Gordon Welchman Read online

  Then, having arranged my discharge, the sergeant gave me a few appropriate papers, one of which I treasured for many years. It urged me to join the Home Guard, where my experience in the Army would be extremely valuable.

  So like many thousands before him, Welchman handed in his pass at the gate and left BP, never to return.

  While pondering on his future at the end of the war, it is unlikely that Welchman ever seriously considered taking on a role in the research and management of business. However, such an opportunity came along through his old BP friend and colleague Hugh Alexander. Before the war, Alexander had held the position of Director of Research for John Lewis and Co., then as now one of Britain’s largest groups of department stores with a staff of about 10,000 at that time. Alexander recommended Welchman as a candidate for his former post and the chairman of the company, Spedan Lewis, made him a very attractive offer, which he duly accepted. As the job was based in London, it was not practical to commute daily from Great Gransden. The Welchmans had to sell their beloved Rippington Manor and moved to a new house in Cookham on the Thames. The road through the village led to Maidenhead which had its own railway station on a branch line about an hour from London. From there, Welchman began the daily commute to London and his new job. John Lewis was not BP and now that he was working normal hours he had more time to spend with his family. His son Nick later remembered his sister Ros being christened around this time and a seaside holiday in Swanage where the family stayed at a hotel with a swimming pool close to an area with an amusement arcade. Katharine could now take steps towards her musical career as a soprano and pianist. She had studied at the Royal Academy, and was thinking about giving recitals, which quite soon she did, at a grand cinema, the Odeon, in Maidenhead.

  The man who had run the John Lewis business throughout the war, Michael Watkins, also lived in Cookham. In discussions with Watkins, Welchman realized what a privileged position he had been in during the war:

  Discussing our respective experiences, Michael Watkins and I found many differences. I lived in quiet surroundings that were never subjected to anything more than trivial accidental bombing. (I do not remember hearing of anyone being hurt.) When I had to travel any distance on business, I was provided with a car and a female chauffeur. Because of the importance of my work I was excused from Home Guard duty. Watkins, though a man of some prominence doing an important, exacting job, was shown no such consideration. He lived at Cookham on the Thames, in normal times a railway journey of about an hour from London. At the end of each day’s work at his office in Cavendish Square near Oxford Circus, he had to struggle across London to Paddington Station in the hope that a train would take him home. The journey was often nightmarish, delayed either by bomb damage or by actual bombing attacks. Yet when he got home to Cookham he would do his full share of Home Guard duty.

  Welchman found the next three years very interesting as he was able to study a wide range of business problems and see them from the point of view of top management. He was Acting Director of Personnel for about nine months and Acting Director of Expansion for about the same period, while other directors were ill. He had the opportunity to represent the company on the committees of a number of national trade organizations. Welchman was very impressed with the vision of Spedan Lewis, whose father, John, had founded the business in 1864. Spedan Lewis had developed power-sharing policies by sharing the profits the business made with the employees. The democratic and profit-sharing nature of the business was developed into a formal partnership structure and, in effect, Spedan Lewis bequeathed the company to its employees.

  In June 1946, it was announced that Welchman along with his BP colleague Alan Turing had been appointed to the Order of the British Empire for their war services. The OBE was standard for civil servants of the rank that Welchman and Turing had officially held. There were rumours that both had been considered for a higher award but that their rank within the Civil Service limited them to the OBE.1 Welchman took his family with him to Buckingham Palace to receive the award but he never felt compelled to use the letters after his name.

  The following year, the Welchmans decided to move to London and they purchased a house at 29 Holland Villas for £7,500. While life was good in London, Welchman began discussing with his wife the possibility of moving to the USA. He had been very impressed with the Harvard Business School Case Study Program as part of his work for John Lewis. This looked at real outcomes while he felt that British universities studying business tended to be more concerned about theory and less about application. The UK was slowly trying to return to normality while in America major advances were being made and opportunities were opening up in fields of interest to him. He also felt that Britain was not expanding its ideas, while due to his wartime experience his head was full of possible innovative projects. The rigidity of the class structures in Britain made it difficult to propose new ideas in fields such as engineering. He confided to a friend years later that, during that time, he felt as though he was hitting his head against a brick wall. Welchman was very interested in the American education system and felt that his children might get a different kind of education there than in Britain. In fact, the only negative thought he had ever harboured about the USA was the delay in its entry into the First World War. His brother Eric had died very early in the conflict on 24 August 1914 at the age of twenty-one.

  In anticipation of persuading Katharine that their future lay in America, Welchman began to contact some of the Americans whom he had worked with during the war. Many of them were now involved in the fledgling computer industry. As he was doing so, an excellent and unsolicited letter of reference arrived from the Director of the Government Communications Headquarters. GC&CS had formally become GCHQ after the war and Travis had stayed on as its Director. The letter, dated 28 October 1948, revealed Travis’s high regard for Welchman:

  Dear Welchman,

  I have recently been reviewing the wartime work of this organization with particular reference to the contribution of individual members of the organization and I should like to place on record my appreciation of the important and outstanding part you played. Your quick mastering of a number of different aspects of the work of which you had no previous experience was most noticeable and your inventiveness and ability in the field of applied mathematics and electronics provided a notable contribution to the success of the organization. From 1943 to the end of the war your services as an Assistant Director gave you an opportunity to display your organizing and administrative ability of which you took full advantage.

  I hope your wartime experience will be of real value to you in Civil life in which I wish you every success.

  Yours very sincerely,

  Sir Edward Travis, K.C.M.G., C.B.E.

  Somehow, Welchman had managed to persuade Katharine to take the plunge and move to the USA. As for many families emigrating to another country, it must have been a difficult and painful decision. It meant abandoning many friends, connections and people and, in Katharine’s case, the foundations that she had been laying for her own career. Welchman flew, on his own, to America in the autumn of 1948. He wanted to establish himself professionally and find a suitable location for the family to settle in. Back in England, the family received postcards from a succession of American cities. Katharine and the three children boarded the Nieuw Amsterdam, a Dutch ship, in December 1948, accompanied by a pedigree Staffordshire terrier. On 19 December they sailed into New York harbour where Welchman was waiting to greet them. They boarded a train for Boston at the beginning of a considerable snowstorm that raged for the entire journey.

  The family’s financial resources were now under some pressure as exchange rates between the dollar and the pound took away more and more of the value of their money and holdings, and the currency controls delayed the transfer of assets from one country to the other. The move was also expensive. Katharine owned two grand pianos and a clavichord, which she had shipped to America at considerable cost. Welchman had rented
a large house for them in Cambridge, Massachusetts, and the children were enrolled in private schools. Nick and Sue started going to Shady Hill School, which was within walking distance of their new home. The school was a standard bearer for progressive education, guided by principles drawn from John Dewey, the American philosopher.

  Welchman’s wartime ‘opposite number’ in the US Navy had been Dr Howard Engstrom. Before the war he had been a mathematician at Yale University. After the war, Engstrom had founded Engineering Research Associates (ERA), a pioneering computer company which ultimately became a division of Remington Rand. When Welchman arrived in America, Engstrom was one of the first people he contacted. However, in 1948, the ERA contracts were too highly classified for a non-citizen, so he had to look elsewhere. After considering a number of possibilities, he decided to take up a research post on Project Whirlwind at MIT under Dr Jay Forrester.

  Project Whirlwind had emerged from the early wartime computer developments in the USA.2 At the same time as the Colossus computer was being developed in Britain, work was also under way in America on an electronic and programmable computer for military purposes. An engineer called Luis de Florez, Director of the Special Devices Division of the Navy’s Bureau of Aeronautics, had approached MIT to build a new kind of machine to help train naval bomber flight crews. The machine that he asked for was not the machine which MIT ultimately built.

  De Florez had wanted a dual-purpose flight simulator, not a digital computer. In 1943 it was taking far too much time and money to train flight crews to man the increasingly complex aircraft in production. British engineers were addressing the same issue and had built a pneumatically operated trainer, the ‘Silloth’. In America, the Navy and MIT came together and using Navy funding and MIT technical expertise, embarked on the development of the Airplane Stability and Control Analyzer (ASCA). MIT’s Servomechanisms Laboratory was tasked with developing the ASCA system. The lab had been set up in 1940 by the Department of Electrical Engineering to develop a servomechanism to link a computing sight to 37-mm guns mounted on merchant vessels to repel attacks from German dive bombers. By 1944 it had around a hundred staff including thirty-five engineers.

  The ASCA project was formally launched on 18 December 1944 and the director of the lab, Gordon Brown, appointed one of his assistant directors, Jay Forrester, to head it up. Forrester had gone to MIT as a research assistant after graduating in 1939 and he eventually spent his entire career there. He in turn brought in Bob Everett to share the responsibility and the technical direction of the project.

  The ASCA development was not going well by the summer of 1945 due to the limitations of analogue computational machinery. Forrester had discussions with Perry Crawford, a young electrical engineer whose 1942 master’s thesis had been called ‘Automatic Control by Arithmetical Operations’. Before leaving MIT to take up a post with de Florez in the Special Devices Division of the Navy, Crawford had told Forrester about digital calculators and a new breed of controlled-sequence devices. At the University of Pennsylvania in Philadelphia an elaborate valve (vacuum-tube in the US) calculator, the Electronic Numerical Integrator and Computer (ENIAC) was under development at the Moore School of Engineering under the direction of John W. Mauchly and J. Presper Eckert Jr. As the development of Colossus in Britain was classified, the two men mistakenly believed that their machine would be the first to use valve-based circuitry to carry out highly complex calculations. The intended application was processing ballistics data for the US Army. A successor to ENIAC was also under development, the Electronic Digital Variable Automatic Computer (EDVAC). Forrester was intrigued by these developments and, years later, both he and Everett would credit Crawford with moving them towards digital numerical techniques.

  Like Colossus, ENIAC was designed for a particular use rather than as a general-purpose machine with wide application. It had 18,000 valves compared to the 2,500 in the Colossus Mark II, and 1,500 electrical relays. The electronic computer tradition that followed on from ENIAC was based on wartime developments in the pulsed circuitry of radar, and the well-developed state of the art in radio valves. It would ultimately be transformed by the invention of the transistor and solid-state circuitry. It would incorporate the mathematical ideas of Charles Babbage, Edward Boole, John von Neumann and Alan Turing. So, by the end of the Second World War, computing pioneers on both sides of the Atlantic were forging ahead, pre-eminently Eckert, Mauchly and von Neuman in the USA and Maurice Wilkes, F. C. Williams and Turing in Britain.

  Forrester attended the historic first ‘Conference on Advanced Computation Techniques’ at MIT on 30–31 October 1945. He also visited the University of Pennsylvania and afterwards was determined to take a digital approach to his project. Forrester was attracted to the fact that patterned arrangements of energised valves symbolized numbers and that these patterns were built up from modular units of two valves. This design could be repeated over and over again like bricks in a wall to build a computer. He started to expand his staff and sought talented graduate students of doctorate degree calibre, experience and ingenuity being deemed more important than the degree itself. This was the team that Welchman would join in 1948 having easily met Forrester’s recruitment standard.

  The Navy raised a new contract with MIT in March 1946. In Phase I, the researchers were tasked with the construction of ‘a small digital computer involving investigation of electric circuits, video amplifiers, electrostatic storage tubes, electronic switching and mathematical studies of digital computation and the adaption of problems to this method of solution’. Phase II of the project would be to design an electronic computer and aircraft analyser based on the outcomes of Phase I. The project was given the name Whirlwind, one of a group of projects funded by the Special Devices Division, the others being Tornado and Hurricane. Forrester set out his vision for the Whirlwind computer in a paper delivered on 29 July 1948 at the Modern Calculating Machinery and Numerical Methods Laboratory at MIT

  Project Whirlwind is directed primarily at computer applications though it will continue to carry a heavy responsibility in machine development. The problems which will receive greatest attention are those requiring high-speed machines of some 20,000 arithmetic operations per second giving high computing efficiencies. The Whirlwind I computer, assembly of which is now under way, will permit such applications and studies and will be a proving ground for trouble location methods and circuits in order that improved high-speed computers may follow. One of the first applications for high-speed computers which will be developed will probably be the high-speed traffic flow network. This is chosen first because it is one of the simpler system applications and one which can be achieved in the near future. The lessons learned can be applied in the problems to follow. While computers appear to be capable ultimately of accomplishing most of the functions with which they are credited, it is almost certain that they will not do so at as early a date as anticipated. Many years and large sums of money lie between this audience and the realization of many of their hopes.

  By 1948, interest in the analyser was waning and the overriding question being faced by Forrester and his team was ‘What is the computer good for?’ To Forrester, the key was finding the right people to take the project forward. As he wrote to a colleague ‘The general type of man whom we need should have originality and what is often referred to as ‘genius’. He should not be bound by the traditional approach … I do not know of suitable prospects.’ Another major issue to be resolved was the problem of ‘internal storage’, known today as memory. For a computer to be general-purpose, memory was essential. Forrester considered a number of options including a mercury acoustic delay line which could store data in a linear arrangement. A two-dimensional storage arrangement was possible by making use of the electrostatically charged surface of, say, a disk in a vacuum tube. A number of insulated ‘spots’ could be placed on such a surface, with each given a positive or negative charge by focusing an electric beam on it. The so-called ‘electrostatic storage tube�
� was first tested successfully in Britain in June 1948. Its inventor, F. C. Williams, integrated it into the University of Manchester’s Mark I computer. However, Forrester had concerns about the speed and reliability of Williams’s solution.

  In 1947, Andrew Booth had delivered a lecture at the University of London in which he discussed the possibility of using magnetic cores as an alternative to storage tubes. Forrester realized the possibilities that this offered and assigned one of his staff, William Papian, to investigate further.

  There were shortcomings with all of the available storage technologies available to the Whirlwind team. They had recognized two broad areas of application for computers: scientific problem-solving, requiring large storage capacity and real-time control, and applications, requiring speed. As Forrester said:

  The fundamental storage problem is not so much that of discovering an element which will retain information; it is the far more difficult problem of switching or selecting among the large numbers of individual elements which make up the storage system.

  During 1948, Forrester had been in discussion with the US Air Force about it providing finance for a theoretical study of the technical problems involved in introducing a computer into air traffic control. This was one of many computer applications envisaged by Forrester. In the autumn of 1948 a new group was set up in the Servomechanisms laboratory with Welchman at its head. The group’s initial remit was ‘to study the mathematical background and coding procedures that will be put on Whirlwind when it is running’. In a memo to staff joining the group, Welchman said that the purpose of the group was ‘to do original work on problems of computer applications’. By the end of the year, Welchman and Forrester had prepared and submitted a proposal for a two-year study of air traffic control problems. In a covering note, Forrester had made it clear that the project would be the first phase of a longer-term project to provide an interim and then a target air traffic control system. Testing of the interim system might be possible by 1952. While the construction and installation of operational equipment was not part of the project, laboratory trials using the Whirlwind computer would explore coding problems and their solutions. Welchman would head the project group with Forrester, Everett and Philip Franklin of MIT’s Mathematics Department operating in an advisory capacity.