In The Soul of a New Machine, engineers compare their work to pinball. Winning a game earns another game; building one computer brings the reward of participating in another design. It can sound as though even the prize is overtime. For them, continuing is precisely part of the attraction.
Tracy Kidder follows such a team in late-1970s Massachusetts at Data General, developing a computer code-named Eagle. It was a minicomputer, not the personal computer later familiar to households. Despite the name, its principal customers were institutions, and its design required complex hardware and systems work.
Project leader Tom West had to catch up with competition. Rivals had new products; another internal team was also developing a next-generation machine. His group had to show that those left here could still build something good. Kidder stayed in offices and laboratories as drawings, arguments, and repeated tests gradually became a functioning computer.
Readers need not understand circuitry first. The driving question is accessible without it: why were these people so willing to invest themselves, and what opportunities did that willingness give others to use them?
The Newcomers Really Have Something to Do
Many Eagle engineers had recently left school. Inexperience usually means support tasks while others decide whether to trust you. Here, short staffing and limited time brought real responsibility quickly.
The hardware group were the Hardy Boys; the microcode group, the Microkids. Microcode can be roughly understood as a finer internal layer of control instructions. Their work could not simply be completed separately and placed in a box at the end. Circuitry and control had to cooperate. A change on one side affected another; a failure required deciding which part had stopped the whole machine.
This gives work immediate appeal. A judgment enters the finished object and is tested by it. An elegant report cannot repair malfunction. Solving a previously elusive problem does not require a supervisor's certificate before the person doing it knows something has been accomplished.
Such chances are unusual early in a career. Responsibility, argument with knowledgeable people, and learning through difficulty are rewards themselves. Bonuses alone cannot explain the experience. Goodreads reader Josh Davis, writing after experiencing a major product launch, describes newly understanding the comradeship, pride, and common purpose. Distance from the engineers may depend less on using their old computer than on having made something difficult together.
Kidder makes that attraction credible. Rather than announcing satisfaction once results are known, he keeps readers where reliability, design changes, and remaining time are uncertain. Success is not a conclusion already supplied. It is work a group can still fail to complete.
Why Must a Good Design Compromise?
A new machine does not permit every part to begin anew.
Eagle needed compatibility with the company's existing software. Programs customers had purchased and practices they had developed could not simply be discarded for an engineer's more elegant machine. The task included carrying old conditions forward, not only improving performance.
That matters for nontechnical readers. Engineering is seldom a problem with one standard answer. Speed, cost, reliability, and delivery can constrain each other. A better solution in one respect may fit the project less well. Arguments are not merely intelligent people correcting fools. Sometimes a better method is known, but there is no time to rebuild everything supporting it.
Design also becomes minute manual work. Hardware leader Ed Rasala and other engineers alter connections on prototype boards. Using tweezers, he picks out a tiny wire fragment and explains how stray pieces amid dense wiring can cause elusive intermittent faults. Hands fixing one problem can introduce another.
The trade-offs therefore do not occur only around conference tables. Knowing how a circuit should work differs from altering it and confirming that it keeps working. Much unglamorous labor lies between. Praising inspiration alone overlooks the insistence represented by Rasala: however troublesome this step, it must be completed. Reliability cannot rest on confidence when a proposal is made.
A team needs attention to the whole and people willing to investigate supposedly trivial problems. West cannot do all the latter himself. Part of his role is securing room for the project within the company. Yet someone who shields a group from external pressure while transmitting pressure inward makes protection and compulsion difficult to separate.
Enthusiasm Is Not a Blank Check
The book leaves uncomfortable material about West's management. He distances himself from young engineers and does not always share what he knows. People already working hard must also guess what their leader thinks of them.
After success, this can be recast as brilliant motivation: withheld approval drove improvement. That makes an outcome prove a method. Producing something good in such conditions does not show that every exhaustion, anxiety, and unnecessary guess contributed to it.
Developer Adam Tornhill's review understands creative excitement while questioning how long the arrangement could last. Its contradiction is straightforward: not working only for money does not make pay and conditions irrelevant. Someone who loves a problem may be especially reluctant to leave it after conditions have become unacceptable.
Calling all the engineers dupes would be no better. Skills, friendship, and creative satisfaction are genuine. Declaring everything exploitation would erase what they valued. The harder question is whether companies and leaders retain duties toward the limits of an investment people genuinely want to make.
If completion requires nobody daring to go home first, voluntariness has become complicated. Refusing extra work can mean leaving it to a colleague or fearing exclusion from the team. Such pressures need not appear in regulations to affect daily decisions. Alongside enthusiasm, readers should notice what stopping might cost.
Pinball now supports two readings. It can express the maker's pleasure: I want to do more. It can also defer an employer's obligations: win this game, then we will discuss the next. Both look like commitment, but who decides when the next game begins and whether to join it is not the same question.
Completion Is Not a Happy Ending
Eagle reached the market as the Eclipse MV/8000. Success did not ensure the original group could continue working together securely. In a 2000 WIRED retrospective, Evan Ratliff revisited changes around the project's completion: the team broke up and West was transferred to Japan. This makes product launch harder to treat as a settlement for everyone involved.
The brief retrospective tests the promise that successful making earns continued making. Engineers can pass technical tests without determining organizational decisions. What they produce enters corporate assets. Keeping a team or deciding their next assignment depends on other relations of power.
Treating the book as a project-management handbook would therefore miss something important. Recruitment and problem-solving techniques are visible, but taking only these away would turn engineers' particular lives back into material for higher output. Why someone finds work attractive and whether a manager should exhaust that attraction are different questions.
The book can inspire excitement and hesitation together. Young people entrusted with real responsibility deserve our pleasure; the dependence of their next opportunity on others prevents pleasure from completing the answer. Creation here is neither pure freedom nor wholly meaningless expenditure. Both experiences occupy the same laboratory and the same people.
Return to the pinball machine. The engineer's desire to continue is worth preserving. What deserves reconsideration is why continued victory and proof must be the price of remaining with beloved work. A functioning machine is a real achievement. How its makers may subsequently work and live remains unsettled.