I'm trying to describe the costs of multitasking. Here's what I've got so far:
There are three parts to multitasking:
- Stopping the work you're doing. The stopping cost is the time it takes to mark your place, save your work, etc. You haven't stopped thinking about what you're doing, but when you stop to take a phone call or answer a question, there's a stopping cost. If you're in flow, this is surprisingly high.
- Swapping out what you're working on. The swapping out is the act of clearing your mind of the work you'd been doing so you have room to swap in the new work. If you were in flow or concentrating deeply, this can take anywhere from 5 minutes to 30 minutes. Sometimes, it takes me even longer.
- Swapping in the new task. The swapping in depends on the complexity of the work and how long it's been since you last touched the task. The more complex and the longer the time since you last touched the task, and the more people you have to talk to, the longer it takes.
I don't know how to give ballparks for each of these. Certainly, for some tasks, it's fairly trivial. If I'm organizing a normal weekday dinner, my swapping in/out is very fast, because there's little knowledge associated with each task. But now when I write chapters of a book (or back when I was writing code,) the costs can be very high, because the knowledge in my head is not yet written down. For me, the stopping the work is defect-inducing. Unplanned interruptions help me make defects. So does the swapping back in, if it's been a long time since I last worked on this task.
Did I miss anything?
I find that the biggest delays in multi-tasking involve waiting for others who are also multi-tasking. In the strictest sense, doing many individual tasks simultaneously takes longer than doing them serially. However, we do not work in isolation. When we are dependent on others who are also multi-tasking, coordination becomes very difficult and often impossible without considerable delays.
Michael, you are exactly right. The time wasted pushes all the projects out.
Shawn, thanks that entry is exactly on target.
You are right, multi-tasking certainly destroys the level of focus we can have on any one project. The cost of these defects can vary, but I agree that it is there.
One additional thought. If you were to diagram out in blocks, 3 or 4 tasks or projects, you would see that to multi-task these four projects actually pushes the completion time out on all of these projects. For example, if each project takes 1 day, we would expect that by the end of the first day, project one would be done. By the end of the second day, project 2 would be done…so forth. However, if I am forced to break into project 1 (halfway) and start project 2, then break into project 2 halfway and start project three, then halfway through project three, go back to project 1, you can see that it actually would take me until the end of the 2nd day to complete project 1. The others get pushed back as well.
So, it would seem that multi-tasking actually delays completion dates on projects or tasks.
Am I right?
I read this blog entry yesterday and it seemed to fit with this discussion:
http://www.agileadvice.com/archives/2006/11/how_two_hours_c.html
Before starting another task, often times you must spend some amount of time deciding whether or not you should switch over to the new task.
Swapping out, can include a physical context as well. Perhaps moving to a conference room, driving somewhere. (As a software developer, reconfiguring a machine to work on another branch or configuration of a software project)
It’s already in your list, but I would suggest that the multiplicative cost of the multitasking delays in an organization are the real killer. Working alone, it is harder to see the costs.
I think you’re missing one very important thing: yes, cost of switching the context is signifacnt, but there are situations when cost of refusing to swotch is much, much bigger.
While I always try to remember about disturbing the flow, swapping to a new task, etc, I never look at swapping task from that perspective only.
Here’s some more about that.
I also experience a higher order cost when multi-tasking too much. I see the cost I’m expending on task switching and get frustrated. This frustration itself costs time and/or defects. (If I’m centred, I’ll take time out to calm down. If I’m not centred, I’ll plough forwards and make mistakes. Either way, I’ll spend time simply being frustrated.)
This frustration is exacerbated by the fact that multi-tasking happens most frequently when I have a lot of work to do… (Could probably draw an interesting diagram of effects about this.)
The phrase “too much” is an important qualifier. Different people seem to have different levels of task switching that is tolerable or even enjoyable. Some people like to get into a task and follow it through; others like to vary their day a bit. I’ve learnt to tolerate more task switching as I’ve got older.
I think this tolerance level is an important part of the issue. Senior managers often have fairly high tolerance levels for task switching – it’s in the nature of their job that people bring them lots of different problems to address. So their frustration cost for task switching is low, and they’ll probably have learned to manage down the other costs too. That means they may not be fully aware of the costs they’re imposing on other people.
Shawn, Johanna
In the unlikely event you didn’t see this rebuttal of the Agile Advice article by Joel – heres the link http://www.joelonsoftware.com/items/2006/11/15.html
It rather eloquently backs up Pawel’s point of making sure that the interrupting task *really* is safe to ignore. But hey, that’s what we have project managers for…!
I think there’s another factor that can impact the length of time needed for a task switch – the risk to the project. I can switch between two tasks relatively quickly if I know at least one of them is low risk, since I don’t feel a great need to consider *everything* about the task. But if I know there is a significant risk – political, technical ramificiations, monetary – then I need to take more time making sure that I’m up to speed on all the potential trouble spots.
I’m thinking particularly of times when I’m in the middle of something technically involved and someone comes to ask me a question about something wildly different. If the question is something low risk like – “where would I find the materials for project X? will they help with question Y?” – then I can pop out with an answer pretty fast, since I don’t need to think about it too much. After all, if I get the thought wrong on a low-risk problem, there’s not much to loose. But if the question is high risk – like dealing with a customer appeasement issue, or trying to fix a serious problem on an old project – then I’ll have to take a significant amount of time to get into the problem and come up with a solution that addresses the risk factors.
I find it also is a factor of which coworker I’m working with – some I know I can trust to double check an “off the top of my head” sort of answer, so they make the risk of a bad answer much lower. Some will just take my word for it, and I had better be right the first time.
By the way, just a phrasing detail: it’s not the cost of swapping out, it’s the cost of swapping IN again when you get back to it.
If I’m interrupted in the middle of writing complicated code (one of the worst cases), I can turn my attention to the interruption within moments. LATER I spend 30 minutes getting my mind BACK into the code before I can be productive again.
— Michael Chermside
As a general rule I totally agree with what you wrote.
There are odd cases though, when you are stuck on something (I had these moments when I wrote code or debugging for example) where it is actually good to swap to another task. When you switch back many times you just see the (sometimes obvious) solution that was alluding you.
Also, there are cases where you are stuck because of dependencies on others stuff/individuals, again, it is good to switch to some other useful task (that why it is good to have a prioritized list of things to do) instead of just waiting.
Arnon
Multitasking – Now I started thinking about it, and I have to sort out these thoughts.
Swap Costs (Stop, out, in, and back)
Well described so far. Can be mitigated by planning. I have a cleaning stunt that is very effective, where I do ten minutes a room/location, over multiple locations and task natures. Because I know I am doing this I can lower the swap costs and get great benefits from the energy I gain from the process.
Opportunity costs (Staying on task, not staying on task, value of other task.)
-Value of other task introduces trade-off calculations. (Couch just caught on fire. Phone rings, but you have an answering machine.)
-Work flow energies (additive, subtractive) I get re-energized by tackling something new. Fresh eyes actually pick-up speed when I return sometimes. Some tasks are actually quite draining and you slow down after a while (e-mail inbox catch-up). Is the energy gained or lost greater than the swap cost? Frequently my processor goes idle on a particular task. If I’m self-aware I know this and another task can be welcome.
-With effort, multi-tasking swap costs can be mitigated, either through training, planning, or better tools. Having a bookmark handy matters when interrupted while reading a book. How about a multi-tasking request queue in your office? When your processor is idle you can easily swap in items in the queue and handle them. (Works best if queue is not visible until you attend to it.)
-Someone noted, the better you know your employees, the better the utility of the response to their interruptions. But it takes extra time to understand how your employees function, especially in multi-tasking interactions.
Interface costs – delays if your multi-tasking depends on others. I think this is a meta category. It impacts both other categories.
-Known interruptions versus unknown interruptions (varying time of “handshake”).
-Joel’s A and B tasks with a processor ( http://www.joelonsoftware.com/articles/fog0000000022.html ) is a great example. If average times go up you run in to interface problems. In his first sequential processing example. Finishing the A task first lets A get a running start on whatever other project they are on. B has to wait the same amount of time in either case. Therefore having a multi-tasking request queue handy for B would help enormously in this case.
-I can multi-task for example, but if the average time of all tasks goes up and I miss getting a package to the mailman, I have to drive to the post office later instead. That throws off all later tasks, and induces a possible cascade of interface costs.
I see an optimization diagram in here. I’m not drawing it, too many other things to do now, but I can see it.
I would like to take a slightly different perspective, based on my personal experience. When we are in Âflow we ARE capable of managing multiple tasks, which do fit into THIS flow. I remember it very clearly. When IÂm in flow I can edit my code, compile, send e-mails, and even play Solitaire smoothly as long as it does fit into definition of MY flow. The real problem is to get into this ÂflowÂ. As long as IÂm not IN I use virtually ANY excuse not to get into: interrupts, phone calls, meetings, e-mails, phone calls, coffee, and bath room. Whatever you want, just not to get into the ÂflowÂ! Why itÂs so I do not know exactly, perhaps itÂs a problem of self esteem, perhaps itÂs just love, one has to forget about her/him self. When we are in the Âflow we do not pay a shit to the rest of the world. But in order to get into it, we have to abandon ourselves, and thatÂs not easy. Then we are looking for any kind of excuses, we cal interrupts.
Asher
Paul / Pawel:
Thank you for showing us a good use of Critical Thinking and Cognition skills. Considering the opportunity cost can not be ignored and that consideration IS the manager’s job.
One does have to consider reputation risk and relationship risk depending on the situation though. Should one postpone a 2-week iteration, a deliverable, a commitment, etc. for one customer because another more lucrative customer has a need? What happens to the moral of a team that is context switching because “management can’t seem to prioritize anything”?
I think the overall emphasis should be that context switching is not efficient as a standard rule of practice but, as with any rule, there are reasonable exceptions.
Somewhat related.. I stumbled upon Microsoft research on Effects of Instant Messaging Interruptions on Computing Tasks (PDF). Among the findings was that the relevance of interruption affects the cost of the interruption. More related = less costly.
I certainly agree that multitasking is just a killer of efficiency.
As a matter of fact, we have a dual core brain (left/ right hemisphere) which is also multiprocessor (spine, reptilian, limbic, cortex), but it is not meant for real multitasking: it is rather designed to handle background functions while applying multi-angle analysis on one only conscious task at hand.
In our world though, it is more and more difficult to resist the urge of multitasking. Personal organization methods like GTD can help regain productivity by splitting wider tasks in elementary actions that will be focused on, one at a time.
I develop this theme on “GTD vs Multitasking”
http://web.mac.com/ericlodi/iWeb/Orgameth/
And if you like to use the full power of your duo-core (left/right) brain, there are more posts to implement GTD on mind map!
Happy GTD and stay away from multitasking!