I stopped measuring my mornings

For about three years I recorded when I woke up, how long I had slept, whether I had gone outside before nine, and how the first two hours of work had gone on a scale I made up. That is roughly a thousand rows, which felt like it ought to be enough to find something.

It was enough to find something. What I found was that the thing I had been optimising had almost no relationship to the thing I cared about.

What the data said

Wake time explained essentially nothing. A 6:10 start and an 8:20 start produced the same distribution of good mornings. Sleep duration mattered a little, and only below about six and a half hours, where everything got worse at once and stayed worse.

The variable that actually moved the number was whether I had decided the night before what I was going to work on. Mornings where the first decision was already made were good almost regardless of how they started. Mornings that began with so what am I doing today were mediocre almost regardless of how well I had slept.

Which is annoying, because it is not a routine

Routines are satisfying to build. You can buy things for a routine. You can tell people about a routine. "Write down tomorrow's first task before you close the laptop" is not a routine — it is a sentence, and it takes eleven seconds, and it does more than the alarm clock ever did.

I stopped logging in March. I do not miss it. The one habit that survived the spreadsheet is the eleven-second one, which is, I suspect, the general shape of how this goes for most people who try it.