A Computer Called Leo (P.S.) (original 2003; edition 2004)

In the 1930s, Lyons (a large catering firm) realised that their business was dominated by a very expensive and unwieldy accounting process; every individual sale made at a teashop would have to be reconciled by an army of clerks. As the profit on each transaction was only a fraction of a penny, the costs of processing it could easily turn the sale into a loss. They began to look for ways to make this process more efficient, a process which culminated - after the War - in an attempt to develop a computer.

At this stage, the late 1940s, rudimentary digital computers were available, but were entirely the domain of researchers. Lyons funded the development of EDSAC at Cambridge, and built a modified copy as "LEO", the Lyons Electronic Office; it came into operation in 1951.

LEO was distinguished from every other machine then in existence - and for some time to come - by being intended specifically to work on "business problems", rather than unadorned mathematical calculations. This required a whole range of new input/output apparatus, to handle the data being processed in a reasonable timeframe, and a great deal of effort put into ensuring it ran reliably - a machine doing research calculations could be offline for a few days with no downsides, but holding up payroll would cause a revolt!

The system developed quickly, expanding to control the company's ordering system (with a marvellously modern "computer call centre" to run it) and renting spare time on the computer to outside customers. This led to the development of two new generations of the machine - LEO II and III - which were actively marketed to other users.

Unfortunately, as with so much of British industry, a revolutionary product was doomed by a flawed approach. The sales process was incredibly time-consuming and involved - a customer would often have engineers seconded to them for several months before a pitch was even made - and they just didn't sell agressively enough in a market dominated by American imports. The company didn't realise the implications of what they'd built, they declined to back it, and when they were given the opportunity to sell up and recoup their investment they took it and walked away. The LEO operation was amalgamated with a number of other producers, which eventually became part of Ferranti, and then bought by a Japanese firm; production of the machines was shut down in the sixties, with the final few soldiering on until 1981.

I kept having this odd feeling that I was reading Backroom Boys again; this is exactly that sort of story. Sad story, but with an underlying note of triumph - this was great, we managed to build it, and it came out of a bakery.

A very interesting book, and an excellent read. Nice bibliography of sources on Lyons, computer history, and office history. ( )

"On Thursday, 29 November 1951, LEO took over Bakery Valuations from the clerks who had previously done the work, and became the first computer in the world to run a routine office job."

LEO wasn't the first computer in the world, nor the fastest. But it was hugely significant -- it was the first that wasn't built by, and for, scientists and mathematicians. It represented the first, uncertain steps down a road that we take for granted these days -- that of business automation. It was a computer designed to be used for the ordinary, the everyday. This book tells its story, and that of the men who dreamed and built it.

It's an interesting story, and Ferry tells it well. Much of it seems unlikely, and the whole episode deserves to be more widely known. One of Ferry's stated aims is to redress the imbalance which has seen LEO remain largely unsung and unacknowledged in histories of computing.

Computing would go on to be dominated by American firms, supported by investment from government and the military that their British counterparts failed to provide. But for a time, the leaders in the field were a handful of visionaries and pragmatists at a British catering company. Extraordinary. ( )

Ferry covers the history of LEO, the Lyons Electronic Office, the first computer designed by and for a business. Along the way, we learn about the history of the Lyons company, best known for its chain of teashops and other catering businesses, and how their management’s quest for ever better and more efficient handling of information led them to choose to develop their own computer, at a time when the only publically known electronic computers were ENIAC and similar machines designed for military use and machines developed in academia.

The Lyons company helped fund Cambridge’s EDSAC machine, then built the first LEO with numerous improvements over EDSAC. Lyons differed in their approach in another way—they wrote practical programs to achieve real business results. The teams would analyze business tasks by working with not only the people who wanted the programs, but also the people who did the work. As a result, their applications tended to work extremely well from the start.

The LEO team’s attempts to expand their business into supplying computers and consulting to other companies were hurt by management’s lack of support and understanding. Lyons’ attempts to get the British government to help fund their computer business to allow them to provide some real competition to IBM and other American companies also failed; the British government first refused to believe that Lyons—a restaurant and catering company—could provide better systems than established electronic companies, then crippled the British computing industry by forcing independent companies to merge into a muddled, ineffective, and noncompetitive company that staggered on for years before being absorbed into Fujitsu.

Ultimately Lyons failed, and LEO was eclipsed by the sales and marketing savvy of IBM and other American companies (helped, of course, by massive American government investment in these companies through the department of defense). But the Brits were there first, and their systems worked very well and continued to run into the 1960s. ( )