From David Bailey
Though I read Peter Bentley’s article on “perfect software”, with interest, I feel I must inject a little realism (6 March, p 28).
Computer programming is unfortunately subject to periodic fads. Each fad begins by being hailed as a new paradigm that could make most programmers redundant. For example, neural networks were supposed to allow the computer to learn like the human brain. Future computers would not require programming, merely a bit of teaching – usually with a list of example inputs and the desired response. Following that, as Bentley acknowledges, the idea of genetic programming was touted as the next great step forward before its limitations were encountered. Each fad, so far at least, ends up all but forgotten.
It is all too easy to hype up new methods, and underestimate more conventional modern software and the people who produce it. Most commercial software is enormously complex – indeed it is one of the marvels of the age. If the guts of software were visible as it operated, like the parts of a machine, the intellectual effort that created it might be better appreciated.
When Bentley has grown a commercially viable piece of software – a word processor, say – I will be much more impressed.
Advertisement
Hyde, Cheshire, UK
