Writing the Reason Beside It

Why reasons decay so fast

Because they live in a conversation rather than in an artefact. Four people each hold a fragment, each fragment degrades differently, and nobody notices because the decision itself is still perfectly clear.

Ask a room six weeks later why something was decided and you will get four confident and partially incompatible answers.

What a reason is for

Deciding the cases the decision does not cover. Every rule meets situations its authors never imagined, and the only guide to those is why the rule exists.

Somebody at a distance meets those cases alone, with no room to read the answer off.

The sentence that does the work

Name the constraint. Not the goal, which is usually generic, and not the process, which is irrelevant: the specific thing that made the alternative unavailable.

"We chose the slower approach because the client's security review could not be scheduled before March" tells a reader exactly when the decision stops applying.

The difference between the two, on the same decision:

We chose the batch import because it is more robust and better suited to our needs.

We chose the batch import because the client's security review could not be scheduled before March, and the live feed needs it.

Why constraints beat rationales

A rationale explains why something is good. A constraint explains why the alternative was impossible, and the second is both shorter and far more useful.

It is also easier to write, because at the moment of deciding everybody knows what the blocker was, and almost nobody can articulate a full rationale without it turning into three paragraphs nobody reads.

Recording what was rejected

One clause naming the main alternative and why it lost. It prevents the same option being raised and rejected three more times over a year.

It costs six words.

The obviousness trap

At the moment of decision the reason feels too obvious to write. That feeling is reliable, universal, and always wrong, because the audience is not the room now.

Writing something that feels redundant is the correct behaviour here.

Reasons that cannot be written

Some decisions rest on something that cannot go in a shared document: a commercial position, somebody's capacity, a difficult relationship.

Say that explicitly rather than inventing a thinner reason. "Decided for reasons I cannot put here — ask me" is honest, and a plausible substitute is worse than silence because it will be relied on at the edges and will mislead somebody.

When the reason is somebody's judgement

Sometimes there is no external constraint and the answer is that somebody experienced thought one approach was better. That is a legitimate reason and it should be recorded as what it is: "chose B on Mira's judgement, no hard constraint either way".

Writing it that way is honest and it is also useful, because it tells a reader exactly what would be needed to reopen it — a conversation with Mira — rather than leaving them hunting for a constraint that does not exist.

Whose job this is

The same person who writes the decision, in the same minute. Splitting the two tasks means the second one does not happen.

If the decision is worth recording, the reason is the part worth recording.

Where it lives

On the same line as the decision. A separate document of rationales is a document nobody opens.

Proximity on the page is what gets it read.

Revisiting against the constraint

Where the constraint is written, somebody can check later whether it still holds. That turns a decision from permanent into conditional, which is usually what it actually was.

Teams without recorded constraints accumulate rules that outlived their circumstances and cannot tell which ones.

The one-line review

Once a quarter, read the recorded constraints and ask which have expired. It takes twenty minutes and usually frees something.

The person outside the room is often the best placed to do it, having less investment in the existing answers.

What a written reason still is not

An account of the thinking. It is one constraint, chosen by one person, from a conversation that contained several considerations and a good deal of uncertainty.

A reader gets the load-bearing part and does not get the texture: how contested it was, what nobody said out loud, which participant was quietly unhappy. Those do not travel, they are covered elsewhere in this collection, and the right response is to carry the part that can be carried rather than to abandon the exercise because it is incomplete.