Sign in

Eric Bailey

@ericwbailey.website
9.4K followers 2.4K following 1.9K posts

#accessibility / #a11y advocate, post enjoyer. 🌐 ericwbailey.website ✉️ ericwbailey.website/newsletter

PostsRepliesMedia
Eric Bailey @ericwbailey.website · 2h
Finally: AI usage. Referencing people wanting more accessible sites over apps, a lot of LLM work seems to be geared toward both addressing the bad alternative descriptions we write, and helping to traverse barriers and friction. Still thinking through this one, but glad it was collected and shared.
A table with 3 columns and 9 rows that is titled, "In what ways do you commonly use AI to improve your online experience?" The first row is the table header row, and its columns are: "Response", "# of Respondents", and "% of Respondents". Row 2 is "To generate image descriptions / alternative text" with 1069 respondents at 60.1%. Row 3 is "To provide in-depth analysis or description of complex images" with 885 respondents at 49.7%. Row 4 is "To generate text or other alternatives for multimedia content" with 510 respondents at 28.7%. Row 5 is "To generate audio alternatives of text content" with 318 respondents at 17.9%. Row 6 is "To provide summaries of webpages or documents" with 660 respondents at 37.1%. Row 7 is "To provide guidance in performing an online task" with 722 respondents at 40.6%. Row 8 is "To perform accessibility evaluations of content" with 347 respondents at 19.5%. Row 9 is "Other" with 290 respondents at 16.3%. Cropped screenshot.
000
Eric Bailey @ericwbailey.website · 2h
Yes, but also consider the language the survey is in, the language of the channels it is promoted in, the primary language of the communities it is shared in, etc.
110
Eric Bailey @ericwbailey.website · 2h
oh no
000
Eric Bailey @ericwbailey.website · 2h
PDFs continue to be absolute garbage trash. No need to use a pie chart to communicate that. webaim.org/projects/scr...
webaim.org
WebAIM: Screen Reader User Survey #11 Results
130
Eric Bailey @ericwbailey.website · 2h
For Finding Information, it continues to be headings as majority. I hate that I keep having to train people about headings. webaim.org/projects/scr...
A pie chart with 5 values: "Navigate Headings", "Read the page", "Use 'Find'", "Navigate Links", and "Navigate Landmarks/Regions". Navigate Headings occupies 67.7%, Read the page occupies 9.7%, Use "Find" occupies 12.5%, Navigate Links occupies 6.7%, and Navigate Landmarks/Regions occupies 3.3%. Cropped screenshot.
101
Eric Bailey @ericwbailey.website · 2h
And speaking of LLMs and the magical thinking that AI will "solve" accessibility: The people just want more accessible websites, not some hypothetical savior super app. webaim.org/projects/scr...
A pie chart with two values: "Better assistive technology" and "Better (more accessible) web sites". Better assistive technology occupies 83.9%, and Better (more accessible) web sites occupies 16.1%. Cropped screenshot.
154
Eric Bailey @ericwbailey.website · 2h
Yeah, that does make sense to me. So much nuance in this data.
010
Eric Bailey @ericwbailey.website · 2h
💜
000
Eric Bailey @ericwbailey.website · 2h
For Web Accessibility Progress, I think "Respondents without disabilities tend to be more positive about recent progress (41% thought the web had become more accessible) than those with disabilities (31.8% thought it had become more accessible)." says it all. webaim.org/projects/scr...
A pie chart with 3 values: "More Accessible", "No Change", and "Less Accessible". More accessible occupies 32.2%, No Change occupies 44.9%, and Less Accessible occupies 22.9%. Cropped screenshot.
121
Eric Bailey @ericwbailey.website · 2h
Yep! I'm a big proponent of progressive enhancement, but wanted to leave it separate to this thread in that I've seen consultants demand the entire experience needs to work without JS enabled in order to "be accessible". It's a non-starter and hurts the overall perception of the practice.
110
Eric Bailey @ericwbailey.website · 3h
100% conjecture on my part, but I wonder if the increase in LLM-powered coding has contributed towards people moving away from mobile apps. LLMs produce inaccessible code by default, and the web is far more malleable—barriers can be overcome via extensions and the like. webaim.org/projects/scr...
A pie chart with two values: "Mobile app" and "Web site". Mobile app occupies 54.8%, and Web site occupies 45.2%. Cropped screenshot.
110
Eric Bailey @ericwbailey.website · 3h
People use screen readers on their phone, another thing I wish more people knew. WebAIM's comment on the correlation between proficiency and mobile OS is also worth noting. Cheap phones run Android, and disabled people are more likely to live in poverty. webaim.org/projects/scr...
A pie chart with four values: "Apple iPhone, iPad, or iPod touch", "Android", "Chrome OS", and "Other". Apple iPhone, iPad, or iPod touch occupies 73.3%, Android occupies 24%, Chrome OS is a hair-thin sliver, leaving Other at roughtly 2%. Cropped screenshot.
120
Eric Bailey @ericwbailey.website · 3h
SAME
100
Eric Bailey @ericwbailey.website · 3h
The main reason for use is a good reminder that at the end of the day it is people who use the technology. The same reason why you're loathe to change your browser is the main reason why most people stay with the screen reader they are most comfortable with. webaim.org/projects/scr...
A pie chart with 5 values: "Existing Comfort/Expertise", "Features", "Availability", "Cost", and "Support". Existing Comfort/Expertise occupies 49.3%, Features occupies 24.3%, Availability occupies 11%, Cost occupies 9.3%, and Support occupies 6.1%. Cropped screenshot.
130
Eric Bailey @ericwbailey.website · 3h
Browsing with JavaScript enabled remains an overwhelming majority. I'm including it here so you can tell a supposed accessibility "expert" that they can pound sand if they parrot the antiquated myth that it needs to work without JavaScript to be accessible. webaim.org/projects/scr...
A pie chart with two values, "Yes" and "No". Yes occupies 99.7%, and No occupies 0.3%. Cropped screenshot.
220
Eric Bailey @ericwbailey.website · 3h
Browser used is directed more towards auditors and testers. Some of the old "canonical" screen reader and browser pairings might not be as commonplace anymore, so testing scripts might need to be updated to more closely reflect reality. webaim.org/projects/scr...
A pie chart with 5 values, "Chrome", "Microsoft Edge", "Firefox", "Safari", and "Other". Chrome occupies 52.3% of the pie chart. Microsoft Edge occupies 23.6%, and Firefox occupies 15.9%. The other two slices don't have visible percentages, with Safari and other's combined slices being about half the width of Firefox. Cropped screenshot.
120
Eric Bailey @ericwbailey.website · 3h
(I will leave the politicking of JAWS and its apparent increasing usage out of this thread)
100
Eric Bailey @ericwbailey.website · 3h
If you're on a desktop or laptop, it's going to be Windows. This is incredibly important in that JAWS and NVDA have different modes of operation compared to VoiceOver. Most people only test on macOS out of convenience, but it is not reflective of majority experience. webaim.org/projects/scr...
A pie chart with 5 values, "JAWS", "NVDA", "VoiceOver", "ZoomText/Fusion", and "Other". JAWS occupies 55% of the pie chart. NVDA occupies 32.9%, and VoiceOver occupies 6.5%. The other two slices don't have visible percentages, with "Other" being the slightly larger than "ZoomText/Fusion", roughly half of the size of the VoiceOver slice. Cropped screenshot.
130
Eric Bailey @ericwbailey.website · 3h
Same general thoughts for internet proficiency. People who are more online probably have a better chance of being aware of, and filling out this survey. webaim.org/projects/scr...
A pie chart with three slices. Advanced is the first slice, occupying 64.6%. Intermediate is 33.0%, and Beginner is 2.4%. Cropped screenshot.
110
Eric Bailey @ericwbailey.website · 3h
Self-reported proficiency is always a fun one. Setting selection bias aside for the moment: It's a good reminder that just because something can technically be accessed and operated via some obscure command that it doesn't necessarily mean someone knows about said command. webaim.org/projects/scr...
A pie chart with three slices. Advanced is the first slice, occupying 57.8%. Intermediate is 38.8%, and Beginner is 3.4%. Cropped screenshot.
100
Eric Bailey @ericwbailey.website · 3h
I'm glad they capture disability types. It's a great reminder that screen reader use is not exclusively for fully blind individuals, and also not exclusively for vision-related disabilities. I wish more people knew this. webaim.org/projects/scr...
A bar chart. The y axis is 0% through 100%, with stops that increment by 25. The x axis values are "Blindness", "Low Vision/Visually Impaired", "Cognitive", "Deafness/Hard-of-Hearing", "Motor", and "Other. Blindness' value is 83.2%. Low Vision/Visually Impaired's value is 18.0%. Cognitive or Learning's value is 3.0%. Deafness/Hard-of-Hearing's value is 7.5%. Motor's value is 1.9%. Other's value is 3.0%. Cropped screenshot.
372
Eric Bailey @ericwbailey.website · 3h
For self-reported disability, I hope the data of the 5.2% of not identifying was discarded. I view this survey as collecting information from daily users, information from people who don't rely on the technology could unnecessarily skew the results. webaim.org/projects/scr...
A pie chart showing 94.8% of respondents self-identifying as disabled, and 5.2% not self-identifying as disabled. Cropped screenshot.
100
Eric Bailey @ericwbailey.website · 3h
For region, a reminder that it's the World Wide Web, not the wealthy Western web. This breakdown is a complicated mix of factors—too large for Bluesky. Here, we should not forget at the end of the day that it's people who use this technology regardless of location. webaim.org/projects/scr...
A pie chart showing regional breakdown. North America occupies 54.1%, Europe occupies 27.7%, then Asia, Africa/Middle East, South America, Australia and Oceania, and Central America and Caribbean occupy progressively narrower slices. No percentages for them are displayed. Cropped screenshot.
220
Eric Bailey @ericwbailey.website · 3h
The 11th edition of the @webaim.bsky.social Screen Reader User Survey is now live! This is a vital resource into better understanding how people who use screen readers actually operate your website or app. webaim.org/projects/scr... My commentary on it as a thread 🧵 #a11y
webaim.org
WebAIM: Screen Reader User Survey #11 Results
12117
Reposted by Eric Bailey
Lynn Stahl @msbtterswrth.altgov.info · 29/09/2026
A letter from us. We want the best for USWDS, and we're ready to help. fedscoop.com/websites-jeo...
fedscoop.com
Websites ‘in jeopardy’ amid U.S. Web Design System takeover, coalition tells GSA
The letter comes after administration officials changed leadership for the system, implemented new AI tooling, and shut out members of an open-source community.
27239
Eric Bailey @ericwbailey.website · 29/09/2026
Joe Gebbia has a lot to answer for
110
Eric Bailey @ericwbailey.website · 29/09/2026
tfw you know more effort went into aligning the input and container border radii than 90% of the underlying infrastructure
The America.gov landing page chat input with 100% rounded corners, placed in a square container whose borders match the curve. Inside the square is a stock photo of a man on an escalator. Above it is the headline, "Hello, America." and a subhead of "Whatever you need from government, start here." Cropped screenshot.
7445
Reposted by Eric Bailey
Jessica Ellis @baddestmamajama.bsky.social · 29/09/2026
I have separated the art from the artist and am now whipping the art to stiff peaks, after which I will gently fold it into the artist before baking at 350 til golden and puffed
3444571066
Eric Bailey @ericwbailey.website · 29/09/2026
Maybe? The accName problem is roughly: - People use title to create a tooltip. - That tooltip can be supplementary content to the control’s accessible name. - title can override the intended accName. - This is confusing for both screen reader and voice control users.
010
Eric Bailey @ericwbailey.website · 28/09/2026
If I could snap my fingers and change it overnight, I'd alias it to a popover, with proper keyboard handling and focus baked in. It would wreck iframe backwards compat, but maybe you just can't give an iframe a tooltip or the like.
150
Eric Bailey @ericwbailey.website · 28/09/2026
Their text content can't be resized, too. It's what I was excited about with your post.
020
Eric Bailey @ericwbailey.website · 28/09/2026
I mean, we'd need to fix the accName problem, but also a boy can dream.
230
Eric Bailey @ericwbailey.website · 28/09/2026
god, it would fix so many WCAG issues
180
Eric Bailey @ericwbailey.website · 28/09/2026
mine is just 16 motion activated Halloween decorations you get at CVS
140
Reposted by Eric Bailey
Anna E. Cook @annaecook.com · 27/09/2026
Once again the Trump regime is attempting to skirt federal laws and hope AI can just magically do things. This will not work but they do not care. fedscoop.com/trump-team-a...
fedscoop.com
A Trump team wants to automate U.S. Web Design System accessibility testing. Experts are worried.
GitHub community and accessibility experts are concerned about a reversal of progress amid potential AI reliance and plans to “automate as much accessibility testing as possible.”
32110
Reposted by Eric Bailey
Manuel Matuzović @matuzo.at · 28/09/2026
This advent calendar is your chance to share your knowledge with a large audience. You have 5 more days to submit your ideas. First-time authors are welcome!
01720
Eric Bailey @ericwbailey.website · 28/09/2026
you published it, yessssss
010
Reposted by Eric Bailey
Josh Black @josh.black · 27/09/2026
Finally got a post up on the GitHuh blog github.blog/engineering/... 👀
github.blog
Improving site performance by shipping more CSS
How we fully migrated github.com away from CSS-in-JS.
4203
Eric Bailey @ericwbailey.website · 25/09/2026
I am also proud to help facilitate an event that is openly progressive at a time when it is not in vogue to do so.
060
Eric Bailey @ericwbailey.website · 25/09/2026
Breaking out of front-end content also foster awareness of new considerations. This helps to remind attendees about the reason behind why we do all this: The people who benefit. inclusivedesign24.org/2026/schedul...
inclusivedesign24.org
Inclusive Design 24 (#id24)
A free 24-hour online community event on accessibility and inclusion
130
Eric Bailey @ericwbailey.website · 25/09/2026
We strive to provide a diversity of both voices and topics. Having a conference talk on your résumé can open doors. That can be huge if you are just getting started in your career. 24 talk slots means there's a lot of chances for that to happen.
130
Eric Bailey @ericwbailey.website · 25/09/2026
Now that hosting is done and I've got a night's sleep and the adrenaline past me, a little love letter to Inclusive Design 24: The goal of the conference isn't profit or advertising, it's access, education, and enablement. #a11y #id24 inclusivedesign24.org/2026/schedule/
inclusivedesign24.org
Inclusive Design 24 (#id24)
A free 24-hour online community event on accessibility and inclusion
1142
Eric Bailey @ericwbailey.website · 25/09/2026
overlay trash
000
Reposted by Eric Bailey
Anthony Moser @anthonymoser.com · 23/12/2025
As a starting point I think the CEO of every company offering chatbots should be *personally* liable for all the text their chatbots produce. as though it was a direct conversation with them
3406102
Eric Bailey @ericwbailey.website · 24/09/2026
"They should dedicate themselves to engaging and listening to the disability community about what problems need solving and what does and doesn’t work." #a11y newmobility.com/nobody-asked...
newmobility.com
Nobody Asked for a Crab Chair
In our last installment of “Good Intentions Gone Wrong: When Disability Engineers Go Bad,” we highlighted a particularly egregious Swiss wheelchair design featuring superfluous drones. Thankfully…
0104
Reposted by Eric Bailey
Ahmad Shadeed @ishadeed.com · 24/09/2026
✍️ New Article: Detect when elements overlap with CSS Can CSS tell when two elements overlap? At first, I thought this was not possible, but it turned out it can be done with scroll-driven animations and anchor positioning. ishadeed.com/article/css-...
ishadeed.com
Detect when elements overlap with CSS
Detecting when two elements overlap using only CSS anchor positioning and scroll-driven animations.
57419
Reposted by Eric Bailey
Inclusive Design 24 (#id24) @inclusivedesign24.bsky.social · 24/09/2026
Our first Inclusive Design 24 talk is a panel discussion with Lainey Feingold, Léonie Watson, Shilpi Kapoor, Meena Das, Jutta Treviranus, and Sheri Byrne-Haber, all speaking about Digital Accessibility Ethics as a a new framework to advance accessibility. www.youtube.com/watch?v=gVTG... #id24 #a11y
youtube.com
Panel: Digital Accessibility Ethics: a new framework to advance accessibility #id24 2026
This session will introduce the Digital Accessibility Ethics Framework and show how it is applied to several aspects of tech and accessibility. These include procurement, development, leadership,…
034
Eric Bailey @ericwbailey.website · 23/09/2026
it’s so good for booping!
010
Eric Bailey @ericwbailey.website · 23/09/2026
she baby
010
Eric Bailey @ericwbailey.website · 23/09/2026
A dog sitting in a chair and using a computer looking down at another dog and saying, "On the Internet, nobody knows you're a dog." Black and white comic.
110