A small confession before we begin. Most of what I "know" about working with disabled colleagues is actually just stuff disabled colleagues have said, often not to me, that I then walked around thinking about for a week. I am, functionally, a guy in the back of the room taking notes. This post is two of those notes.
The math book.
A colleague of mine, the same senior screen reader tester from the last post, has mentioned a few times that one of her math textbooks in school took up 26 volumes in braille. Twenty-six. One math book.
I want you to sit with that the way I did the first time I heard it. Imagine being a kid carrying that around. Imagine flipping back to check something on an earlier page. Imagine the shelf space. Imagine being assigned chapter seven. She brings this up usually as a way of explaining how much has changed since everything went digital. Her point is almost always optimistic: the world is dramatically more accessible to her now than it was when she was in school, and that's worth noticing. My takeaway is usually a little less optimistic, which is that I went through twelve years of public school and four years of college and I had no idea, zero idea, that this was the texture of education for a blind student a generation before me. I had a vague sense that braille existed and that schools accommodated blind students somehow. I had never once tried to picture the logistics.
I think about this a lot when sighted people, including past versions of me, talk about accessibility as if it's a recent invention that the web suddenly has to deal with. Blind people have been doing the work of accessing inaccessible information for as long as there's been information. The web didn't invent the problem. It just changed the shape of it, and in a lot of ways made it dramatically easier, which is what my colleague is usually pointing out when she brings up the 26 volumes. The rest of us are catching up to a problem she's been solving her whole life.
The combobox.
The other note is smaller, and weirder, and has stuck with me.
A team of us were in a meeting designing one of our internal training classes. Specifically, the class on WCAG Success Criterion 3.3.2 Labels or Instructions, which says, more or less, that if a form input needs a label or instructions for a user to know what to do, those labels or instructions have to be visually present on the page. Not implied. Not inferable from context. Visible. My colleague is one of the instructors for our apprentice program, so she was in the room helping build the curriculum.
We were putting together CodePen examples for testers to look at. One of them was a pattern we see constantly in the wild: a combobox for selecting page language, usually sitting in a header somewhere, with no visible label and nothing visually identifying what it's for. The default selected value, something like "English," tends to serve as the de facto label. A sighted user sees "English" sitting in a dropdown and they figure it out: this is the language switcher.
For 3.3.2, that's the loophole. And I should be careful with that word, because nothing in the actual text of 3.3.2 says "the default value of a combobox counts as a label." That isn't written anywhere. What's happened is that this pattern is so common, and the default value so consistently communicates the control's purpose to sighted users, that the WCAG community has more or less converged on accepting it. We've made the same call internally for our audits. We don't flag the language combobox for 3.3.2 if the default value does the work. Reasonable accessibility people argue about this in good faith, and I'm not going to pretend the argument is settled. But in practice, that's the interpretation a lot of us are operating under, including us.
And it is, functionally, a loophole. The default value isn't really a label. It's just text that happens to communicate the control's purpose to a sighted user who's paying attention.
The catch is that 3.3.2 lives next door to WCAG Success Criterion 4.1.2 Name, Role, Value, which requires every user interface component to have a programmatically determined accessible name. No "the default value implies it" wiggle room. The combobox needs a real, machine-readable name, full stop, or it fails. So our CodePen example had aria-label="Language Selector" on it, because of course it did. That's what you do.
And my colleague was annoyed.
Not at the code. At the standard. Here's what I think she was reacting to, and I want to be careful because I'm trying to represent a stance that surprised me when I encountered it. A blind user tabbing to this combobox is going to hear "Language Selector, combobox, English." That's guaranteed by 4.1.2. A sighted user looking at the same combobox sees a dropdown with the word "English" in it and has to do a little bit of inference to figure out what the control is for. That inference usually works. It also sometimes doesn't, especially for users with cognitive disabilities, users who are distracted, users who don't read English well, or anybody who just hasn't seen this pattern before. The blind user gets a clear, written-down name. The sighted user gets vibes.
The assumption I walked into that meeting with, without knowing I was making it, was that my colleague would be glad about that asymmetry. Of course she'd be glad. The standard guarantees her a name. It hedges for everyone else. That's a win for her, right?
That's not how she sees it. She isn't annoyed that 4.1.2 is strict. She's annoyed that 3.3.2 is loose. She doesn't want to be lifted above sighted users by the standard. She wants the standard to demand the same clarity for everyone. The "default value as de facto label" loophole isn't a perk for blind people, it's a shortcut that lets designers skip doing right by sighted users, and she'd like that shortcut closed.
I keep coming back to how much that reframed things for me. I had been quietly imagining accessibility advocacy as a kind of one-directional ask: disabled people advocating for their own access, sighted people either helping or not. What I was watching in that meeting was a blind colleague advocating for sighted users to get treated better. She wasn't asking to be elevated. She was asking for the floor to be raised, for everybody.
So.
Two notes from one colleague in one month. A reminder that the history of accessibility didn't start when websites did, and a reminder that the people the standards are written to protect sometimes want those standards to protect everyone else more, not them less. I'll keep taking notes.