I recently read the book Slack: Getting Past Burnout, Busywork, and the Myth of Total Efficiency. My previous post on Working Smarter, Not Harder already discussed the main premise of the book. However, DeMarco writes compelling arguments on other issues, one of which is overtime.
Much analysis has been done to determine the correlation between software function points / lines of code and work effort. This analysis has shown that using workdays is more reliable as an indicator/predictor than using workhours. The result is that "overtime is explicitly ignored in projecting effort required to perform new work" (Slack, page 64). This means that on average, working overtime does not mean more is accomplished than working a normal work day. DeMarco identifies four reasons why overtime hurts enough to offset the effect of the added hours:
- Reduced quality
- Personnel burnout
- Increased turnover of staff
- Ineffective use of time during normal hours
(See Slack, pages 65-70 for further elaboration on these four points.)
This analysis matches my personal experience. When doing lots of design & coding, I am often mentally exhausted after 7 or 8 hours - I'm no longer at my 'A' game: I get slower, make more errors, and take longer to figure things out. I can sometimes push myself to go as long as 10 hours, but usually that only works when I've had meetings or other non-mentally-draining activities to break up my day. And I certainly don't try to do that more than one day in a row. Why not? Because I've experimented with working 9-10 hours for multiple days in a row, and I don't find it to be a maintainable speed.
Working regular overtime (more than a week or two) has a significant negative impact on a person. Factors such as increased stress, strains on family and personal relationships, less sleep, and less exercise all add up over time to affect an individual's mental, emotional and physical health. Over time, this can result in increased physical illnesses, lack of motivation, and eventually complete burnout. During my university years at a summer manual labour job, one supervisor who had been working extreme amounts of overtime (70+ hours per week for several months) had a nervous breakdown and had to be hospitalized.
This ties in to DeMarco's discussion on turnover, which I found quite interesting. He makes the point that turnover has a cost in terms of hiring & training, and this 'human capital' cost is seldom explicitly measured and tracked. A manufactoring company makes capital investments in factories and equipment in order to produce goods, and these investments are carefully tracked. Knowledge companies invest time & effort in hiring & training employees, but how often is this tracked? A quick test for managers: what is the annual turnover rate for your company / department, and how does it compare with the average for the industry? How much does this turnover cost your company? Think about this before continuing with the next paragraph. If you don't know, then try making a guess.
DeMarco mentions studies which show that "companies whose turnover places them in the best third of the sample are experiencing less than half the turnover loss of those in the worst third" (Slack, pages 67-68). The annual turnover rate was measured at 35% in the computer field as of 1990. An estimated 20% of the average company's total labor expense is turnover cost. (Steve McConnell. 1996. Rapid Development, page 293). I'm interested in seeing more recent statistics on this - let me know if you come across any.
I feel that overtime is harmful, both to the workers engaged in it, and to the company itself - especially regular overtime. As a call to action, just say no to overtime and make our industry a better place to work.
(The title of this post is adapted from the title of a famous article titled Goto Considered Harmful by Edsger W. Dijkstra.)
If you find this article helpful, please make a donation.