A good screen reader should be capable of reading the unicode description of the emoji. Unfortunately, some don’t. More concerningly, I just realised that Discourse’s emoji substitution is really inaccessible - Discourse substitutes unicode emoji with its own pictures. Now, personally, I think that’s bad enough (emoji is not meant to be pictures, but rather a changeable font, but that’s a whole different debate and one I’m losing in many circles). However, Discourse is doing something really nasty… if they’re going to do emoji substitution, the alt text for the substituted image should be the unicode description (so a screen reader can handle it gracefully). Discourse isn’t doing that, instead it’s putting the code you can use to type the emoji, which may not read as well due to the colons.
Fundamentally, though, I agree. But part of the issue here is how Discourse is handling emoji substitution. Proper unicode emoji shouldn’t be an issue to a screen reader that is well designed and knows how to read them. That’s the point of unicode, well, part of it…
As for the original question:
friendly, excellent, efficient
Monzo’s been absolutely perfect in this front.