Product.

The framework

1. 01

   Every product assumes an ability, and nobody writes it down.

   Start by naming what your product quietly requires. Two hands, probably, and steady ones. A quiet room, or headphones. Colour vision good enough to tell the red state from the green. Fluent reading in the language you shipped, at the size you chose. Sustained attention for as long as the flow runs. None of it appears in a spec, because it never occurred to anybody to say it out loud — and that's Holmes's point about exclusion: it isn't malice, it's product people solving a problem the way their own bodies happen to work. Write the list down before you argue about a single line of it.

2. 02

   Three populations, one mismatch.

   Reading an ability across the spectrum is a mechanical exercise, and it takes about ten minutes per ability. For each one, write the permanent person, the temporary person and the situational person, then write what each of them actually experiences at the specific step where your product asks for it. They all hit the same wall. What differs is how long they're standing at it, and how likely they are to tell you — the permanent person has met this a thousand times and can describe it precisely, while the situational one just puts the phone down and gets on with their day.

3. 03

   Solve for one, extend to many — and check the second half happened.

   The toolkit's own principle runs in two parts, and teams reliably do the first and forget the second. Solving for one gives you a clean specification, because a person permanently excluded can tell you exactly where the thing fails and won't politely work around it. Extending to many is a separate piece of work: you take the fix that was built for the edge and make it the default rather than a setting buried three screens down. OXO didn't ship a special peeler for arthritic hands and a normal one for everyone else. There was one handle, and it sold.

4. 04

   The reach number is bigger than the permanent count.

   This is where inclusive design stops being an ethics conversation and starts being an arithmetic one. Count the permanent population if you like, but the temporary and situational ones behind it are the size of the prize, and they're the groups nobody has been counting at all. Day 43 measured physical availability as the steps between wanting and paying; the spectrum tells you whose hands those steps are actually stopping. So when a fix gets scored for reach, don't score the permanent group. Score everybody the mismatch touches, then let it compete on that number.