Where the old dashboard lied
It said 41%. The metric hadn't moved since the 4th of August.
A number was on the screen. It looked current. It was three weeks stale, and nothing about how it was displayed said so.
The sync stopped. The dashboard didn't notice.
A north star metric on one initiative is fed by a sync — something outside StratOps posts a value, and progress is worked out from that value against a target. The sync delivered a reading at some point, worked out to 41%, and the old dashboard drew a bar at 41% and left it there.
Then the sync stopped delivering. No error, no alert — it simply went quiet on 4 August. The bar stayed at 41%, because 41% was the last number anyone had written down, and nothing was checking whether it was still true. A stale reading and a current one were rendered identically: same bar, same percentage, same confidence.
Same initiative. Same underlying silence. Two different screens.
What it used to show
A progress bar, roughly 41% full. A percentage, printed with two significant figures of false precision. Nothing on screen distinguished a number computed this morning from one computed three weeks ago and never revisited. The metric was gated on whether a value existed at all — and a stale value still counts as existing.
What it shows now
North star
No reading yet
No bar. No percentage. The tool that answers get_initiative_status now asks a second
question before it answers the first: not just "is there a value" but "when did the value last change,
and is that recent enough to mean anything." When the answer is no, it says so in words instead of
drawing a bar that implies otherwise.
Refusing to answer is the feature, applied elsewhere too.
The north star isn't the only number in this system that would rather say nothing than say something confident and wrong.
- 112 decisions, 0 outcomes recorded.Calibration — how often a call turned out
the way it was expected to — needs closed decisions with a recorded result. There are zero, so
get_calibrationreports nothing rather than a score built on an empty sample. It will start answering the first time an outcome is written back against a closed decision, not before. - 190 observations, 0 triaged.Nothing has moved an observation into a decision, a task, or an idea yet. A report that summarized them as if they had would be describing work that hasn't happened.
- 51 ideas, 2 acted on.The lineage from idea to solution is tracked explicitly —
get_idea_lineage— so "acted on" means a real link exists, not that an idea is old enough to assume someone must have gotten to it.
Nothing here runs on a schedule.
Health, momentum, and the north star are all worked out when you ask — get_initiative_status,
get_initiative_momentum, get_initiative_health_breakdown — and nothing is cached
between asks. That has a cost: no alert fires the moment a sync goes quiet, because nothing is watching in
the background. It also means there's no stored number that can quietly drift out of date the way the 41%
did, because there is no stored number at all. Every answer is computed fresh from what's actually there,
including the honest answer that nothing recent is there.
This page describes one initiative's history, not a claim that every sync fails. Ask for the full account and I'll send it: jake@aifrontdesk.com.