Neurodivergent-first is not accessibility later

Most software is designed for a mind it was never tested on, and then patched for everyone else with an accessibility pass, a focus mode, a settings toggle. The patch model has a quiet premise: there is a normal user, and there are exceptions.

This program rejects the premise, not just the patch.

The retrofit ceiling

Retrofitted accommodation can only ever soften what the core design decided.

If an interface's information architecture assumes sustained linear attention, no focus mode recovers the person whose attention is associative and bursty. If a system's notion of done assumes executive function that initiates on command, no reminder feature closes the gap. The decisions that exclude are upstream of the settings page.

There is a second cost the patch model incurs, and it compounds. Every accommodation added after the fact is a parallel path: a second code path, a second set of states, a second thing to test and to break. We call the accumulated total accommodation debt, and unlike technical debt the interest on it is paid by people rather than by engineering teams, which is why it so rarely appears on a roadmap.

The disclosure tax

The patch model also charges a fee that is easy to miss because it is not financial.

To receive behaviour that would have been better as a default, a person has to declare something: a diagnosis, a need, a status as an exception. That cost is social, it is charged before the benefit is delivered, and a great many people decline to pay it and go without.

Designing from AuDHD outward

This house includes AuDHD cognition, not as a market we studied but as the minds doing the studying. That changes the work in a specific way: the program's primary data is not borrowed.

When we treat attention, sensory load, and executive function as first-class design inputs, we are describing constraints we live inside, then asking what software would look like if it had honored them from the first sketch.

Some working positions, held loosely and tested by building:

• Attention is a budget the interface spends. Every element is a withdrawal. Most interfaces are overdrawn by default.
• Sensory load is a cost, not a style. Motion, sound, and density are materials with a price; the price varies by nervous system, and the design should know its own price.
• Initiation is the hard part. Systems obsess over completion and measure nothing about starting. For many minds, starting is the entire problem.
• Interruption is the normal condition. A design that punishes a person for returning an hour later is broken for most people and unusable for some. The whole question is what the system preserves, and what it makes a person reconstruct.

The bet

The bet of neurodivergent-first design is that it is not niche. Interfaces built for minds with the least tolerance for noise, ambiguity, and friction tend to be better for everyone. It is the curb-cut effect, upstream of the curb.

If that bet is right, the neurodivergent-first path is not a mode selected in settings. It is the path everyone uses, because it is the better one, and the case for accommodation as a separate track weakens considerably.

We research by building, and the systems in this program are in daily internal use. Their unit of publication, as ever, is a working thing.