💾 Archived View for jacksonchen666.com › posts › 2023-08-27 › 00-35-14 › index.gmi captured on 2024-07-08 at 23:40:23. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-11-04)
-=-=-=-=-=-=-
2023-08-27 00:35:14Z (last updated 2024-03-18 07:35:41Z)
CORRECTION (2024-03-18T07:35:41+00:00): I previously said that it was basically impossible to move around in the alt text word by word in VoiceOver and possibly other screen readers as well. However, I forgot about Rotors in VoiceOver, which do in fact allow you to move around in alt text word by word, character by character, and by other semantics.
It is also possible to move around alt text in Tusky with TalkBack, so other screen readers do indeed support moving around in alt text, not only VoiceOver.
Also, *this blog post is coming from a person who's not visually impaired at all* (apart from slowly worsening shortsightedness). Therefore, I do not recommend linking this blog post further. Instead, speak to people who's actually visually impaired or uses a screen reader.
With the pretty big usage of Mastodon (including the Fediverse as well), there's quite a big culture towards accessibility too. That's great!
Except... The results aren't exactly the best.
Sure, people are now describing the images that they post (at least, most people do that, some just don't do it for any number of reasons). But it's kind of problematic for screen readers.
I'm not saying "don't use alt-text", in fact, you should add alt-text to images even if you're not on the Fediverse (e.g. writing an HTML page).
Seirdy's blog post on inclusive accessibility has a section on "going beyond" alt-text. To over-summarize the good practices:
1. Conciseness
2. Under about 100 characters
3. Separate information which isn't in the image
Seirdy's blog post, "Beyond alt-text"
Now, who's going to get every single one of those (not set in stone) practices?
https://mastodon.scot/@bodhipaksa/110956786907386848
Not the above linked post on a Mastodon instance. It features 2 images, and the contents of the post is:
Reposting this with alt text. In fact the entire description is in the alt text.
And yes, I checked, the description is part of the alt-text.
So, compare that to the guidelines above:
1. It's not quite concise in my opinion, because it goes to describe the picture in detail.
2. The first image also contains alt-text with 378 characters, which is quite far away from 100 characters.
3. Information outside of the image is not separated from alt-text.
Now, let me sit you through an experience called: Reading what VoiceOver said for the alt-text.
Here's what VoiceOver said for the first image's alt-text:
Exhibit by Michael Rakowitz in Vienna. A car-shaped tent has been constructed by the roadside, occupying a parking space. Someone in a yellow sleeved jacket is leaning out, fastening a zipper. In the background, pedestrians are walking on the sidewalk. Space.
(VoiceOver adds the word "space", they're not literally part of the original alt-text, just included it in for transcript completeness)
Here's what VoiceOver said for the second image's alt-text:
Exhibit by Michael Rakowitz in Vienna. A split image. On the left, the person looks out of the tent door. In the second, the person sits inside the tent, on the cobblestones, reading a newspaper. Space.
We're conditioned to see this behavior as selfish and as taking public space away from others, and yet if they were in a car most people would think nothing of the same behaviors.
Exhibit by Michael Rakowitz in Vienna. A split image. On the left, the person looks out of the tent door. In the second, the person sits inside the tent, on the cobblestones, reading a newspaper.
We're conditioned to see this behavior as selfish and as taking public space away from others, and yet if they were in a car most people would think nothing of the same behaviors.
Notice that on the second image, the alt-text has been repeated. More on how that happened later.
Now, here's further issues (or potential issues) with putting all the information in alt-text:
Further alt-text/`title` attribute related (potential) issues:
For some potential solutions (as a non-expert in everything) to try to fix some of these problems (potentially poorly):
1. Make the textbox a text input, making it single line instead of multiple lines, to discourage long alt-text.
2. Further limit the amount of characters available for alt-text. Won't work well for images of just text.
3. Maybe only use alt-text for alt-text, not `title` as well!
4. Add captions for stuff related to the image, shown close to the image. Keep alt-text for actual alt-text.
Of course, those solutions might be a HORRIBLE idea, and I'm not a super expert in accessibility. There seems to be great recommendations on writing alt-text too, so here's some links to recommendations with alt-text writing:
https://accessibility.huit.harvard.edu/describe-content-images
https://axesslab.com/alt-texts/
https://seirdy.one/posts/2020/11/23/website-best-practices/#beyond-alt-text
(Note to Gemtext readers: I have replaced the audio part with just the transcript. Hugo make it a pain to manage files outside of HTML outputs basically, still haven't figured out shortcodes so it's not coming to here.)