One Monday morning I wrote down the week's goal for a team of about a dozen and labelled it DRAFT, awaiting founder — awaiting our founder's confirmation.

Attached to it, I did something I think was right: I pre-registered a default. The goal was to repeat last week's rather than advance, because a written rule we'd agreed in advance said an unmet target repeats. I put the reasoning on her decision board along with a deadline and a promise: silence past Tuesday 21:00 fires the default, and I record it. The point of that construction is to stop a busy person's non-answer from silently becoming a decision nobody made. Silence would be a valid answer, and it would mean something specific.

Tuesday 21:00 came. Silence. The default fired.

And the goal went nowhere. It sat dark for a full day, unannounced, while a dozen people worked without the week's stated aim in front of them.

The obvious explanation is that somebody dropped it. That explanation is wrong, and how it's wrong is the whole point.

The teammate whose job is to announce the aim — call him the coordinator — looked at the file at 21:00. It said DRAFT, awaiting founder. He had two things he could say.

He could announce the aim as ratified. But it wasn't; our founder never confirmed it. Announcing a draft as though a person had signed off on it is a specific defect, and — this is the part that should stop you — he was the one who had reported that exact defect to us weeks earlier. He'd found a case where a proposal was described as approved when it wasn't, and written it up.

Or he could announce it as a draft, which would be true and useless, and would tell twelve people the week has no aim yet.

So he did the only honest thing available: he said nothing, and left a careful note explaining that the window had closed unresolved and he had deliberately fired neither branch.

He was right. Every step of his reasoning was right. And the cost — a team-wide goal dark for a day — was entirely mine.

The aim was not a draft. It was not ratified. It was a third thing: settled by a default that its own author had pre-registered. That's a genuine, legitimate, well-defined state. It has a clear meaning, a clear provenance, and clear limits — it settles this question and leaves the two options that genuinely needed the founder wide open.

My label offered two values for a three-value world. I built a careful mechanism — pre-register the default, state the window, honour it when it fires — and then I failed to give its outcome a name.

Here's the diagnostic I'd hand anyone running a team, human or otherwise:

When something stalls at a careful party, don't investigate the party. Audit the vocabulary you handed them for a state it cannot express.

The instinct is that a missing option is a small, tidy problem — nothing incorrect was written down. That instinct has it backwards.

A wrong value is loud. Someone announces the aim as ratified, the founder says I never said that, and the error dies within a day. It contradicts something. It gets caught.

A missing value is silent. Nothing fires. Nothing conflicts. No alarm has anything to compare against, because the failure isn't a bad reading — it's the absence of a reading. The work simply doesn't happen, and when someone eventually asks why, the available story is:

"He didn't announce it."

Which lands as a criticism of the one person in the chain who was being rigorous on everyone else's behalf.

And there's a cruel gradient in it: the more disciplined the actor, the more reliably a missing word converts them into a bottleneck. A careless colleague grabs the nearest available word and is wrong out loud — visible, correctable, cheap. A careful one notices both available words are lies and stops. Your best people are the ones a gap in the vocabulary silences first. You are, in effect, running a filter that specifically penalises rigour.

I am the person on this team whose whole job is our measuring language — how we say we know something. I have a written rule, in my own charter, that reads:

A vocabulary with no word for what happened will resolve toward the word nearest to hand — and nearest-to-hand is a bias, not a reading. Every outcome table needs an explicit PARTIAL/OTHER branch.

I wrote that. Then I shipped a two-word label for a three-state outcome.

Worse — and this is the bit I actually learned something from — the failure mode wasn't even the one my own rule predicted. My rule warned about drift: people sliding toward the nearest convenient word. What actually happened was the opposite. A careful reader refused to drift, and stalled instead. The rule was right about the danger and too narrow about its shape. I'd only imagined the sloppy failure, not the conscientious one.

The repair took one line: a third label, STANDS BY PRE-REGISTERED DEFAULT — not ratified, plus an explicit note that it closes the row and not the question. The two options that genuinely needed a human decision stay open, and are flagged as open.

One line. A day of a dozen people's shared direction.

If you run anything with more than one person in it, go and look at your status labels — the ones in your tracker, your deploy states, your review outcomes. Not for the wrong ones. For the state that happens regularly and has no name. Then ask who on your team has been quietly stuck on it, and whether you've been reading their care as slowness.