People who are excluded belong inside the design process

What it is

A week before launch, a Blind friend is asked to “take a look,” unpaid, with no power to move the information architecture. The exclusion cards and the curb-cut are still on paper; the door still opens the old way. Design with excluded people means the people blocked by that assumption enter problem framing and option choice, not only a one-shot crash test at acceptance. Nothing about us without us is not a slogan here. It is a claim about where knowledge comes from — a checklist cannot infer that person’s task structure.

If participation is one person, only at the end, and only as unpaid labour, exclusion repeats in another form: “we asked a disabled user” goes in the report, and decision rights stay with people who were never blocked.

Why it happens

Excluded people hold local knowledge no checklist has: how their own assistive technology is paired, which wait will collapse switch scanning, which error will fight the speech cursor for focus. That knowledge cannot be reverse-engineered from success criteria, and it cannot be simulated by a designer turning the display off — simulation removes a channel, not long-term strategy or device fluency. Put it at the front of the process and the option still cheap to turn; put it at the end and the structure is locked, so participation can only rewrite copy.

The second layer is representation and pay. Disability is not a monolith: one Blind VoiceOver user does not answer for low-vision magnification, Deaf users, or cognitive disability. Unpaid “a friend will help” selects people with spare time in the team’s own circle and distorts the exclusion structure. Paid, several people, across impairments, with a channel that can change the spec — then participation is a design activity. Otherwise it is public relations.

Studying it

Inspect traces of participation on the decision timeline: at which stage excluded people first appear (topic choice / option comparison / final acceptance), whether they were paid, whether the spec was rewritten because their task failed. Use the methodological cautions from testing with disabled users: their own devices and AT, not a unified lab setup that irons difference out.

Independent variables: stage of involvement, number and impairment coverage, whether paid, whether a channel existed to change the spec. Dependent variables: count of spec changes and whether they happened before structural lock, share of defects that could not be moved when involvement was only final.

A small n is not the danger. Writing n=1 as “disabled users signed off” is. Qualitative notes should read “these participants, these tasks, these failures,” and name impairment types not covered.

Where it stops holding

On confidential, safety-critical, or pre-prototype work, participation may have to become a disclosable task fragment or a proxy, rather than handing an unreleased system to someone who cannot sign an NDA. Disabled organisations and specialist evaluators can bridge a gap when target users cannot be found; they must not permanently replace them. Participation with children or people with cognitive disability has consent procedure; “must participate” does not skip it. Participation is not transferring design responsibility onto excluded people — the product side still decides, but the grounds for the decision must include their task evidence.

Applying it

• Schedule paid sessions with excluded people before an option is chosen, on their own devices and assistive technology. The output must be able to move information architecture, not only colour.
• Cover at least one visual, one motor, and one cognitive or auditory-related participant on the same product. Do not let a single “disability representative” sign.
• Final acceptance can be a further round; it cannot be the only round. First contact after structural lock gets labelled “too late,” with a list of defects left unmoved.
• How to check: open the research log — is the first date excluded people appear earlier than option freeze, is there a payment record, and can a spec line be pointed at that changed because of them. Missing any of the three, participation is still an acceptance ceremony.

Related

• Same group: J1.05.1 Inclusive design starts by naming who the current design already shuts out · J1.05.2 Solving for one person at the edge often produces a curb-cut for many
• Nearby: J5.14 Testing with disabled users · Q1.05 Participant recruitment · J1.07 Social model of disability
• Search terms: participatory design · nothing about us without us · paid disabled participants