156013916

David Berlinksi is a polymath. Like a proud peacock, he is not afraid to put his erudition on full display. I suspect, however, that he has (innocently, I'm sure) somehow missed Somerset Maugham, who aphorized: "to write simply is as difficult as to be good."

Berlinsksi valorizes the journey over the destination. I first stumbled across his writing years ago in A Tour of the Calculus--a book I enjoyed despite its rambling (some may say pretentious, verbose, off-topic, unedited) storyline. Tour, indeed.

Advent is not dry intellectual history: au contraire, mes amis (a Berlinskian interjection if ever there was one). The proof in the pudding is the closing Epilogue: Wallace Stevens' "The Idea of Order at Key West." Order, algorithms--you get the idea. Like Dover sole garnished with crisp, skinned Granny Smith apple flesh (because both are vaguely white). And I love Wallace Stevens.

Berlinski can be enjoyable if you know this going in. Like grandparents with spoiled, smarter grandkids, readers can leave--or love. ( )

Horrible prose. Long digressions and vignettes where the author tries to get some point about algorithms or meta-algorithmic thinking across through truly unfortunate unrequited-love stories. Yes, as bizarre as it sounds. There's some interesting history of Leibniz, Frege, Turing, Church, etc., buried in here, but eventually the book turns into an Paley-ian argument from design for the complexity of the universe. ( )

Berlinksi's premise with this book seems to be to explain the idea of what an algorithm is and the history of the discovery and application of algorithms. At least I think it is. The author's prose style is at times so impenetrable that I find it difficult to work out what his intention is at times. There's an air of breathless enthusiasm and continued climaxes which demand to be followed by even greater ones:

'Now! at last! the breakthrough is finally here! Until the next even better one!'

It's putting me off a subject with which I am fascinated and about which I would consider myself a reasonably well-educated amateur. There are parts which are wrong and there are other parts which are just misleading. Consider the discussion, towards the end of the book, of the P/NP problem. Consider also that you either recognise what I mean by the P/NP problem or you don't. Either way, you're potentially amongst the audience for this book. If you don't already know what a polynomial problem is and how it differs from a non-polynomial problem you aren't going to be enlightened by the explanation here; you might even end up with completely the wrong idea. If you *do* already know about these concepts you'll just be frustrated at the appalling explanation.

You'll be annoyed that we have 4 pages of whimsy to describe a form of the travelling salesman problem which is followed by a cursory description of an 'algorithm' to solve it which is anything but. What's more, this is towards the end of a book in which the concept of an algorithm should by now be firmly implanted in the mind of every reader, whether or not it was something they understood before. Yet Berlinski still writes as if we don't expect an algorithmic solution and as if we need explained to us yet again in the most basic terms why one might be desirable.

It's not all this bad. The description of Godel's Incompleteness Theorem is accessible and relatively compact and the same can almost be said of the material on Church's work on the lambda calculus. But overall I find the tone and the lack of direction infuriating.

Yet there's obviously an audience for Berlinski's work as he appears to be a well-selling writer of popular science. You may be amongst that audience, but based on this example I shan't be seeking out any of his other work in the future. ( )

Algorithms