6/11/2008

The penalty for task switching

In today's demanding business world we are expecting more and more out of our employees. They should be cross-trained and should be "wearing many hats" and handling many tasks. There is a point in which multitasking (the habit of working on multiple projects at any given time) takes a toll on productivity.

Your intuition tells you this is true but to what extent does it compute into reality? Let's examine the most efficient of task handlers: the computer.

If two tasks take 10 seconds each to computer then executing them sequentially would look like this:

Process 10 seconds of task 1
Process 10 seconds of task 2

If we decide to multi-task them, the outcome would be as follows:

Process 1 second of task 1
Process 1 second of task 2
Process 1 second of task 1
etc...

In an ideal world, each task would consume the same 10 seconds of processing time, although as you might have observed, Task #1 was completed within 10 seconds and Task #2 though completed in 10 seconds time, had to wait for task 1 to complete. If you were able to prioritize task 1 over task 2 then sequential process let you complete it much sooner.

Lets take another look at the multi-processing timeline for tasks 1 and 2:

Time: __1__2__3__4__5__6__7__8__9_10_11_12_13_14_15_16_17_18_19_20
Task: __1__2__1__2__1__2__1__2__1__2__1__2__1__2__1__2__1__2__1__2

Task 1 will complete after 19 seconds after it began, and task 2 will complete after 20 seconds.

Now consider this: in order for the processor to switch between task 1 and task 2 it needs to save the data in memory for the first one and load up task 2. If task switching only takes half a second then the total time from start to finish for task #1 is 19 seconds + 18 * 0.5 second = 28 seconds and task #2 will take 29.5 seconds.

What about human task switching?
A friend of mine, Jochen Krebs, asked me to review his new book to be released July 2008 on Agile project management (highly recommended, buy it on Amazon). Jochen references to human task switching penalty as he discusses resource utilization in project portfolio management. Research has shown that humans spend on average 15% of their time switching between tasks when they are multi-tasking. To use the numbers in the above example, a project where key resources are shared (multi-tasking), will take more than twice the amount of time to complete, in fact at least 115% more in order to accommodate for task-switching. In fact on average, task switching costs 20 minutes to 1 hour a day of "mental time" needed to refocus on the task at hand.

Gurus in space of project management have come to the conclusion that it is better to have one person working less than maximum capacity than 4 people interrupted daily.

Should we never multi-task?

Studies have shown that productivity is significantly impacted when multitasking more than 2 activities. It is for this very reason that organizations that adopt a rapid delivery methodology such as Agile - stay away from sharing resources across multiple projects. Merely the multitasking that occurs in our regular work activity is enough to take a toll on our productivity. To compound that with inherently different roles (projects) is a direct shot at productivity.


If you want to read more on this subject, here are a few recommended books:

Critical Chain by Eliyahu M. Goldratt
Slack: Getting Past Burnout, Busywork... by Tom DeMarco
Peopleware: Productive Projects and Teams by Tom DeMarco and Timothy Lister

1 comment:

Clay said...

Great Article! But as I read it I was also proofing a pretty boring proposal at the same time. I've always been wild and crazy like that. Great Blog Tiran. Keep up the good work!