It seems pretty obvious that accessibility is always going to require extra thought Actually, I think this extra thought could help everyone by treating disabi

It seems pretty obvious that accessibility is always going to require extra thought Actually, I think this extra thought could help everyone by treating disabilities as just special cases of the general idea that different people prefer different input/output devices and having apps (mostly) stop processing low-level input (keystrokes, pixel hit testing) and switch to handling more abstract "primitives". The OS would have a set of more abstract primitives: enter text, zoom in, zoom out, rotate, 3d move, 3d rotate, select item, select 2d region, select 3d region, deselect, grab, cancel, undo, etc. You, the user, would need a set of I/O devices that provided one or more implementations of each primitive in the checklist. One user would send an "undo" to an app with a click of the tongue, another with a hand gesture, another by pressing the ESC key, but the app just gets an "undo" event from the OS. For 2d region select, for example, one would use a joystick, another uses voice, another uses eye movement. The app doesn't process pixel hit testing to do its own region select, it lets the OS tell it what region of the 2d surface was selected. The app makers don't make the app "accessible" themselves, they just make it respond to the checklist of higher-level primitives. You could provide at least the core functionality of most apps by this means after which an app could possibly add some lower-level processing in the form of "shortcuts". When I'm forced to use a watch interface away from my keyboard, I am in a sense "disabled". When driving in my car with eyes on the road, I am "disabled". I'm not a gamer, so relative to a gamer with mad joystick skillz, I am "disabled". Well, fine, just change the approach to software interaction to a more abstract type of interaction that treats us all as being variously abled.