VORSPRUNG TROTZ TECHNIK? *

Dick Pountain/14:48/06 August 1995/IDEALOG 12

A front-page story in my Sunday paper today informs me that high-street banks rip off their customers by charging £17 for sending letters to tell them they're overdrawn. Surely with all their computer technology, and having recently 'downsized' about 20,000 of their staff onto the dole, the banks must be able to produce such letters for a few pence? Er, well, probably not if you believe a new book "The Trouble with Computers" by Thomas K. Landuaer. His thesis - backed by many impressive sources - is that the £4 trillion dollars spent on computer technology over the last 20 years has contributed only a negligible, perhaps nil, increase in productivity.

Landauer is not your run of the mill cyberphobe - currently Professor of Psychology at the University of Colorado, former Director of Cognitive Science at Bellcore, he's spent 30 years designing, writing, commissioning and using software from mainframes to PCs. His arguments deserve to be heard and understood. First let's be clear that Landauer is not using 'productivity' in the woolly sense that software marketeers do, but in the precise sense of the economist - a quantity measured either as value added per person-hour employed ('labor productivity') or dollars worth of new output for each dollar spent on labour and capital combined ('multifactor productivity').

He also distinguishes two phases in the deployment of computers since their invention just after WW2. Phase 1 employed the extraordinary calculating power of computers to automate tasks that human beings just couldn't do - controlling oil refineries or high-speed machine tools, calculating complex insurance premiums - and achieved large increases in productivity comparable to those of previous inventions like the wheel, the plough, the machine screw or the steam engine. It's Phase 2, the application of computers to 'white collar' jobs in the service sector (insurance, banking and other office work), where the computer is supposed to be an aid to the human mind, that Landauer claims has flopped. The first half of the book is devoted to weighing evidence that productivity in the service sector has essentially been static from 1975 to 1992, despite huge investment in computers.

The spinning jenny caused an 300% increase in output per hour, and reduced the cost of cloth by 80%. Progress in computing has been even more spectacular if you measure it in MIPS, but the fact is that, apart from scientists and engineers, most people don't need many MIPS (except to play Doom.) The cost of writing a letter has shown no equivalent reduction.

Landauer deals in depth with the excuses and the reasons for this so-called 'computer paradox'. Among the excuses: it's too soon to tell (20 years is not enough?); it's a different sort of productivity (the kind you can't measure); it's an accident that worldwide recession has concealed the benefits; computers buy you market share, not increased output; that we've ceilinged out and just can't get any more productive; that managers are complacent and don't care about productivity.

Among the reasons: computers may just enable expansion without increasing productivity (eg. the car rental business); they do things that are irrelevant to productivity (like adding 10 fonts to a simple letter); they enable a huge proliferation of pointless products (eg. airline fare deals) or analyses (that wad of graphs at the board meeting) that soak up any productivity gains; or that most software much too hard to use, which is Landauer's overall favourite. He devotes the second half of the book to a prescription for making software useable via 'User-Centred Design' techniques. One of his most devastating points is that software designers and programmers are much TOO bright, which makes them incredibly poor predictors of how difficult ordinary folks will find it to use their creations; one of the unhealthy effects of computers has been to drastically increase the gap between the best and worst practitioners for many tasks, from 2:1 to around 10:1.

My own feelings fluctuated wildly as I read the book, indignantly re-discovering each of the excuses only to find them shot down two pages later. Writing is a white collar occupation, and my own productivity must have increased enormously since the days I submitted long-hand copy written with a blue Biro. Except, it hasn't; it still takes me almost exactly as long to write an article as it did back then, but I save someone else the job of retyping it. My articles are certainly better (though in a hard to measure way) because I can now change things almost effortlessly, equivalent to doing dozens of drafts the old way. But even by this slightly fuzzy criterion, are my efforts better now under Windows than they were using Wordstar 3.4 under CP/M? Absolutely not. I could type just as fast in Wordstar (in fact those much-hated Wordstar keystrokes were cleverly designed to aid touch-typists like me, and I still use several of them, implemented as macros.) All that's changed is that everything looks prettier now.   

I have a few qualms about Landauer's arguments. It's my gut feeling, on no real evidence, that productivity may have taken an upturn in the last two years as Windows integration and client/server technology finally begins to work (after a fashion) - Landauer quotes few figures later than 1992. Also he has little to say about the social choice that needs to be made, between using increased productivity to increase output, or to reduce the number of people employed; US and UK employers currently seem to favour the latter. That's not to diminish the importance of this book, which all computer professionals ought to read.

"The Trouble with Computers: Usefulness, Usability and Productivity"
Thomas K. Landauer, 1995, The MIT Press. ISBN: 0-262-12186-7

* "Forward despite technology", with apologies to Audi.