Peopleware

The problem it starts from

Tom DeMarco and Tim Lister were software consultants who had seen hundreds of projects. In 1987 they published Peopleware to argue something that still sounds slightly heretical in technology companies. When projects fail, they wrote, it is rarely because of the technology. It is because of people: how they are hired, where they sit, how they are managed and whether they ever become a team. They put the point in one sentence, that the major problems of their work were sociological rather than technological in nature, and the rest of the book is a guide to taking that seriously.

The book was revised in 1999 and again in 2013, and the core has hardly changed. Its chapters are short, often funny and frequently angry at what managers do to the people they manage.

The practices

Stop treating people as interchangeable. Much management, the authors argue, carries the mental model of a production line, in which workers are parts to be swapped and pushed to work faster. Knowledge work does not behave that way. Pressure to work faster mostly produces lower quality, and people forced to deliver work they are ashamed of become less committed. The authors argue that quality, set high by the people doing the work, is a source of productivity rather than a cost to it.

Measure the workplace, and fix it. The book’s best-known evidence comes from the “Coding War Games”, a set of programming exercises DeMarco and Lister ran between 1984 and 1986 with more than six hundred developers from ninety-two organisations, each working in their own office during normal hours. Performance varied enormously, both between individuals and between organisations. Experience, salary and the programming language used explained surprisingly little. What did correlate was the workplace. The best performers worked in spaces that were quieter, more private and larger, and could stop interruptions. People from the same organisation tended to perform alike, which suggested the environment was doing much of the work.

Protect flow. Concentrated work requires a state of immersion that takes about a quarter of an hour to reach and can be broken by a single phone call. Every interruption therefore costs much more than its own length. The authors propose measuring how many uninterrupted hours people actually get, rather than how many hours they are present, and they are scornful of open-plan offices designed by “the furniture police” to look tidy rather than to let people think.

Hire well, and keep people. Turnover is expensive in ways most companies never count, including the months it takes a new person to become fully productive. Companies with low turnover, the authors observe, tend to invest in their people’s growth and make clear that they expect them to stay.

Let teams jell. A jelled team, in their sense, is a group so closely bound that it achieves more than its members could separately. It has its own sense of identity, low turnover and a feeling of eliteness. The authors admit that jelling cannot be ordered. What managers can do is stop preventing it. They list the practices that kill teams, which they call “teamicide”: defensive management, bureaucracy, physical separation, fragmented time, quality reduction of the product, phony deadlines and clique control. Their advice is to give teams the chance of early, real success together, to protect them from interference and then to step aside.

The idea underneath

The book’s argument is finally about respect. Its methods are practical, but nearly all of them follow from treating knowledge workers as people whose judgement, concentration and pride in their work are the organisation’s main assets. Many of the practices that harm productivity, such as constant interruption, open offices, unrealistic deadlines and motivational posters, come from managers who do not trust their people and want to see them working. Much of what helps comes from leaving capable people alone to do work they care about.

It is also an argument about measurement. Managers measure what they can see, such as hours present, lines of code or tickets closed. DeMarco and Lister repeatedly show that what matters, including uninterrupted time, quality and the cohesion of a team, is harder to see, and that what is easy to measure can be pushed up while the real work gets worse. The site’s idea page When a Measure Becomes a Target traces the same problem in other settings.

Since it was published

Many of the book’s claims have been supported by later research. Studies of open-plan offices, including a widely cited study by Ethan Bernstein and Stephen Turban published in 2018, found that moving to open plans reduced face-to-face interaction rather than increasing it. Research on interruptions by Gloria Mark and others found that people take substantial time to return to an interrupted task. The authors’ emphasis on teams anticipated later work by Amy Edmondson and by Google’s Project Aristotle, which pointed to psychological safety as the strongest predictor of team effectiveness. The rise of remote work has changed the specific problem of the office, though not the general one of fragmented attention, which now arrives through chat and video calls rather than through a neighbour’s phone.

A first step

For one week, count how many blocks of at least an hour you and each person on your team spend without interruption. Then pick the single biggest source of interruption you control as a manager, whether a recurring meeting, a status request or an expectation of instant replies, and remove it.