There Is No Deep Work on a Ward, So I Stopped Trying to Build a Fortress

The core recommendation for concentration is to construct a period nothing can enter: no notifications, no meetings, a closed door. For a lot of jobs this is achievable and it is genuinely the best answer. For mine it is not available at any price, and for about three years I treated that as a personal failure rather than a description of the work.

Once I stopped trying to eliminate interruptions, a more useful question appeared: what does one actually cost, and which parts of that cost can I reduce? Because the cost is not the two minutes of the interruption. It is the re-entry, and re-entry is where the leverage is. A daily plain-text log offers one place to leave the context you will need on return.

I timed mine. Over four weeks I noted, roughly, how long after an interruption it took before I was properly back in what I had been doing. The median was about seven minutes, and the spread was enormous — some were near-instant and some were twenty-five. The difference was not the length of the interruption or how urgent it was. It was whether I had left myself anything to come back to.

That is the whole finding, and it is small enough to state in a sentence: an interruption to a task that is mid-thought costs three times an interruption to a task that has a written next step. So the practice is to always be leaving a next step, and specifically to leave it at the moment you are pulled away rather than trying to reconstruct it later.

Concretely, when someone says my name, I now finish the sentence I am writing — not the paragraph, the sentence — and type four or five words about what comes next before I look up. This costs about eight seconds and feels rude for roughly a week until you realise nobody has noticed. On return, those five words remove almost the entire reconstruction cost. My median re-entry went from about seven minutes to a bit under three across the following month, which over a shift with fifteen or so interruptions is roughly an hour.

The second change was in how I choose what to do with the fragments. The old advice implies fragmented time is useless for anything demanding, so save the hard things for a stretch that never comes. What I found instead is that fragments are fine for a specific kind of demanding work — anything where I am producing rather than deciding. Writing a handover, drafting a policy, working through a roster: all of these tolerate being cut in half. What does not tolerate it is anything requiring me to hold several options in mind at once, and those I now genuinely will not start unless I have a real block, which I get about twice a week.

The third thing, and the one that took longest, was accepting that some of my attention cost is not recoverable and should be planned around rather than fought. After a difficult event on the unit, I am not going to do careful work for the next hour, and pretending otherwise produced worse output and a longer recovery. Now that hour goes to things where being slightly diminished does not matter — restocking, filing, the kind of work that is genuinely just work.

None of this makes my job into one with deep work in it. What it does is stop me from spending my planning energy on a fortress I cannot build; time blocking around interruptions starts from the same constraint. The literature on attention is largely written by people who can close a door, and I do not think it is wrong; I think it addresses a situation and mine is a different one, and there is more written about the first than the second.