Inside Intel. Tim Jackson

the more capable an engineer, the more need there is to explore not only accomplishments but to try to analyse problems, failures and all that is negative about the engineer’s job performance.’

      He then offered a paragraph of measured praise, and then six further paragraphs demolishing Reed’s working practices point by point. He complained that the ‘problems the [1103] suffers from could fill a book’. Reed had ‘disengage[d] himself’ from day-to-day problems. He showed a ‘complete lack of initiative in taking (or identifying the need for) action’. It was ‘inexcusable’ that he failed to keep himself familiar with vital technical data. Reed was ‘disorganized’, and left ‘devices and data scattered about’. He always wanted ‘to move on too fast to new projects’.

      ‘I want to emphasize’, Vadasz concluded, ‘that I respect John’s engineering capability very much. John can solve any circuit problem we will have … My objections above all were all related to John’s working habits. Habits can be changed. I hope that John will be able to use this criticism in a constructive manner.’

      Vadasz was wrong. The review made Reed incandescent with rage. He instantly wrote a furious memo for his own personnel file, headed, ‘To Whom It May Concern’, complaining about it. Reed opened his reply by pleading guilty to the charge of needing corrective feedback from above to keep him working on matters that were important. Then he went on the attack.

      ‘Here is a typewritten, signed off, filed-in-my-records report,’ he wrote, ‘citing specific examples of bad things I have done in my eight and one-half month tenure here, each one of which should have been directly criticized on the day it occurred! … The practice of saving up all these “constructive” criticisms in one package, to be delivered below the belt (in the form of a copy of an already filed report!), after the fact, is an indefensible management policy.’

      So angered and disillusioned was Reed by the incident that he immediately cut back the eighty hours a week he had been spending at the plant, and started devoting more time to seeing his family and singing in a local choral group. Within another year he had left Intel for ever. When Andy Grove, hearing of his departure plans, asked him what could be done to make him stay, Reed looked at him uncomprehendingly. ‘This company is like a woman,’ he told Grove bitterly. ‘You’re madly in love with her, but then you find she’s been cheating on you. After that, you just can’t give your hundred per cent any more.’

      ‘We’d be happy with ninety-five,’ replied Grove.

      But it was no good. Reed was determined to leave the company, and the only way to make the best of a bad job was to circulate the Vadasz review and the Reed response around the company as a lesson in how the business of reviewing performance should be approached. Intel would later institute a series of regular lunches at which engineers and technical staff would have a chance to make their grievances known to the company’s top management. They became known as the ‘John Reed Memorial Lunches’.

      It was understandable that tempers inside Intel should run high. The conventional rule of thumb in the industry was ‘one man, one chip, one year’. Individuals took personal possession of design projects in a way that would be impossible today, where teams of 100 or more people can work on a single chip. A circuit designer would spend months drawing up his design. He – there weren’t any women engineers at Intel in 1971, or at most other electronics companies either – would work long hours working with a ‘layout girl’, a designer whose job was to cut the image of the circuits into a giant sheet of a red plastic-like substance called ‘rubylith’. He would check the design again and again over a period of weeks, following lines of different colours around the circuit with a finger to make sure that everything connected up properly. Finally, after months of perfecting the process, the first prototypes would be fabricated, with the designer waiting tensely outside the wafer sort area, where the devices would be tested and then sliced into individual chips, as if outside a delivery room. There were two possible outcomes: the horror of a stillbirth, or the elation of a working device. But it was an intensely personal experience. And nothing would make an engineer more angry than the hint that credit for an idea or a piece of work was going in the wrong direction.

      One day in August 1970 Joel Karp opened the latest edition of Electronics magazine on his desk. To his astonishment, he found an article in the magazine about the 1102 chip written by Ted Hoff. True, Hoff had devised the dynamic memory cell that the chip was based on. But Karp felt the 1102 was his chip. Incensed, he rushed into Gordon Moore’s office and threw the magazine on Moore’s desk, open at the beginning of the offending article.

      ‘I promise you,’ said Moore with tears in his eyes, ‘I promise you that as long as I am in this company, nothing like this will ever happen again.’

      But the jealousies inside the MOS group were nothing to the rivalry between the MOS group and the group of engineers working on bipolar circuits. As the 1103 began to make headway in the market, the company shifted resources away from bipolar technologies towards MOS. The bipolar engineers began to find it harder to book time on the testing equipment. Their prototypes would always take longer to emerge. Their projects would receive less attention from Noyce and Moore.

      Commercially, this was probably the right decision. Even the most diehard enthusiasts for bipolar, such as H. T. Chua, whose faith in the technology had brought in the $10,000 from Honeywell, would admit later that MOS was the more promising process. At the time, however, the downgrading of the bipolar operation was taken by the team as a personal slight. How could they take it otherwise, when they had devoted so many late nights, so much effort, so much emotion, to making their bipolar projects work?

      Vadasz, as head of the MOS engineering team, did not help matters. He had a knack of saying hurtful things to the bipolar people – not in meetings but just in corridors, in the company cafeteria, in the parking lot – that could spoil an entire day for them. Some of his greatest venom was reserved for Dick Bohn, the bipolar team leader. Bohn was under considerable stress to make bipolar perform. At the same time, he was left in no doubt that the company considered his beloved bipolar process no more than a transitional technology that would soon be phased out until it was used only for expensive niche products working at especially high speed. In the end the pressure became too much. Bohn began to drink heavily, and his work started to suffer. He was later eased out of the company, into a downward spiral of alcoholism and mental illness from which he did not recover for some time.

      The fate of Dick Bohn was a terrible reminder to everyone at Intel of the high human cost of the conditions they worked under. Even Andy Grove, who was largely responsible for setting those conditions, felt a pang of conscience. A few years later, at a company retreat in the resort town of Pajaro Dunes, Grove would confide to a colleague that his contribution to the departure from Intel of Dick Bohn was the one thing in his life that he regretted most.

7

       Yellow Snow

      IT WAS BECOMING CLEAR that Intel wasn’t a startup any more. The company’s sales had grown from a token $566,000 in 1969 – largely earned from the bipolar circuit built for Honeywell – to a more serious $4.2m. Although a sharp downturn hit the electronics market in late 1970, forcing the company to lay off some of its workers, the introduction of the 1103 in October made it inevitable that the company would expand. Its workforce, which had passed the 100 mark at the beginning of the year, was approaching the point where Noyce, Moore and Grove could no longer expect to know every employee by name.

      Intel was also on the move. Anticipating that it would be impossible to meet demand for the 1103 from the old Union Carbide plant in Mountain View, the company had bought twenty-six acres of orchards further south in Santa Clara, where property prices were lower. By spring 1971 the plum trees, apricots and almonds that covered the site were uprooted to make way for a large new fabrication plant – and the company was ready to move its manufacturing operations into Santa Clara 1, as the new plant was to be called.

      Other companies near by were expanding too. As the Bay Area became