Note that the author partly justifies it by the writing she produced later, which wouldn't have happened without her "unproductive" time.

Applied to programming, it seems to me the author is against measuring lines of code per day, and in favor of going for long walks and letting solutions pop into your head that don't require so much code. Or longer walks that cause you to realize the whole program is bunk and you ought to be building something different. It's pretty difficult to measure the productivity of this sort of thing (except perhaps after a lot of hindsight), but it can have a huge effect on your long-term results.

Richard Feynmann wrote about this. He got burned out trying to be scientifically productive, started screwing around, got curious about the physics of something trivial, and ended up figuring out something important.