Seated movement, and why five variants or it does not ship

Independently authored, not degraded

The tempting implementation is to write the standing version and generate the rest by removing things. It produces a seated movement that is worse than it needs to be, because a good seated movement is not a standing one with the legs deleted — it uses the chair.

So each of the five is authored, reviewed and dosed on its own terms. The equivalence multiplier that lets them be compared is computed, not assumed.

The five

Standing. Seated. Chair-supported, where the chair takes load rather than just holding the person. Bed or recliner, for people who are not reliably transferring. And adaptive single-limb, for anyone working with one usable side.

Independence and Vitality modes default to chair-supported rather than standing. In Vitality, standing requires an explicit clearance rather than being the assumption somebody has to opt out of.

Why there is no override

Because an override is used. It gets added for a genuine emergency, it stays, and within a year the library has forty movements that only work if you can stand. The people affected are the ones least likely to complain and most likely to just stop opening the app.

A partial movement stays in draft indefinitely. That is the entire enforcement mechanism and it is enough.

What it costs

Roughly four times the authoring effort per movement and a slower library. In exchange, the sentence "every body qualifies" is a property of the system rather than a claim on a marketing page.

The equivalence multiplier

Five independently authored variants create a problem: if a seated version is not a reduced standing version, the two are not directly comparable, and the whole progress model depends on comparing them.

So each variant carries an equivalence multiplier computed against a reference effort, not assumed from its posture. A well-chosen chair-supported movement can score higher than a lazily performed standing one, and the numbers reflect that rather than encoding a hierarchy where standing is simply worth more.

This has a pleasant consequence: somebody who moves to a seated movement during a flare-up does not watch their progress collapse. Their effort is measured on its own terms.

Substitution goes down the ladder, never up

There is a support ladder — standing, chair-supported, seated, bed or recliner — and the engine may substitute downward without asking. If the context suggests standing is a poor idea right now, it offers the supported version.

It may never substitute upward. A person who has been offered seated movements does not suddenly get a standing one because their step count looked promising, because the cost of being wrong in that direction is a fall.

Upward movement happens only when the person changes their capability profile, or when a clinician-facing clearance is recorded. In Vitality mode, standing requires that clearance rather than being the default somebody has to opt out of.

Six cue sets, as well as five variants

The gate checks more than posture. Every movement also carries text, audio, captioned video, haptic and voice-only cue sets, plus an Easy Read register, because a movement somebody cannot understand is as unavailable as one they cannot perform.

Captioned video is the one that catches authors out. An uncaptioned demonstration is not a video with a missing feature, it is a movement that does not exist for a deaf user, and the gate treats it as such.

What this looks like from outside

The publishing contract is a public endpoint. Ask `/movements/gate` and it returns the five variants, the six cue sets and the screening requirement, in full, as data. Then attempt to publish a movement with four variants and read the refusal.

A claim you can call is worth more than a claim you can read.