John Naughton
The greatest thinkers are always those who can learn from their mistakes. Take Fred Brooks, the engineer who led development of the operating system for the IBM 360 mainframes in the late Sixties.
It was the largest software project undertaken up to that point and, like all such projects, it ran over time and budget. But eventually the series was one of IBM’s most successful products.
At which point, most project managers would have breathed a sigh of relief, updated the CV and moved on.
But not Brooks. Instead he quit IBM, became a professor at the University of North Carolina in Chapel Hill and began to reflect on what had gone wrong in the OS360 project and why. The result of his cogitations was an elegant collection of path-breaking essays, The Mythical Man-Month, published by Prentice Hall in 1975. It has never been out of print since.
The curious title comes from Brooks’s intuition that the traditional way of measuring programming effort did not make sense. Whenever a software project falls behind, the instinctive managerial response was and still is to throw more programmers at it. But Brooks’s experience on OS360 suggested that, in practice, adding programmers to a failing team actually slowed it further.
Rereading the book is salutary, for although most of the technology mentioned is nowadays found only in museums, many of Brooks’s central insights are as relevant as ever.
A couple of years ago I met him for lunch. He turned out to have a wry sense of humour and a rare insight into the computing industry. You could caricature it, I suppose, by calling it the view from 18000m. After lunch he gave a lecture on how the entire history of the computer industry can be explained as a series of ultimately unsuccessful attempts to escape from the fundamental architecture for digital computing laid down by John von Neumann in the late Forties.
To reach this insight, Brooks had to strip away all the superficial gloss of successive generations of computers all the stuff about memory, instruction sets and pipeline caches that impress the computer press to get at the underlying characteristics of each machine and compare it with the axioms laid down by Von Neumann.
It was an awe-inspiring performance. I came away thinking that we need more people like Brooks to help us think about the sea changes under way on the Internet, from new authentication technologies for e-commerce to Microsoft’s dot-Net initiative, which will alter the network’s architecture in fundamental and potentially troubling ways.
We are bombarded daily with press releases, information about new technologies and processes, White Papers and predictions about the New, New Thing. The problem is how to winkle out from this blizzard of information some real understanding of what it all means. Or, put in older communications technology, how to see the wood for the trees.