What the list cost to keep

What it actually was

A spreadsheet. Five columns:

1. What stopped.
2. The week it stopped.
3. Whether anybody said anything at the time — argued, asked, or nothing.
4. What became of it.
5. A sentence, where there was one worth writing.

The first three columns were filled in as the transition ran, in the same meeting where the removal was agreed, taking a minute or two each. That is about an hour in total across eleven weeks.

The fourth column was filled in on three occasions afterwards: at roughly six months, at a year, and at eighteen months. Each pass took under an hour, because most rows did not change.

Why almost nobody keeps one

Not cost. Three hours is nothing against a transition that occupied a quarter of an organisation's attention for three months.

The reasons are structural and they are worth naming because each one has a cheap answer.

The list has no owner after the transition ends. Whoever coordinates a change is reassigned when the change is done. The list belongs to a project that no longer exists. On this record it survived because one person kept it in their own files and happened to still be there.

Filling in the fourth column is not anybody's job and does not feel urgent. Nothing goes wrong if it is not done. It is the purest form of work whose absence is invisible — which is, with some irony, exactly the category this record identifies as the dangerous one.

The list is most interesting at the point when everybody has stopped caring. At six months it says very little. At eighteen months it says most of what is on this site. The interval between those two is exactly the interval over which organisational attention decays.

It was never a management document, and that is why it is honest. Nobody reported from it. It did not go to a board, did not appear in a status update, and was not used to show progress. A field recording whether anybody objected, kept in a document that will be read upwards, quietly becomes a field recording whether the objection was reasonable — which is a judgement, made by somebody with an interest in the transition going smoothly. For eighteen months nobody had any reason to shade this one.

What makes it cheap

Three things, all of which are decisions made at the start:

• Write it in the room where the removal is agreed. A list assembled afterwards from memory is missing the unproposed removals and the third column entirely, and the third column is where the finding came from.
• Keep the third column to three values. Argued, asked, nothing. Anything richer does not get filled in. This was the cheapest column to maintain and it turned out to be the one that predicted something.
• Diarise the read-backs before the transition ends. Three calendar entries, at six, twelve and eighteen months, owned by a named person. Without this the list does not get read, and an unread list is an hour wasted rather than three hours well spent.

What it bought

Concretely, on this record:

Two of the five costly removals were identified by the read-back rather than by the problem itself arriving.

The eighteen-month pass is what connected the repeated rediscovery of a refusal reason to a log that had stopped sixty-four weeks earlier. Without the list, that would have been four people independently solving the same problem and nobody ever knowing why it kept recurring.

It also bought the ability to say something true about the transition afterwards. The alternative account — the one available to an organisation without a list — is a recollection, assembled a year later by people with an interest in the thing having worked, about events nobody wrote down. That account is not dishonest. It is simply made of the wrong material.

The smallest useful version

If three hours is too many, the version that still works is one hour: the first three columns, filled in during the transition, and a single read-back at twelve months.

That loses the distinction between the twelve-month and eighteen-month picture, which on this record is two entries. It keeps the list, the dates, the third column and the read-back, which is everything the finding on this site was built from.

Below that there is nothing useful. A list without the third column carries no signal, and a list nobody reads back is an hour spent producing an artefact with no consumer.

What a spreadsheet is good at here

The format matters slightly more than it sounds, and the reasons are not about tooling.

A spreadsheet sorts. The finding on this site — that quiet removals were the slow ones — emerged from sorting thirty-eight rows by a three-value column, which is an operation that takes two seconds and is unavailable in a document.

A spreadsheet also tolerates a column being half empty. The outcome column was blank for most rows for most of the eighteen months, and a half-finished table looks like work in progress rather than like a neglected report.

And it is a file rather than a system. It survived the end of the project by being in somebody's own folder, which is a weakness if you want institutional memory and the reason anything survived at all here. A record held in a project management tool would have been archived with the project, which is the same as deleted.