes something. You could make tiny changes to normal letters instead. These changes would be much less pronounced.

Modern ciphers operate on bits, not characters. Operating on bits makes it possible to create a cipher that can operate on any well defined plaintext.

$\endgroup$
  • 3
    $\begingroup$ I'm not counting the single character substitution in "Commander Keen" as "modern cryptography" :) $\endgroup$ – Maarten - reinstate Monica Jul 21 '15 at 8:47
  • $\begingroup$ "However in your example you make very strange patterns; many persons will wonder if the pattern encodes something" = isn't that a thing as well? :) $\endgroup$ – MicroMachine Jul 24 '15 at 7:22
1
$\begingroup$

As the others answered, this of course is not a good idea. However, something similar which can be done properly is called "visual cryptography". This was introduced by Moni Naor and Adi Shamir at Eurocrypt 1994. This has its own Wikipedia entry, so I'll leave it for you to go there and start investigating.

$\endgroup$
  • $\begingroup$ Visual cryptography is not really an encryption method, but more like a secret sharing. Of course you can always consider this as a cipher if you consider one share to be the key and one share to be the ciphertext. $\endgroup$ – tylo Jul 23 '15 at 9:23
  • $\begingroup$ That is the idea. Indeed the initial motivation for visual crypto was exactly this. The receiver would take a transparency with a random pad printed on it and later the ciphertext could be posted publicly and only the receiver with the transparency can decrypt by laying the transparency over the ciphertext. $\endgroup$ – Yehuda Lindell Jul 23 '15 at 12:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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