Sorting an hour while it is happening

Why it has to be this small

Because anything larger does not get done, and a method that does not get done produces nothing.

Every richer version of this has been tried by somebody and abandoned in the second week. Rating each hour out of five requires a judgement you do not have at four o'clock. Categorising by activity produces a list of activities, which you already knew. Writing a sentence per hour is fifteen minutes a day and feels like homework.

One letter is about three seconds. It survives a bad day, which is the only real test, and bad days are the ones the record most needs.

The one hard case, and the rule for it

The hard case is the hour that was pleasant. Lunch with three colleagues; twenty minutes in the kitchen; the walk to the station with somebody you like.

The rule is that pleasant is not a category. The question is whether the building was required, and for most of those hours the honest answer is yes — they could not have happened anywhere else. They are n, and they should be, and the record is more honest for it.

What they are not is a work argument dressed as one. If a quarter's n hours turn out to be mostly lunches and corridors, that is a finding, not an embarrassment: it says the building is doing something real and social rather than something operational, and that is worth knowing before anybody writes a policy about collaboration.

What a fortnight looks like

Ten days, seventy hours, about three minutes of effort in total. Two things usually emerge, and they emerged here.

The n hours cluster. They do not spread evenly across the days. On this record seventeen of the thirty-one fell on Wednesdays, which was the day most of the team happened to pick.

There are more w hours than expected. Nearly everybody who has kept this record is surprised by the fourth column, and the surprise is always in the same direction.

What to do with it afterwards

Keep it to yourself first, and read it before showing it to anybody. The point is not evidence for an argument; it is finding out what your own quarter actually contained, which most people cannot say.

If it does become evidence, the thing that makes it usable is that it is specific and small. "Seventeen of my thirty-one useful hours last quarter were on Wednesdays, and I was required in on Tuesdays" is a sentence a manager can act on. "The office does not work for me" is not.

Ten days of one record

The raw material, as it was actually written, for one week:

Mon  (home)   s s s s s s s
Tue  (office) n s s b s s w
Wed  (office) n n b s s s w
Thu  (office) s s s w s s s
Fri  (home)   s s b s s s s

Five lines. The thing that jumps out of even one week is the fourth column, which appears three times in three office days and not once in the two days at home.

The two mistakes people make in the first week

Classifying the day instead of the hour. The whole method depends on the hour being the unit; a day is too coarse and the answer comes out as a mood. If you find yourself writing the same letter seven times without thinking, you are rating the day.

Rewriting the morning in the afternoon. The letter has to be written at the end of the hour it describes. Reconstructed at five o'clock, the morning is remembered through whatever the afternoon was like, and the error runs in the direction of whatever you currently feel about the office.

When to break the rule about not reflecting

Once, at the end of the fortnight, for about ten minutes.

Look at the n hours and ask what each of them needed. The answers come out as a short list — a specific person, something to stand in front of, an afternoon that was not full — and that list is more useful than the totals, because it is what you would have to arrange to get more of them.

On this record the list came to four items and they are the four conditions set out elsewhere on this site. They were not designed; they were what was left after reading thirty-one hours and asking what each had required.