If you’re a software developer it’s not unusual to have crunch periods where you’re working into the evenings and at weekends.

The practice of overworking developers is particularly bad in certain fields, such as the video-game industry, where working weeks can stretch to 100 hours as games near completion.

The issue with this workload is that evidence suggests working significantly more than 40 hours a week for a prolonged period actually makes you less productive, and there are plenty of developers on record about the mental and physical cost of sustained crunch periods.

Despite these issues, it is reportedly increasingly common for developers to work more than 40 hours a week in China, where a number of tech firms encourage employees to follow the ‘996’ work schedule, where staff work 12-hour days, six days a week. The trend has already triggered a backlash among some Chinese developers, who have started a GitHub page protesting the practice and its impact on the well-being of developers.

Now the effect of long working hours on developers has been raised once more by a new analysis of the Stack Overflow Developer Survey 2019, which finds no significant spike in job dissatisfaction among devs working 60 hours a week, with satisfaction levels not really dropping off until developers hit 80-hour weeks.

How the number of hours worked correlates with job satisfaction. Job satisfaction is shown by green, neutral feelings by peach, and dissatisfaction by blue.
Image: Stack Overflow / Count

The annual Stack Overflow survey is one of the most comprehensive snapshots of how programmers work, with this year’s poll being taken by almost 90,000 developers across the globe.

Commenting on the data, Robert Pozen, senior lecturer for technological innovation, entrepreneurship, and strategic Management at MIT Sloan School of Management, said although many “white-collar professionals” are content to work for longer than the standard 40-hour week, working hours can only be extended so far before it will negatively affect them.

“Many professionals are quite happy working 40 to 55 hours per week,” he says.

“But if professionals work for 70 to 80 hours per week on a regular basis, their productivity will gradually deteriorate on average. They will lose focus, and the long work hours will undermine the rest of their lives.

“Of course, professionals can have fruitful work spurts on projects they like or think are important. But that is the exception, rather than the rule.”

For developers, that fall in productivity is often mapped to an increase in poor quality and buggy code that will need to be fixed at some point, actually costing companies more in the long run.

It’s possible the Stack Overflow graph above reflects a willingness among developers to work around 55-hour weeks.

However it’s also likely the data is being skewed by how few survey respondents work these long hours, with the majority working about 40 hours a week and only a fraction working 60 hours or above.

The average hours worked by respondents to the Stack Overflow Developer Survey 2019.
Image: Stack Overflow

It may be the survey respondents who work these long hours are simply a small subset of developers who are happy to do so.

Julia Silge, data scientist for Stack Overflow, says a complicating factor in interpreting the data is that “some of the types of developers who work long hours are also among those who earn the most and also have the highest job satisfaction”. The Stack Overflow survey found these high earners typically reported being “very or highly satisfied” with their role, as was the case for 77% of senior executives, 70% of engineering managers, 69% of site reliability engineers, 68% of data scientists and 68% of DevOps specialists.

Confounding factors and the small sample size for those working above 60 hours a week make it is difficult to draw meaningful conclusions from the data without further analysis.

Altogether, this makes the connection between hours worked and satisfaction somewhat difficult to tease apart,” she says.

“You’d need some careful multivariable modelling to do this well, and look at what contributes the most (at what confidence level) to job satisfaction: salary, hours worked, type of coding work done etc.”

Silge says the data from this year’s Stack Overflow Developer Survey is publicly available for this kind of follow-up analysis and she’d be interested in seeing the results of such work.

For those curious about how average hours worked correlates to a developer’s primary programming language, here are the figures. Again, most developers work around 40 hour weeks.

The average number of hours worked by developers’ primary programming language.
Image: Stack Overflow / Count

Also see