Jeff Huang schedules everything on his calendar, even work that isn’t time-bound:
I put everything on this calendar, even things that aren’t actually for a fixed time like “make a coffee table at the workshop” or “figure out how to recruit new PhD students” - I’ll schedule them on a date when I want to think about it.
The forcing function is visibility. When someone asks for your afternoon and you’ve already blocked “architect search redesign,” you have language for no. But Huang also notes it took years to accurately estimate what fits in a day. Calendars expose estimation failures rather than fix them.
Kahneman’s planning fallacy shows people systematically underestimate how long tasks take, even with historical evidence. Scheduling forces you to see that refactoring won’t fit alongside three meetings and code review. You’ll still underestimate the duration, but at least you’ve acknowledged the work exists and needs time.
The harder problem is organizational. Engineers need to be available to unblock people and make decisions, but also do deep work that requires sustained focus. Your calendar can say “architecture review 2-5pm” but if your team is blocked on a database migration, that block evaporates. Protecting your time requires organizational capital to enforce boundaries when they conflict with immediate needs.
What You Can Control Link to heading
Interruptions are inevitable. Attention residue means each context switch degrades performance on both tasks-part of your attention stays anchored to the interrupted work. Leroy’s research tested a mitigation with “ready-to-resume” plans where you note where you were and what’s next before switching. For engineers, this means leaving a comment like “was verifying exception paths in payment processor lines 145-203, check error propagation next.” That note reconstructs context faster than scrolling through files trying to remember your mental state. It has been highly effective for me.
The other controllable variable is break timing. Research on systematic breaks found predetermined breaks (24 minutes work, 6 minutes rest) produced higher concentration and lower fatigue than self-regulated breaks. Past about 60 minutes, review effectiveness falls off. The calendar should encode both focus time and recovery, not assume you’ll power through.
What You Can’t Link to heading
I like this quote from Kevin Kelly:
“Your work will be endless, but your time is finite. You cannot limit the work so you must limit your time.”
But limiting your time often means limiting your impact. When your job is unblocking others, your time becomes a shared resource. The calendar may say “deep work 2-5pm” but the team needs a decision at 2:30pm. You can protect the block and delay the team, or abandon it and fragment your work.
The orgs I’ve seen handle this best treat certain blocks as truly protected (booked as “Busy / Do not Schedule” explicitly marked in team norms) and others as “prefer not to interrupt but will flex” or “Focusing / Reachout before scheduling”. Making that distinction explicit helps, but it also requires broader buy-in that focus time is valuable, not just nice-to-have that gets lip service.
Huang’s pattern-breaking work into what fits in a day and moving the rest-only works if protected blocks actually survive. Otherwise you’re optimizing individual time allocation in a system where allocation is determined by whoever has the most urgent interrupt. The calendar techniques assume you control your time. The job assumes your time is a shared resource. One of those assumptions has to give.