I'm kinda disappointed in the lack of support in the Open Source world for this kind of thing, but I've been guilty of it myself in the past. These days, I always ask, "How does this effect users with no or limited sight? Limited mobility? Limited hearing?" I don't always know how to resolve the problems, but I'm always thinking about it and trying to learn. I recently implemented a rule for myself that I won't publish any new video without complete and accurate captions (the automatic captioning at YouTube is insufficient to satisfy this requirement, but it's wonderful that they make captioning so easy and fast). Likewise, we have a new theme that's been developed by a third party which is miles ahead of any UI we've had in the past, but we're not gonna make it the default until we've at least made a clear path for reader users to get to a theme that will work well for them (our old mobile theme uses a hierarchy of menus and is surprisingly useable by screen readers, so we continue to support it just for that reason).
This may be something I should try just so I can wrap my head around how one would use our software without being able to see it. It's been around for a long time. I wonder, however, if Emacspeak would give me anywhere near a realistic average user experience.