The Soul of a New Machine

A group of engineers, many of them recent graduates, building a 32-bit minicomputer in roughly a year, under a manager who has committed to a schedule nobody believes.

The machine matters less than the people. Kidder embedded with the team for the duration and wrote about what building something difficult does to the people doing it.

The machine is the occasion. The subject is what building something hard does to the people building it.

Devmark’s reading — not a quotation from the book

The book's central concept and its most enduring contribution to the vocabulary of technical work. Engineers were recruited by being offered a difficult problem and near-total autonomy, and in accepting they signed up — committing to whatever hours the work required, without being asked directly.

Kidder is clear-eyed that this was a deliberate management technique, and that the engineers largely knew it and accepted anyway.

The two sub-teams: the hardware engineers and the younger engineers writing microcode, working in parallel on a machine that had to work as one thing.

The interfaces between them, and the arguments at those interfaces, are the book's best organisational material — and describe a dynamic that anyone who has shipped a technical product will recognise immediately.

The project's manager, and one of the more complicated figures in technology writing. He is manipulative, evasive, absent, occasionally dishonest with his own team, and he also creates the conditions in which they do the best work of their lives.

Kidder does not resolve this. West is presented as effective and not admirable, which is a harder and more honest portrait than the genre usually attempts.

The team's own name for West's approach: keep them in the dark and feed them manure. Information about corporate politics, competing projects and the machine's actual prospects was deliberately withheld.

The effect was to protect the team from distraction and to deny them information they had a legitimate interest in. Kidder reports both without adjudicating.

Keep them in the dark. It protected the team from distraction and denied them information they had a right to.

Devmark’s reading — not a quotation from the book

The book is honest about what the project took. Marriages strained, health neglected, several engineers leaving the company immediately after shipping, and a collective flatness at the end rather than triumph.

West's line about the process being a matter of no one being interested in the outcome once it arrived captures the deflation, which is the part of technical work least often written about.

The technology completely — minicomputers, the specific architecture, the competitive landscape — and the workforce, which is almost entirely young, male and American in a way the book notices only faintly.

The deadline-driven heroics are also presented with less scepticism than a contemporary account would apply, and the book reads in places as celebrating a culture that has since been examined more critically.

The organisational dynamics, entirely. Ambiguous authority, informal status, the politics of interfaces between teams, the manager who manufactures urgency, and the specific psychology of committing to something difficult are all exactly recognisable.

Kidder's central insight — that people will work extraordinarily hard for an interesting problem and a sense of ownership, and that this can be elicited deliberately — remains the mechanism most technical organisations run on.

It is the reason the book won a Pulitzer. Kidder makes a computer's internals legible without simplifying them, and he is exceptional at conveying an engineer's relationship to a problem.

It is also a genuine work of literary journalism rather than a business book, which means it withholds lessons and lets the reader draw them.

The signing-up dynamic, recognised for what it is: a management technique that produces extraordinary effort by offering autonomy and an interesting problem, and that has a cost the person signing up rarely prices correctly.

And the flatness at the end, which is worth knowing about in advance by anyone who has attached a great deal to shipping something.