Imagine yourself performing two tasks at a time. Let’s say you are multitasking squeezing oranges and making an argument over a phone.
You think it would take the same time if you would do this simultaneously or sequential.
The thing is, it’s not. Your brain can only really focus on one task at a time.
The switching time costs you. And it costs you badly.
Because multitasking essentially is performing two tasks sequential and continuously switching between them.
Focussing on multiple things at the same time is impossible.
Every time you switch, you experience what one could call ‘switch time’.
Don’t waste time multitasking.
Go for sequential processing from now.
First choose what you are going to focus on.
Then do it. Finish it.
Software Engineer Joel Spolsky was one of the first to write about this in his book ‘Joel on Software’. It was in the year 2004.
Suppose you have two separate computations to perform, A and B. Each computation requires 10 seconds of CPU time. You have one CPU that, for the sake of this problem, doesn’t have anything else in the queue.
On our CPU, multitasking is optional. So you can either do these computations one after the other by using sequential processing:
Let the results speak for itself.
If you multitask, on this particular CPU, tasks run for 1 second at a time, and a task switch takes no time at all.
Which would you rather do? Most people’s gut reaction is that multitasking is better. In both cases, you have to wait 20 seconds to get both of your answers. But think about how long it takes to get the results to each computation.
In both cases, the results of Computation B (shown in black) take 20 seconds to arrive. But look at Computation A. With multitasking, its results take 19 seconds to arrive… yet with sequential processing they already are in only 10 seconds.
In other words, in this nice contrived example, the average time per computation is lower (15 seconds rather than 19,5 seconds) when you do sequential processing rather than multitasking. (Actually, it’s not such a contrived example-it’s baseed on a real problem Jared had to solve at work.)
If you are not convinced yet, try this exercise with pen and paper. Use a timer on your mobile phone.
First attempt
- Start timer
- Write 1-10
- Write A-J below the 1-10
- Stop timer
Second attempt
- Start timer
- Write 1 – A – 2 – B – 3 – C … – 10 – J
- Stop timer
Please be so kind to e-mail your findings to blog@reiniervaneijk.nl, I’m thrilled to learn how you did.
Sequential processing
Stel je voor dat je twee taken tegelijkertijd uitvoert. Stel dat je multitaskend sinaasappels perst en een discussie voert over een telefoon.
Je denkt dat het dezelfde tijd zou kosten als je dit tegelijkertijd of opeenvolgend zou doen.
Het ding is, het is niet zo. Je brein kan zich maar op één taak tegelijk concentreren.
Switchtijd nekt je. En het nekt je behoorlijk.
Omdat multitasken in essentie is twee taken na elkaar doen en continu switchen tussen de taken.
Aan twee dingen tegelijkertijd aandacht geven is onmogelijk.
Elke keer dat je switcht, ervaar je wat je ‘switchtijd’ zou kunnen noemen.
Verspil geen tijd aan multitasken.
Gebruik vanaf nu sequential processing.
Kies eerst waarop je je wilt concentreren
Doe het dan. Maak het af.
Software Engineer Joel Spolsky was een van de eersten die hierover schreef in zijn book ‘Joel on Software’. Het was in het jaar 2004.
Suppose you have two separate computations to perform, A and B. Each computation requires 10 seconds of CPU time. You have one CPU that, for the sake of this problem, doesn’t have anything else in the queue.
On our CPU, multitasking is optional. So you can either do these computations one after the other by using sequential processing:
Ik laat de resultaten voor zich spreken.
If you multitask, on this particular CPU, tasks run for 1 second at a time, and a task switch takes no time at all.
Which would you rather do? Most people’s gut reaction is that multitasking is better. In both cases, you have to wait 20 seconds to get both of your answers. But think about how long it takes to get the results to each computation.
In both cases, the results of Computation B (shown in black) take 20 seconds to arrive. But look at Computation A. With multitasking, its results take 19 seconds to arrive… yet with sequential processing they already are in only 10 seconds.
In other words, in this nice contrived example, the average time per computation is lower (15 seconds rather than 19,5 seconds) when you do sequential processing rather than multitasking. (Actually, it’s not such a contrived example-it’s baseed on a real problem Jared had to solve at work.)
Als je nog niet overtuigd bent, probeer dan de onderstaande oefening met pen en papier. Gebruik een timer op je mobiele telefoon.
Poging 1
- Start timer
- Schrijf 1-10
- Schrijf onder 1-10 blokletters A-J
- Stop timer
Poging 2
- Start timer
- Schrijf 1 – A – 2 – B – 3 – C … – 10 – J
- Stop timer
Stuur je bevindingen naar blog@reiniervaneijk.nl. Ik ben nieuwsgierig hoe je het er vanaf hebt gebracht.