scvalex.net

43. Deep Work

I recently finished reading Deep Work by Cal Newport. It was an interesting, if personally depressing, read. My takeaways were a practical definition of deep work and an obvious-in-hindsight observations on what prevents deep work.

My reading is that deep work is work that requires all of our focus. It's the best work we can do. It's work that pushes us to our intellectual limits.

If we accept this definition, then we should clearly strive to do as much deep work as we can since this is the only time we use ourselves to our fullest. Put differently, why should we spend our lives on trivialities when we could be spending it on big things instead? What exactly constitutes a big thing is a personal matter — it's probably not "world peace" for most of us, but it could be "write a book", "write a game", or "further scientific research".

This definition also hints at a simple heuristic to distinguish deep work from the rest: How many months would it take to train a reasonably smart college grad (e.g. our past selves) to do the work? If the answer is a small number, then the work is probably not deep.

Interestingly, the depth of work is not directly related to how tiring or how time-consuming it is. For instance, acting as a human search engine in person or over email is tiring. Being a bystander in meetings is time-consuming. Manually doing automatable tasks is both. But none of these require the best of us, so none of these are deep work.

A point raised by the book is that research suggests we are only capable of about four hours of deep work in a day. On one hand, this means there is time for all the administrivia surrounding work, but on the other, it means we should try to max out our deep work everyday because catching-up later isn't really possible.

Google's Site Reliability Engineering book approaches this problem from the opposite direction and defines "toil" which loosely maps to non-deep work for programmers. The book goes into more detail, but toil is roughly repetitive work of no enduring value. In other words, if there is repeated work to be done on a system, but the system is left in the same state afterwards, then the work is probably toil. The book claims that Google SREs spend no more than 50% of their time on toil. Coincidentally, that's about four hours in a regular working day, leaving the remaining four hours for deep work.

Deep Work points to interruptions as the enemy of deep work. This makes sense since deep work requires focus, and it takes more time to get into a focused state than to be broken out of it by an interruption. Personally, I've found that even the threat of an interruption is enough to prevent me from focusing properly. For instance, why even bother trying to get into solving a problem if there's a meeting in 20 minutes or if there's some running job to check in on in 5 minutes?

The SRE book points to toil as the enemy of interesting engineering work. In my experience, it's easy to get lost in the busywork. Sometimes, even when I recognize that what I'm doing is repetitive and could be automated, the end frequently seems close, so it feels like it would be a shame to discard the progress I've made and start from scratch in a more automated way. Other times, the repetitive work has been going on for so long, that it's hard to imagine a world without it.

When I became aware of deep work and toil, I found I was spending almost all of my time on toil and interruptions. The two would feed into each other: if there are only 15 minutes until the next interruption then the only possible work doable in that time is that little bit of repetitive work that's been popping up for weeks.

The steps I took to minimize interruptions were to turn off audible notifications, only read news and Reddit before 09:30 in the morning, read my email in batches three times a day, and pack all of my meetings into two days a week. As for toil, I started booking blocks of time in my calendar just to think about how things could be improved, and I've committed to fixing a bit of toil regularly. If nothing else, these changes have definitely made me happier.