The removals nobody proposed

How a thing stops without being stopped

The mechanism is straightforward once it is named. A procedure does not exist in the abstract. It exists in a slot: a time of the week, a person's habit, a gap between two other things. Remove a day from the week and some of those slots stop existing, and whatever lived in them stops with them.

Four of the seven lived in Friday afternoon. The workshop clean-down, the diary check for the following week, the colour-coded wall planner being brought up to date, and the informal half hour in which the week's refused jobs used to get talked about. Friday afternoon was the day that went. Nobody discussed any of the four.

The other three lived in slack rather than in a slot: the second quote on small purchases, the informal shadowing of new starters, and the re-sending of the rota as a reminder. All three were things somebody did when there was a moment. There stopped being a moment.

Why this matters more than it sounds

An unproposed removal has three properties, and all three are bad.

1. It is not on the list unless somebody goes looking. These seven were added to the record retrospectively, four of them more than a month after they had stopped, when somebody noticed they were not happening.
2. Nobody argued about it, by construction. There was no meeting. This puts every one of them straight into the category that the rest of this site identifies as the slow one.
3. It has no owner, because nobody decided it. If a proposed removal turns out badly there is a person who proposed it and a discussion to reopen. If an unproposed one turns out badly there is nothing to reopen.

Of the seven, two surfaced later as problems: the shadowing of new starters, at a lag of forty-seven weeks, and the log of refused jobs, at sixty-four weeks — the two slowest on the entire record.

That is the sharpest result on this site

Seven entries is a small number and two is a smaller one. But the two slowest failures on a list of thirty-eight both came from the seven that nobody decided, and both came from the group that carries knowledge between people.

The mechanism stacks. A thing that nobody proposed is a thing nobody argued about, which is a thing nobody is watching. If it also happens to be the kind of thing whose failure is invisible, there is no channel at all through which the organisation can find out, and the lag becomes whatever the lag of the underlying risk happens to be.

The two removals that cost the most were not removals. They were absences that nobody had noticed were absences.

How to catch them

The practical answer is a question asked once, a month after the transition finishes, and it is not the question anybody expects.

Not: what have we stopped doing? People answer that with the list they already have.

Instead: what did you used to do on the day that went?

Asked of each team, that question produced five of the seven on this record, retrospectively. The other two came up much later. It works because it attaches the question to a slot rather than to a decision, and the unproposed removals are exactly the ones that lived in a slot rather than in a decision.

A second, narrower version is worth asking of anybody who has been there less than a year: who showed you how to do this? If the answer is a procedure that no longer runs, you have found a knowledge-carrying removal that nobody logged.

What this does not mean

None of this says the seven ought to have survived. Five of the seven are among the fourteen that nobody has asked for, which is a better rate than the list as a whole.

Things that live in slack are, on average, things the organisation can do without — that is roughly what it means to live in slack. The problem is not that they stopped. The problem is that they stopped without being recorded, so that the two which mattered had no entry to be read back, no owner to ask, and nobody expecting anything.

Write them down afterwards and they become ordinary rows on an ordinary list. Leave them off and the list is not a record of the transition; it is a record of the parts of the transition that were discussed.

The seven, listed

Set out plainly, because they are the category most likely to exist unrecorded in anybody else's transition.

Four in one afternoon and three in the gaps. Three were never missed, two drifted back, and two are the slowest failures on the record.

The spread is the point: these are not systematically worse removals than the proposed ones. They are removals with no paperwork, and the two that mattered had nothing attached to them that anybody could have read back.