The quiet ones
Nothing happened is not a result
An entry with no event looks like evidence of a successful removal. It is the absence of evidence, and four of them were about to become expensive.
The quarterly reconciliation of the project ledger
week 3 · surfaced after 31 weeks
Shadowing new starters through their first week
week 6 · surfaced after 47 weeks
Reading supplier contracts before renewal
week 2 · surfaced after 22 weeks
Logging why a job was turned down
week 9 · surfaced after 64 weeks
Surfaced 4
Four removals were written down with the week each of them stopped. Every one of them produced an event inside the 78 weeks watched. Of these, four surfaced as a problem. The four removals nobody said a word about, all of which later surfaced as problems.
At week twenty-six, nineteen of the thirty-eight entries on this list had produced no event of any kind. Nobody had asked for them back. Nothing had crept back. Nothing had gone wrong.
That is the largest single group in any snapshot of the record, and it is the one that does the most damage to the reading of a transition, because of how naturally it converts into a claim.
The conversion
It happens in one step and the step feels like reasoning:
- We removed nineteen things.
- Nothing has happened.
- Therefore the nineteen were unnecessary.
Step three does not follow, and the reason it does not is that step two is not an observation about the nineteen things. It is an observation about the reporting channel.
"Nothing has happened" means no one has reported anything. Whether anybody would report anything depends on whether anybody is in a position to notice, and for a substantial share of any list of removals, nobody is.
What the nineteen contained
On this record, the nineteen silent entries at week twenty-six contained:
| what they turned out to be | n |
|---|---|
| genuinely not needed, still gone at week 78 | 14 |
| came back quietly, later | 2 |
| surfaced as a problem, later | 3 |
Fourteen out of nineteen is a good hit rate, and anybody who inferred "mostly unnecessary" from the silence would have been mostly right.
The three are the issue. At week twenty-six they were not merely unknown; they were sitting inside the positive total, indistinguishable in every respect from the fourteen. Same silence, same absence of complaints, same appearance of a clean removal.
Why there is no tell
It would be useful to be able to say that the three looked different — that they were larger, or older, or more contested, or more central.
They were not. One was removed in week two with a one-line description. One stopped without being proposed at all. One was removed on sensible grounds by somebody outside the team.
Their only common property is the one described throughout this part of the site: the person who would notice the absence was not going to be in a position to notice for a long time. That is not a property visible on the day, which is precisely why it cannot be used as a tell and why the silence is uninformative.
How to handle a silent entry
Not by keeping it. The fourteen that stayed silent all the way to week seventy-eight are the best thing on this record and the strongest argument for shorter weeks anywhere in it. Treating silence as a reason to reinstate would throw those away to protect against three.
The handling that works is to stop treating silence as a terminal state and start treating it as an open file:
- Count them, by name, at every reporting point. Nineteen at week twenty-six, sixteen at a year, fourteen at eighteen months. The series itself is informative; it is the only number in the transition that reports on what is still unknown.
- Give each one a date — the next external occasion on which its absence would be visible. Most will have one within weeks. Those can be forgotten.
- Flag the ones with no date. On this record there was one: the log of refused jobs, whose absence had no scheduled occasion at which it would show. It took sixty-four weeks and it is the hardest case on the record. A blank in that column is the clearest warning the method produces.
The sentence to avoid
"We removed thirty-eight things and nobody noticed."
It is the sentence everybody reaches for, it is true in a narrow sense at any point in the first six months, and it asserts something about the removals that is actually a fact about the observers.
The accurate version is longer and much less satisfying:
"We removed thirty-eight things. Nineteen have produced no information yet. Of the rest, seventeen came back in some form and two were problems."
Nobody will put the second version in a slide. It is, on the evidence of this record, the only one of the two that was true at the time.
The two that came back quietly after week twenty-six
Of the nineteen silent entries at six months, two eventually crept back: the induction folder at week thirty-seven and the Friday clean-down at week thirty-one.
Both are worth a line because they are the most benign thing that can happen to a silent entry and they still cost the organisation its own account of itself.
Neither was decided. Neither was recorded at the time it resumed. Both appear on the list as removals and both are running. The organisation believes it stopped two things that it did not stop, and it found out only because somebody read the list back eighteen months later and asked.
A silent entry is not only an unknown. It is an unknown in both directions: the thing may be unnecessary, it may be about to cost something, or it may simply have resumed without anybody noticing. Three states, one observation, and the third is the one nobody anticipates because it does not sound like a risk.