Deep Work and time blocking: what Newport actually argues

The argument, in one paragraph

Deep Work (2016) argues that sustained concentration is becoming rarer and more valuable at the same time, and that the work requiring it gets crowded out by default rather than by choice. Shallow work is legible: it arrives in an inbox, carries someone else's name, and produces visible motion. Deep work produces the things careers are actually built on and, left unscheduled, reliably loses. Newport's conclusion is structural rather than motivational. You do not defeat this with willpower; you defeat it by deciding in advance what every hour is for.

How Newport's time blocking works

His method is specific enough to try today. At the start of each working day, divide the day into blocks and give every block a job, including the shallow ones, so email gets a block instead of leaking into all of them. When the day breaks the plan, and it will, redraw the remaining blocks rather than abandoning the exercise. The plan is not a prediction that failed; it is a decision you keep remaking. He pairs this with a depth budget: decide how many hours of genuinely deep work a week you can sustain (most people plateau near four a day) and treat those hours as the scarce resource the rest of the schedule serves.

Two details are usually lost in summaries. First, the blocks are coarse on purpose: half-hour granularity invites fiddling, and fiddling is itself a form of shallow work. Second, the redrawing matters more than the drawing. Newport treats the daily replan as the skill being trained: an honest accounting of where the time went, made while there is still day left to spend.

Where the book disagrees with automatic scheduling

Newport wants you to do the blocking yourself, by hand, every morning, because the deciding is the discipline. Outsourcing the decision gives up the part that changes your behaviour. On his account, an auto-scheduler is a labour-saving device for the one labour that was doing you good. Diarized is exactly the thing this paragraph warns about, and we would rather say so than pretend the book blesses us.

There is a second, sharper conflict. Diarized books whatever duration your task app declares, so a day of half-hour estimates becomes a day chopped into half-hour pieces, the fragmented schedule the book is written against. Nothing in our scheduler will infer that your work needed three unbroken hours; if it does, the estimate has to say so. Where we think the trade runs our way: the calendar Newport redraws by hand each morning assumes a day mostly yours to redraw. A calendar full of other people's meetings changes the arithmetic, and recomputing the gaps around twenty meetings is exactly the kind of work software should do and people should not.