meta re: OCRbot, longish, -~
i think i approached the topic of accessibility through image captioning the wrong way. the intent behind OCRbot was to say "hey, some people need descriptions for images. please remember to write them out." but that's not the message that came across. what ended up happening was that people who previously didn't bother to caption their images despite being fully capable of doing so continued not to, and instead simply tagged OCRbot to do it for them. that feels like the most handwavey, uncaring thing to me. just "oh i don't have to caption this, the bot will probably get it right". it frequently doesn't, by the way. i recommend DMing the bot, then copy and pasting its response so you can edit it and clean it up, rather than making vision impaired users dig through the replies and open a content warning to view a poorly transcribed mess that approximates the original.
what i *should* have done was made a video, or something similar. something that illustrates the issue and why you should care and fix it in a more engaging format than a simple post. something that gives you an emotional *and* logical reason. something that really sticks with you.
but i can't do that. i'm a programmer. all i can do is code. so that's what i did. i made OCRbot as a vehicle for a message, but that message clearly didn't come through. i may have achieved success in the form of making an easy to use text transcribing bot, but i failed in what i actually set out to do.
re: your OCR bot toot, suggestion
I think you can still achieve what you want.
You can't force people to caption images properly, but when they use your tool you could force them to review the output by instead DMing the result to them.
That way a human is put back in the loop as last decision maker instead of delegating responsibility entirely to the bot.
@firstname.lastname@example.org's anti-chud pro-skub instance for funtimes