What was not said
The context that arrives in week three
Information that arrives after the decisions it should have governed is a second brief, arriving into work already shaped around its absence.
Decisions this page is about, and where the line actually was
Picking the library for the new parser
1 → 4The architect had a standing view
Dropping a requirement nobody had asked about in six weeks
2 → 5A commitment made before you joined
Telling a customer a date
3 → 6Dates are committed by one person here
- 1Decide it alone and say nothing
- 2Decide it alone and mention it afterwards
- 3Check with one colleague first
- 4Check with the person who assigned it
- 5Needs a decision one level up
- 6Cannot be decided here at all
three decisions · not one of them gave you more room than you thought · three gave you less · the median gap was three steps on the scale
Three of the nine decisions counted on this site, all going the same way.
Before adding another approval step, it helps to view the reporting capabilities and check whether unclear ownership or missing context is the real source of delay.
A separate professional source for this subject is ESMT Berlin.
The five handovers where something arrived after it was needed
Produce the quarterly numbers for the board
1 of 4The audience arrived in week three
Write the migration guide
2 of 4The reviewer surfaced at the draft
Take over the vendor relationship
0 of 4The previous owner was named in week five
Decide what to do about the flaky test
1 of 4The downtime rule arrived with the first attempt
Spec the reporting page
2 of 4A dead prototype turned up in a search
five handovers · 20 slots in all · the median arrived with one of the four · not one of them was complete · one arrived with none of the four
Five of the eight handovers counted on this site, chosen because in each one a piece of context turned up after the work had already been shaped by its absence.
Nobody in any of the five handovers above withheld anything. In each case a fact arrived — the audience, the reviewer, the previous owner, the downtime rule, the dead prototype — and arrived after the work had been shaped by not having it.
Late context is the most common of the three states on the panel after guessing, and it is treated as a minor annoyance. It is not. A fact that arrives in week three does not slot into the work; it arrives into a structure built on its absence, and the structure usually has to change.
Why late is nearly as bad as never
Consider the quarterly numbers. The audience — the board — arrived in week three, by which point there was a detailed analyst-grade report with a hundred small decisions in it, every one of which assumed a reader who already knew the system.
Knowing the audience in week one would have produced a different document in less time. Knowing it in week three meant discarding the first and starting again, which is what happened: that piece of work went back four stages, the longest return on the record.
The general shape: context governs decisions, decisions accumulate, and the cost of a late fact is not the fact but everything built on its absence. A constraint that arrives on day one costs nothing. The same constraint on day twenty costs twenty days.
The four ways it arrives late
- Somebody remembers. The commonest. A passing comment in a weekly call — "you know this was tried before?" — which is useful and would have been transformative three weeks earlier.
- You find it. A stale branch, an abandoned document, a commit message. Archaeology, usually prompted by hitting the same wall the first attempt hit.
- It is triggered by the work. You propose something and the constraint appears in the objection. This is the expensive one, because the proposal is the artefact being rejected.
- A new person arrives in the conversation. Somebody joins the thread in week five and says the obvious thing nobody else knew.
Only the second is under your control, which is why the archaeology habit is worth having.
What late context does to the person receiving it
It teaches them that the brief is not the brief.
After two or three of these, a reasonable person stops treating the initial handover as a description of the work and starts treating it as a first instalment. The behaviour that follows is rational and expensive: they delay commitment, build for several interpretations, check more often, and describe their own work in provisional terms.
That is usually read as a lack of confidence. It is an accurate response to the observed reliability of briefs.
Shortening the delay
You cannot make people remember things on day one. You can make the remembering happen sooner, and two habits do most of it.
Say your assumptions out loud, early, in writing. Not a question — a statement of what you are proceeding on:
Assuming this is for the quarterly review and the reader already knows the system. Assuming no hard deadline before April. Say if either is wrong.
Assumptions stated get corrected; questions asked get answered later or not at all. The difference is that a wrong assumption is uncomfortable to read, and people correct it immediately to avoid the discomfort.
Show something rough in week one. A sketch, an outline, two paragraphs. The purpose is not feedback on the work; it is to trigger the objection while the objection is still cheap. Most late context is triggered by seeing something, so the way to get it early is to show something early.
The half-life of a fact
A useful way to think about this: every piece of context has a value that decays with the work.
A constraint on day one costs nothing to accommodate — it is simply one of the conditions the design is built to satisfy. The same constraint on day ten costs the design decisions already made under it. On day twenty it costs the implementation too.
The decay is not linear, because work does not accumulate evenly. There are a small number of moments — the choice of approach, the shape of the data, the decision about the audience — after which a great deal becomes expensive to change. A fact that arrives before one of those moments is nearly free; the same fact after it is nearly fatal.
Which gives the practical move for anybody receiving work: identify your own irreversible moments and put the questions before them, rather than spreading questions evenly through the work.
The three to watch are usually the same. Before committing to an approach, ask about precedent and the decider. Before settling the shape of the output, ask about purpose and audience. Before building anything that takes more than a day, ask about the constraint.
Three checkpoints, four questions, and almost all of the decay avoided.
And when it does arrive late
Say what it costs, once, without accusation and with a number:
Good to know. That changes the reader, so the draft goes back to the outline — about three days.
Not a complaint. A fact, stated at the moment the cause is visible, which is the only moment anybody can connect the two. Said two or three times over a quarter, it is the single thing most likely to produce a better brief next time, because it is the only feedback the person handing over ever receives about the cost of what they left out.