Once upon a time, when BlueSky was much younger--almost thirty-eight million users younger, in fact--people started to become conscious of using the alt-text field to describe their posted images. At its peak, alt text was posted with about three out of every ten images. Doesn't sound like much, but it was a remarkably high usage rate when it came to people embracing accessibility accommodations on social media platforms. In those days, a few tools were developed to facilitate and encourage people to add descriptions to the images in their posts. One of those tools is still around today and has come to be widely known as "the alt-text bot."

So. How did this miracle of technology come to be? What does it do? How is it used? And...what can't it do? So if I'm starting from someplace close to the beginning, I have to mention the bot's creator, @holden.bsky.social. Believe it or not, Holden doesn't maintain much of a presence on BlueSky. Never really has, as far as I'm aware (except for a short period waaaay back in the Elder Days). But he manages the bot to this day. It operates on the site under the handle @alt-text.bsky.social...don't forget the dash! It was built to use OCR to extract text from posted images. It can handle up to four images per post with up to 1000 characters in each post. According to its creator, it's also designed to use threaded replies to accommodate any overflow past that 4000-character count. As for how it's used, all you have to do to get it to work is to invoke its handle as a reply to the post that contains a text-based image. Seems simple, yes?

It's really too bad that nothing's ever as simple as it needs to be, but on the other hand it's always important to understand exactly what your expectations of the tool should be in the first place. The bot was never designed to do image descriptions. If there are no words in the image, the bot will not return a response. For memes or other images that are captioned, the bot will return the words only without any context as to how they may relate to any of the other elements that exist in the image. Some images in posts are contained in what are described as "preview cards." If someone puts a link in their post, these previews will be generated automatically from the content on the destination page and will often contain an image. The bot won't retrieve that remote content to get image details, and even if it did, it probably wouldn't be able to describe it...remember, it's only a text transcription tool. And it can't do text in quoted images. It has to be invoked in a direct reply to the post that includes the image content. Lastly, if the original poster has restricted who can reply to the post, the odds are pretty high that invoking the bot will do absolutely nothing for you.

This bot has become super-popular. I don't truly know how often it's invoked, but I can certainly see it being asked to do all kinds of stuff all the time. One key thing to remember is its original purpose. It was built specifically for screen reader users to use when someone has neglected to add alt text to their post or for when they themselves have made a post containing a text-heavy image that they can't transcribe easily. Now...I can't say that the people invoking the bot aren't all screen reader users. But my general impression is that there may be a lot of people who aren't using screen readers who use the resource.

And now, I'm done talking. I hope that what I've written has been in some way useful to you. Have fun. Happy BlueSky-ing.