There have been some things about alt text and the way it’s used on BlueSky that I’ve been thinking about writing down for quite some time now. Unfortunately, I’m pretty lazy about stuff like that, so all the shit in my head has been swirling around for quite some time and it’s clogged up my brain a bit. So what happens is stuff like this: I’ve got a few notes about what you’re about to read jotted down in a note on my iPad. Then while my wife and I were walking around the supermarket, I had a fabulous idea about how I was going to actually start this post and pull those notes into something coherent. Did I write the idea down? I DID NOT. Did I forget it almost immediately? I DID. I guess this is what you get instead.
alt text: it’s EASY fucking HARD sometimes
I have said this before, and I will say it again. Alt text is an important accessibility tool. Without it, people who use screen readers are cut off from all the visual content that provides context and support for written information online. A picture’s only worth a thousand words if you can actually understand what’s depicted. So, to include as many people as possible in our conversations and to facilitate the clearest communication, it’s important that everyone gets to work with the same information. And it’s not just users of screen readers who need that context. It’s folks who are neurodivergent and it’s people who have sensory processing issues, and it’s people who may have limited experience dealing with visual context cues, and the list goes on. I’m a person who on a good day has a cultural literacy score that hovers someplace around a two out of ten…so for me, any context I can get out of alt text is awesomely helpful. So my motives are sometimes pretty selfish when it comes to my hope that more people attach alt text to the images they post.
But there is also this, and this is also something I’ve said before. It’s important to remember that it’s not always that easy for some people to just type some fucking alt text and that it doesn’t always make sense to be aggressive about the way we approach people who don’t always add descriptions to their posted images. Conflicting needs exist for conflicting disabilities, and the do better, you asshole approach isn’t always necessarily the right call. Some people definitely deserve to hear it in that specific way, but maybe that shouldn’t be the default assumption.
yes, the alt text you put on your video is there
You just can’t see it. Screen readers can see it, but it’s not easily accessible to human eyes either in the browser or in the app. I don’t know how easy it might (or might not) be to put a similar text overlay over video as the one that’s displayed over images, but for now the alt text is for all intents and purposes invisible. Sorry about that. But hey…since you can’t have alt text, you know what WOULD be cool to include in your videos? Hear me out: captions. Instead of getting super-technical in this post about what those are, what they should look like, and how to generate them, I’m going to point you to a couple of places. Official government entities can be vocal about these kinds of things these days, so there are plenty of resources to check out. Caveat: These are North American-based links. Guidelines elsewhere will most probably differ from what I’m listing here.
- (US) Department of Education’s Described and Captioned Media Program Guidelines (related mostly to educational video, but it’s an exhaustive listing of requirements that can be used to create highly visible captions)
- (US) FCC Caption Guidelines
- (US) Web Content Accessibility Guidelines
- (US) Americans With Disabilities Act Rules and Guidelines
- (CA) CRTC Guidelines for TV Captioning
- (CA) Closed Captioning Standards and Protocol for Canadian English Language Broadcasters (this is a PDF, not a web page…but it’s an exhaustive resource).
don’t tell anyone i said this
Screen readers on some mobile platforms/devices and some applications for your computer can recognise and read the text in images. Some can also describe images without text. Do not–for the love of whatever you deem holy–take this as me saying that screen readers can see and describe everything in images these days and that you’re now free to give up using alt text because the machines can do it themselves. I am in fact saying the exact opposite. You should continue to use alt text as much as you can, according to your ability to do so. But while you’re here, let’s look at a few things I found out. I only know about Windows, iOS, and Android…so please forgive my ignorance and please do some research for yourself.
Windows: According to the User Guide, NVDA (Non-Video Desktop Access) version 2025.2 can leverage the OCR capabilities in Windows 10 or newer to recognise text from images. You should always read these guides carefully to understand the limitations of the tools and to get an idea of what you might be able to expect if you decide to use such a thing. I have not tested this myself, so I have no idea at all how good it is, nor how it works.
Apple: Apple’s native screen reader is called VoiceOver, and it can be enabled in the device Settings under the Accessibility sub-menu. Apple will tell you that their VoiceOver Recognition will work on certain devices that run iOS 14 or newer. I can’t test it on anything older than an iPhone 13 Pro, but I can say that it does work without having to rely on Apple Intelligence. I can also say that not only does it read text from images, it also can provide brief image descriptions. I haven’t tested it extensively, but it does seem to be fairly accurate with uncomplicated images.
Android: I have a Samsung Galaxy A53 5G, which is a couple of years old but still capable of running Android 15. The native Android screen reader is called TalkBack, and version 15.1.02.0 (which lives on my device) is definitely capable of reading text from unlabelled images. This was apparently introduced in Talkback 13.0. And it gets better, maybe? In version 14.1 of TalkBack, an image description capability was introduced. Sadly, it’s beyond the power of my simple device to access that functionality. From what I’ve been able to find, it’s flagship phones like the Pixel series that get those abilities.
Again, just because those things are there doesn’t mean you should give up writing alt text. And yet again for y’all in the back (because I see you shady motherfuckers back there): do not give up writing alt text. OCR is only truly accurate if the text is formatted in a way that the OCR engine can understand. Anything other than a straightforward text box can be a comprehension challenge. Human interpretation of a digital graphic will almost always provide a much more accurate interpretation of the more entertainingly-formatted text blocks in such a graphic than OCR can. And there’s also this: image description engines typically have difficulty understanding complex images. They may not be entirely trustworthy when it comes to giving accurate data. Bottom line? Human generated descriptions are still the best way to get alt text attached to images online.
what about ai? am i training ai with my alt text?
Who cares? I mean, I guess some people care, but it’s a choice. On the one hand you can worry about whether or not your image description is gonna help or confuse whatever LLM can scrape this data. On the other, you can worry about whether or not your image description is gonna be a thing that helps someone to engage socially on a social media site. Aren’t choices cool?
it’s the end of this post (as we know it) and i feel fine
As usual, I hope that this has at least been somewhat helpful. Maybe you’ve even *gasp* Learned A Thing.
Thanks, y’all. Have fun. Happy BlueSky-ing.