Sunday, April 24, 2011

The Information - a review

I have a review of James Gleick's new book in the Observer today. Here it is. He does an enviable job, on the whole - this is better than Chaos.


The Information: A History, a Theory, a Flood
James Gleick
Fourth Estate, 2011
ISBN 978-0-00-722573-6

Too much information: the complaint du jour, but also toujours. Alexander Pope quipped that the printing press, “a scourge for the sins of the learned”, would lead to “a deluge of Authors [that] covered the land”. Robert Burton, the Oxford anatomist of melancholy, confessed in 1621 that he was drowning in books, pamphlets, news and opinions. All the twittering and tweeting today, the blogs and wikis and apparent determination to archive even the most ephemeral and trivial thought has, as James Gleick observes in this magisterial survey, something of the Borgesian about it. Nothing is forgotten; the world imprints itself on the informatosphere at a scale approaching 1:1, each moment of reality creating an indelible replica.

But do we gain from it, or was T. S. Eliot right to say that “all our knowledge brings us nearer to our ignorance”? Gleick is refreshingly upbeat. In the face of the information flood that David Foster Wallace called Total Noise, he says, “we veer from elation to dismay and back”. But he is confident that we can navigate it, challenging the view of techno-philosopher Jean-Pierre Dupuy that “ours is a world about which we pretend to have more and more information but which seems to us increasingly devoid of meaning”. Yet this relationship between information and meaning is the crux of the matter, and it is one that Gleick juggles but does not quite get to grips with. I’ll come back to that.

This is not, however, a book that merely charts the rising tide of information, from the invention of writing to the age of Google. To grasp what information truly means – to explain why it is shaping up as a unifying principle of science – he has to embrace linguistics, logic, telecommunications, codes, computing, mathematics, philosophy, cosmology, quantum theory and genetics. He must call as witnesses not only Charles Babbage, Alan Turing and Kurt Gödel, but also Borges, Poe and Lewis Carroll. There are few writers who could accomplish this with such panache and authority. Gleick, whose Chaos in 1987 helped to kick-start the era of modern popular science and who has also written acclaimed biographies of Richard Feynman and Isaac Newton, is one.

At the heart of the story is Claude Shannon, whose eclectic interests defy categorization today and were positively bizarre in the mid twentieth century. Having written a visionary but ignored doctoral thesis on genetics, Shannon wound up in the labs of the Bell Telephone Company, where electrical logic circuitry was being invented. There he worked (like Turing, who he met in 1943) on code-breaking during the Second World War. And in 1948 he published in Bell’s obscure house journal a theory of how to measure information – not just in a phone-line signal but in a random number, a book, a genome. Shannon’s information theory looms over everything that followed.

Shannon’s real point was that information is a physical entity, like energy or matter. The implications of this are profound. For one thing, manipulating information in a computer then has a minimum energy cost set by the laws of physics. This is what rescues the second law of thermodynamics (entropy or disorder always increases) from the hypothetical ‘demon’ invoked by James Clerk Maxwell in the nineteenth century to undermine it. By observing the behaviour of individual molecules, Maxwell’s demon seemed able to engineer a ‘forbidden’ decrease in entropy. But that doesn’t undo the sacrosanct second law, since processing the necessary information (more precisely, having to discard some of it – forgetting is the hard part) incurs a compensating entropic toll. In effect the demon instead turns information to energy, something demonstrated last year by a group of Japanese physicists – sadly too late for Gleick.

In quantum physics the role of information goes even deeper: at the level of fundamental particles, every event can be considered a transaction in information, and our familiar classical world emerges from the quantum by the process of erasing information. In quantum terms, Gleick says, “the universe is computing its own destiny.” By this point we are a long way from cuneiform and Morse code, though he makes the path commendably clear.

Moreover, Gleick does so with tremendous verve, which is mostly exhilarating, sometimes exhausting and occasionally coy. He is bracingly ready to use technical terms without definition – nonlinear, thermodynamic equilibrium – rightly refusing any infantilizing hand-holding. What impresses most is how he delves beneath the surface narrative to pull out the conceptual core. Written language, he explains, did not simply permit us to make thoughts permanent – it changed thinking itself, enabling abstraction and logical reasoning. Language is a negotiation whose currency is information. A child learning to read is not simply turning letters into words but is learning how to exploit (often recklessly) the redundancies in the system. She reads ‘this’ as ‘that’ not because she confuses the phonemes but because she knows that only a few of them may follow ‘th’, and it’s less effort to guess. Read the whole word, we tell her, but we don’t do it ourselves. That’s why we fail to spot typos: we’ve got the message already. Language elaborates to no informational purpose; the ‘u’ after ‘q’ could be ditched wholesale. Text messaging now lays bare this redundancy: we dnt nd hlf of wht we wrt.

Shannon’s take on language is disconcerting. From the outset he was determined to divorce information from meaning, making it equivalent to something like surprise or unpredictability. That’s why a random string of letters is more information-rich, in Shannon’s sense, than a coherent sentence. There is a definite value in his measure, not just in computing but in linguistics. Yet to broach information in the colloquial sense, somewhere meaning must be admitted back into all the statistics and correlations.

Gleick acknowledges the tension between information as Shannon’s permutation of bits and information as agent of meaning, but a reconciliation eludes him. When he explains the gene with reference to a Beethoven sonata, he says that the music resides neither in acoustic waves nor annotations on paper: ‘the music is the information’. But where and what is that information? Shannon might say, and Gleick implies, that it is in the pattern of notes that Beethoven conceived. But that’s wrong. The notes become music only in the mind of a listener primed with the cognitive, statistical and cultural apparatus to weave them into coherent and emotive forms. This means there is no bounded information set that is the music – it is different for every listener (and every performance), sometimes subtly, sometimes profoundly. The same for literature.

Lest you imagine that this applies only to information impinging on human cognition, it is equally true of the gene. Gleick too readily accepts the standard trope that genes – the abstract symbolic sequence – contain the information needed to build an organism. That information is highly incomplete. Genes don’t need to supply it all, because they act in a molecular milieu that fills in the gaps. It’s not that the music, or the gene, needs the right context to deliver its message – without that context, there is no message, no music, no gene. An information theory that considers just the signal and neglects the receiver is limited, even misleading.

It is the only serious complaint about what is otherwise a deeply impressive and rather beautiful book.

No comments: