When I was at The Economist, there was a policy to rotate everyone every three years. The idea was that fresh eyes were more important than experience. "Foreign everywhere" was the mantra, and around your second year in Cairo, you could expect to get a call from the editor asking you to consider Mumbai or Sao Paolo--ideally two places you'd never been to and knew nothing about.
The first year after arriving to your new assignment was terrifying and exhilarating. It was a vertiginous learning curve, but you could ask dumb questions without fear and note that the emperor has no clothes.
In the second year, after the emperor had invited you in a few times to explain the subtle political dynamics that require him to go garbless for the ultimate good of the nation (but surely there were more important things to write about, such as his new elevated rail project), you would find yourself writing sophisticated analyses, traveling easily through the region, admiring your bulging rolodex and otherwise feeling very productive.
In the third year, you'd find yourself returning to stories with a certain cynicism and worldweary accounting of endless process. The elevated rail project has been delayed once again because of infighting within the opposition party. The emperor has no fiscal discipline. You understand everything all too well. It's time to move on.
For the first ten years of my career, I changed jobs every three years. Then, for the seven years I was at The Economist, I changed countries every three years (London, Hong Kong, and New York, although sadly not long enough at the last). Here at Wired, I seem to have achieved the same rhythm by publishing a book after my fifth year and, next summer, my eighth. Each time it changes my life and puts me back on a steep learning curve with a new subject to immerse in and a new pace of travel and speaking. I've got a new foreign land to explore.
I was thinking about the three-year rule while reading about Malcolm Gladwell's observation that it takes 10,000 hours to become truly expert at something. If you really throw yourself into a job, you'll spend 60 hours a week working. That's 3,000 hours a year (allowing for vacation), which means you'll hit the 10,000 hour mark a few months after your third year.
So maybe that's where the three-year rule comes from. You're now expert at what you set out to master. Great. Now go do something else.