Sunday, March 19, 2006

LEO: alternative future history

Just been given a copy of Georgina Ferry's book on LEO, the pioneering family of business computers developed by Lyons in the late 1940s. I was broadly aware that they had been a long way ahead of everyone else, having been the first company to use a computer to do anything useful and also having based their design directly on Maurice Wilkes' Cambridge EDSAC. (They have a website, here.)

What I didn't know, but didn't surprise me, was the speed with which a recognisable geek culture evolved as soon as the pooter was installed. The Lyons and Cambridge Maths Lab developer teams were exactly the sort of people, with exactly the sort of working methods and (non-)hierarchy, who develop good software and engineering now. Not necessarily engineers or mathematicians, but driven by enthusiasm and committed to open exchange of ideas, practical experiment, and meritocracy. The bad sides were just the same, too - tightly-wound geek freakouts and personality clashes, absurd workloads, maddening eccentricities and poor social skills.

What did surprise me was how many of the current tech issues they faced and sometimes prefigured. Service-oriented architecture (although it wasn't called that then) was in place from day one. Software-as-a-service and the integration of business process analysis and software design, too, were strengths. Indeed, the LEO team survivors now claim that Where It All Went Wrong was that after the English Electric takeover, they stopped selling computers as part of a whole-system engineering/consulting service and just began trying to sell boxes.

Even more surprising was the unfamiliar programming culture that the LEO team created - very unfamiliar, that is, to the IT world. Some other industries, however, would have instantly recognised it. And if they were Japanese car makers, they would have wondered who was stealing their ideas. Whenever a program was tested on the computer (which had to be minimised because of the machines' packed schedule), if a fault occurred the test was instantly stopped and not resumed until it had been fully documented, analysed and fixed. However senior the author, no line of code could be tested until another programmer had peer-reviewed it. It was a software version of Toyota's now famous decision (around the same time - 1958) to allow any worker to stop the line if they discovered a faulty part, and keep it stopped until the fault was traced. Quality through lean production, for code.

Which begs the question - if Leo Computers, not IBM, had cracked the market, and if Lord Weinstock had decided to keep Marconi's Rochester microchip plant (the biggest in Europe, shut down and sold for a supermarket site), what would a British-dominated geek culture have looked like?


Anonymous said...

Boffins rather than geeks.

But would that have been a difference? I find that the movie The Dish suggests something, but that's Australia.

Anonymous said...

"Java" would be called "Tetley".

ohdawno said...

Have you read The Difference Engine by Gibson and Sterling? They created very some interesting British geek characters - and the British Empire does dominate technology in this alternate timeline story(if I recall correctly - it's been awhile since I read it).

kostenloser Counter