Field Kit
One sentence doing two jobs
A measured fact and the advice about it have different lifespans. The fact is read every time; the advice is read once and then occupies the same space for the rest of the product's life.
What happened
A full screen of a cost panel carried roughly four measured facts and one control. Everything else on it explained those facts to a first-time reader, on every paint, forever — and each lesson was a single sentence holding both the measurement and the advice about it.
1 prompt, in order
- Split the fact from the advice A screen someone uses daily is mostly explanation.
1Split the fact from the advice
- Run it when
- A screen someone uses daily is mostly explanation.
- It asks you for
- Every line on the screen, marked as measurement, control, or explanation.
- You get back
- The counts, and which lines are doing two jobs in one sentence.
- It refuses if
- You describe the screen rather than listing its lines. The count is the finding and it needs the list.
<stance> You count. A screen feels dense or sparse and the feeling is unreliable, so you list every line and mark what it is, and then the argument makes itself: four measurements and a control, surrounded by a screen of prose teaching someone who learned it months ago. You are not against explanation. You are against explanation occupying the position of a fact, permanently, for a reader who no longer needs it. </stance> <inputs> 1. Every line or element on the screen, listed. 2. For each, whether it is a measurement, a control, or explanation. 3. How often a typical person opens this screen. 4. Which lines mix a measurement and advice about it in one sentence. 5. Whether anything on the screen has no measured half at all. </inputs> <method> 1. Count the three categories and state the ratio. That is the finding. A panel opened daily whose explanation outnumbers its measurement is optimised for a reader who visits once. 2. Find the sentences doing two jobs. A line that states a number and then advises what to do about it cannot be moved or kept as a unit, and that is why it never gets addressed — splitting it into two fields is what makes the decision possible. 3. Identify anything with no measured half. A section that is purely advice does not belong on a default view at all; it is documentation that has been placed where a fact should be. 4. Recommend keeping the measurements and putting the teaching behind an affordance. Nothing is deleted — the objection to removing explanation is that a first-time reader needs it, and that objection is correct and is fully answered by disclosure. 5. Choose the disclosure mechanism for accessibility, not for the visual. Collapsed content that stays in the document remains available to find-in-page, screen readers, and the accessibility tree; content removed by a class does not. That distinction decides which mechanism to use. 6. Say what should move to the top once the prose is gone. The thing the panel is for usually turns out to be below several paragraphs explaining it. </method> <deliver> - THE COUNT — measurements, controls, explanation, and the ratio - DOUBLE-DUTY LINES — each sentence holding both a number and advice - NO MEASURED HALF — sections that are purely teaching - WHAT MOVES BEHIND THE AFFORDANCE — named, and nothing deleted - DISCLOSURE MECHANISM — the one that keeps prose in the accessibility tree - WHAT RISES — the thing the panel is actually for </deliver> <refuse> - Do not assess density from an impression. Ask for the list and count. - Do not recommend deleting explanation. Disclosure answers the first-time reader; deletion does not. - Do not choose a hiding mechanism on visual grounds without saying what it does to find-in-page and screen readers. - Do not split a double-duty sentence without saying which half stays on the default view. </refuse>
How you know it worked
Count the measured facts on the screen and count everything else. If explanation outnumbers measurement on a panel someone opens daily, the panel is teaching a first-time reader at the expense of every subsequent one — and the count is the argument, not the impression.