At the end of the previous post, I mentioned a suspicion that merely deciphering the correspondences between the glyphs on Fringe and the letters they represent might not be the whole of the glyph puzzle. Being new to the show, I didn’t know a whole lot about it previously, but I did decide to poke around a bit, and found a recent interview with the show’s producer, claiming that the solution to the glyph cipher would “speak to some of the larger controlling mythology behind the show.” Clearly, the words embedded in each episode so far don’t do anything of the sort, suggesting there’s something more to discover.

Possibly that just means that the words encoded in the first season’s episodes are provided mainly as raw data to enable folks to start cracking—using words like “Observer” and “Bishop” makes it easier to run a targeted attack and readily confirms you’ve got the correct key—while in future episodes, the code words will be more informative or provide the basis for a new puzzle.  But eyeing the glyph key, there’s some reason to think there’s a second puzzle already built in:

Now remember, the folks behind the show are making up these correspondences. They could’ve matched up glyphs and letters totally randomly.  Or they could’ve gone with a more straightforwardly linear pattern: Apples (with shifting dots) for A–D, Flowers for E–H, and so on.  Instead they went with this weird-looking hybrid (after last night’s show, we can apparently add “W” which is a mirror image of “V”), clearly somewhat patterned, but with interesting deviations. The first two (all plant!) lines open with “high dot” and “low dot” mirrored pairs of the leaf glyph. On the first line, though, the Apple glyph itself doesn’t mirror shift; only the dot moves. On the second line, the flower is mirrored, and the dot shifts in a different order. Line three looks like it’s getting ready to move from the 2-4 structure to a 2-2-2. And on the last line, the smoky face assigned to T disrupts the symmetrical pairs we’ve come to expect.

To me, that suggests potential information content. A completely patterned key—like a text rearranged with all its characters in alphabetical order—would be low-information. A completely random key assignment would technically be “high information,” but without some underlying structure or pattern, you’d have no basis for extracting any specific meaning from it. The reason “I’m going to go to the fridge and get a beer” is less surprising (and thus “lower information” in the technical sense) than “I’m going to go to the fridge rhinoceros teacup insipidly jellybean ” is that the very meaningfulness of the first set of terms lets you form reasonable expectations about the likely content of the rest of the phrase, based on the assumption that the sentence forms a coherent expression of an idea.  The sweet-spot mix of pattern and randomness is what hints at meaning and communication.

So how could this work?  Well, those of you who know anything about cryptography know that it’s not only possible to encipher an informative text with a certain cryptographic key, but to embed information in the key itself.  So, to pick a very simple example, suppose I create a basic monoalphabetic substitution cipher with the following key: “J=A, U=B, L=C, I=D, A=E, N=F, B=G, C=H, D=I…” Now I can encipher a text with that substitute alphabet, and given a long enough text,  a combination of frequency analysis and word-guessing should make it simple enough to crack the code and work backwards to the original key.  Anyone who steps back from the particular text and writes down that key in alphabetical order will not only learn the content of the particular enciphered text, but they’ll readily see that I used the word JULIAN as the basis of the substitution cipher. More complex forms of encryption, like the polyalphabetic Vigenère cipher, can also make use of (potentially much longer) keywords or phrases as the basis of their cryptographic keys.

That’s not to say that’s what’s been done here; I just offer it as an example of one way that a cryptographic key can embed information above and beyond that in any particular enciphered text. Someone who wanted to be especially devious could encipher a banal plaintext with the actually significant information embedded in the structure of the key. This key, say, might itself contain another enciphered word, where the letter sequence can be read as (LEAVES)(APPLES)(LEAVES)(FLOWERS)…—or LALF….—though that seems unlikely to be the method here, if only because a single long word doesn’t provide enough raw data for cryptanalysis, unless there’s some other means of guessing the key to the metacipher. If folks want to play with that idea for  while, I’d be interested to see if people discover anything.  I’m going to leave it for now, but perhaps when we’ve filled out the complete key, I’ll take another look.