The British Lyons

Amazing though it is, the world's first business computer was developed in the UK. Called LEO, it went live in 1951. It was unlike the other machines of its era, because LEO was not developed by an electronics company, but by tea shop chain Joe Lyons, to manage the stock control of its buns.

Most of the members of the team which developed LEO have written a book, User-Driven Innovation. Their work will strike a chord with many of today's IT professionals.

LEO was designed by business people, not techies, and the authors are adamant that this was the secret of its success.

Indeed, the few triumphs clocked up by the UK computer industry in the last 50 years have been user rather than supplier-driven. Colossus was designed to satisfy a user need - to win the war by cracking German codes. LEO is part of this pragmatic tradition.

The LEO story began in 1947, when two managers from Lyons toured the US in search of ways to increase efficiency. They found that universities and research institutions considered computers to be useful only for scientific work. They also rejected an electro-mechanical punch card machine from a firm called International Business Machines, because it was too complex for their needs. The Lyons team was after a fast and cheap device, not rocket science, to solve the company's stock control and payroll bottlenecks.

The managers returned to the UK and offered # 3,000 to Maurice Wilkes, head of the mathematics laboratory at Cambridge University, to develop a system. Wilkes had impressed the Lyons team with his Electronic Delayed Storage Access Computer (Edsac), which was one of the first stored-program computers in the UK. On the day Edsac ran its first program in 1949, the Lyons board decided it wanted to build its own machine. This also signalled the beginnings of many partnerships between business and academia.

In those days, universities did not have to be tempted by government 'technology transfer' grants to cooperate with outsiders. Partnerships happened because the participants had the wit to realise they needed each other. Pairings flourished between Lyons and Cambridge, Ferranti and Manchester, British Tabulating Machine and Birkbeck, and English Electric and the National Physical Laboratory. As a result, they produced more world-class computer technology in the space of 10 years than the UK computer science establishment has done in the previous 50.

Lyons did not intend to sell LEO, but because of its clear technical lead over other developments in the UK and US, it soon excited interest from other firms. By 1954, Lyons decided to market its machine formally by setting up LEO Computers. The second version of LEO was built in 1958.

It was much faster than the original and had an I/O channel linked to magnetic tape. Sales of LEO II were soon made to blue-chip companies including car manufacturer Ford Motors and steel makers Stewarts & Lloyds.

LEO was redesigned and in 1961, a multiprogramming version, LEO III, was released. It beat IBM's first multiprogramming computer, the 360, to market by three years. LEOs were used as far afield as South Africa, Australia and the USSR.

Despite success in industry, local government, customs and the Post Office, Lyons found it hard to sell LEO to central government. David Caminer, main author of User-Driven Innovation, still feels bitter that ministers and civil servants of the 1950s did not buy many computers. In the US, the Department of Defense poured $400m (# 258m) into IBM contracts, which sowed the seeds of the country's computer industry.

By 1963, the writing was on the wall for LEO. Lyons was going through a bad financial patch and felt it could no longer afford the expense of developing new models. It sold the rights to LEO to English Electric.

In 1965, English Electric launched System 4, a range of IBM-compatible machines to compete with IBM's juggernaut, the 360 Series. Despite its technical superiority, LEO was dropped.

Three years later, Tony Benn, the then minister of technology, performed a shot-gun wedding between English Electric and International Computers & Tabulators (ICT) to form ICL. This was against the wishes of Arthur Humphreys, who was then managing director of ICT.

Humphreys believed IBM compatibility was a big mistake. He argued that although it made it easier for customers to defect from IBM, compatibility also made it easier for the same customers to return to the IBM fold.

He believed that to beat IBM, you had to be better, and different.

Although LEO was dropped by English Electric, LEO Computers' staff continued to work there to support the machine's existing customers. Indeed, the systems continued to be used until 1981. However, the LEO team found it difficult to fit in with the English Electric culture. They were regarded as subordinate to sales staff but saw themselves as consultants, bringing business as well as technical knowledge.

John Aris, one of the book's authors, recollects: '(LEO) programmers designed systems, managed projects, taught, sold, in a gentlemanly sort of way, negotiated contracts, staffed exhibitions and invented.' A team designing a payroll system for the Army in the late 1950s, was far-sighted enough to consider, albeit briefly, the year 2000 problem.

LEO people were taught to 'rethink (customers' business processes) rather than automate what is there ... make the data sweat', says the book. Aris, who for some years was the director of the National Computer Centre, points out that this thinking is what we now call business process re-engineering.

But it was tough carrying out the process in machine code on LEO II. It only had a 2048x19bit word memory and the card-reader/punch and printer worked at 100 cycles a minute.

The LEO level of service could also be regarded as the pioneer of outsourcing. LEO people were expected to implement and manage the system for the customer. However, as the systems grew in popularity this level of service could not be sustained. And anyway, customers were keen to stand on their own two feet.

Some 40 years on, the wheel has turned full circle and customers are happy to have their computers managed by someone else. The LEO approach is back in fashion.

User-Driven Innovation highlights the fact that so much in the industry is not new, although ideas are dressed in new clothing. It is a shame that we do not consider the lessons to be learnt from the industry's forefathers.

The LEO team can certainly teach us a thing or two.

- User-Driven Innovation, The World's First Business Computer by David Caminer, John Aris, Peter Hermon, Frank Land and other members of the LEO team, is available from McGraw-Hill (01628) 23432, price # 35.

LEO: THE TIME AND THE PLACE

1947 Lyons agrees to provide Cambridge University with aid to complete Edsac

1949 Edsac successfully completes first live operation. Lyons decides to build LEO

1951 Lyons Bakery valuations system becomes world's first routine office computer application

1954 Lyons builds a payroll application for up to 10,000 staff. LEO Computers is set up

1958 LEO II is built as first commercial version

1961 First LEO III with multiprogramming installed

1963 LEO Computers sold to English Electric

1965 English Electric decides to discontinue LEO in favour of System 4 (IBM 360-compatible)

1968 English Electric and ICT merger forms ICL

1981 The last LEO in use at the Post Office is decommissioned

WHERE ARE THEY NOW?

All the principal players in the LEO story have gone on to enjoy varied and distinguished careers. We've tracked down some of them.

David Caminer, a director of LEO Computers, joined Lyons as a management trainee in 1936. At ICL, he headed the team which launched ICL's 2900 Series of computers. He then led the implementation of computers at the EC in Luxembourg.

John Aris joined LEO Computers in 1958 as a programmer, rising to senior consultant. After LEO, he assumed a number of marketing posts at ICL and later become head of computing at Imperial Group. In 1985, he was made director of the National Computing Centre.

Peter Hermon, another LEO programmer, joined the company in 1955. He installed a LEO system at Dunlop, and became worldwide head of computing for the company. He then joined Boac - which later became British Airways - where he masterminded the Boadicea and Babs reservation systems. He was then made managing director of BA's European division.

Frank Land joined Lyons' statistical office in 1951 and rose to senior consultant level at LEO. In 1967, he joined the London School of Economics, first as a lecturer and then as professor of systems analysis. Later, he became professor of information management at the London Business School.

Ralph Land, Frank's twin brother, joined Lyons in 1953. After helping to develop LEO and ICL systems, he joined Rank Xerox to help set up its Eastern European business.

After LEO and ICL, Mike Gifford moved to Cadbury Schweppes, where he rose to financial director, before becoming chief executive of the Rank Organisation.

Another former LEO employee, Tim Holley, is now the head of Camelot.

After joining LEO in 1960, Ninian Eadie was one of the few to stay the course with ICL. He was given a seat on the ICL board and helped steer the company from being a proprietary mainframe manufacturer to an open systems supplier.