PHOTOLIBRARY OF BABEL

Dick Pountain/17 September 2007/11:24/Idealog 158

I first encountered Jorges Luis Borges story "The Library of Babel" as a student in the early 1960s, and it blew my mind then as it has blown so many since. It describes a mysterious library, organised as a presumably infinite collection of connected hexagonal rooms whose walls are filled with uniform volumes containing random text. Everything that could ever be written in every language is in there somewhere, but the wretched librarians who patrol the rooms are lucky to see two consecutive intelligible words in their whole lifetime, which may be short since jumping down the infinite stairwells is a popular way out... An allegory, even a satire, about society and communication, the story also raises profound philosophical and mathematical questions that could provide a lifetime of amusement. The library must contain all our literary works (and our advertising, and our telephone directories) including Borges' story itself, translated into every language on this planet and others. It must also contain all these with one letter changed, one word wrong, or two, or three - you get the picture. And by what criteria would you decide they were "wrong"?

It's awfully tempting at this point to write a column about the blogosphere as our modern Library of Babel, but with enormous self-control I'll resist and pursue a quite different tack. I've recently been reading up about the human visual system and the way the brain handles imagery, for a book I'm preparing, and one day I suddenly wondered what would be the visual equivalent of Borges' library: the Photogallery of Babel if you like? This turns to be as interesting as the original, but different in rather revealing ways.

First of all we have to decide on the uniform format for pictures in the gallery, and while I could choose a JPG file of some size, I prefer instead the biological option of "one human retina full". The retina is a finite array containing around 126,000,000 rod and cone cells, and for our gallery format their output would need to be sampled at some finite resolution - 32 bits would do since that approximates photorealistic colour. Each picture in the gallery would therefore be a bitmap of around 4 billion bits, and the gallery would contain every possible 4 gigabit image. I see the pictures as being hung on the white walls of a spiralling ramp, rather like the New York Guggenheim but taller. Not infinitely high though because there would actually be a finite number of pictures, around 10^1,000,000,000 to be exact - its ten-to-the-billion floors would stretch some way out of the solar system, probably the galaxy, which would make planning permission tricky. This gallery would contain everything that it's possible to see, or to imagine, which would be nice. The only problem is that every picture you could see in your lifetime would look like an almost identical reddish-bluish-greenish fog, in which you might occasionally discern some faintly recognisable object.

Borge's library wasn't actually infinite either because he specified that each book was 410 pages with 40 lines of 80 characters, from an alphabet of 25 symbols including space and punctuation: that's 25^1,312,000 unique books. However the question of what language those characters represent complicates things enormously: Kevin Kelly has suggested that every single book, however random looking, could be made intelligible if you discovered the correct language or encoding, which is true but of little consolation to the librarians. Is this true of pictures too? Actually it is - you could choose a random gallery picture, take a copy of the Mona Lisa, then use morphing software to turn the former into the latter. What you would have achieved is not a lot.

Both of these nightmarish institutions are really about meaning, the difference between information and noise, but the way that text is anchored to the real world is utterly different from the way a picture is. Words on a page don't refer directly to anything in the real world except the puff of air generated by your voicebox when you read them aloud. And when you utter, say, the sound "pan" an English person will think of a cooking utensil, whereas a Spaniard will think of a loaf of bread and French person a loud bang. The latter case is an example of onomatopoeia, pure imitation, which is the closest words ever come to real world objects. It's only once you string several sounds together that such folk can tell which language you're speaking.

Pictures however are biologically linked to the real world because they represent light reflected from external objects. The human visual system contains immensely powerful feature detection mechanisms that will try to see meaningful objects even in a random bitmap. Certainly you can shut your eyes and imagine objects that don't exist, but probably you'll be reusing, recombining and recolouring memories of light you once saw reflected from real objects. Abstract painters might bridle at such a handcuffing of vision to the real, but they shouldn't be upset - it's precisely our brains' automatic effort to intepret visual inputs that gives them the power to disconcert and surprise us. In the Photolibrary of Babel, pictures that look like Jackson Pollock's will be just as rare as Mona Lisas...