RE: Thoughts on professional portrait (headshot) images?

“In cases of doubt I believe the user, not the author, is the one who should decide if an image provides useful information for them”

Your argument is inconsistent. You say “we cannot know what an author's intention was”, yet you claim to somehow know what users want. In most cases you actually can ask the author what their intention was – we frequently do that. But it’s virtually impossible to ask users what they want unless you go to the expense of conducting some kind of user testing or survey. Have you ever asked a user what they want while conducting a WCAG audit?

Screen reader users have a trade-off between the speed of task completion and their ability to read all the content. It takes way too long to read everything. My experience is that the majority choose navigation and reading strategies that optimise speed while running the risk that they may miss something important, just like when sighted people skim read a document. Adding incidental fluff in “alt” attributes makes the job more difficult for them.

Steve


From: Guy Hickling <guy.hickling@gmail.com>
Sent: 23 August 2022 20:18
To: WAI Interest Group discussion list <w3c-wai-ig@w3.org>
Subject: Re: Thoughts on professional portrait (headshot) images?

> My view is that the author’s view should prevail. If the author says the image is decorative, then that’s what it is.

But in many cases (maybe most cases), we cannot know what an author's intention was when they display some content. All outsiders (i.e. anyone not the author) have to work on is the content itself, and the rest of the surrounding web page as the context. They cannot know the author didn't intend them to take in any information they contain. Accessibility testers, likewise, can't second-guess the mind of the author, as to whether they left the alt text blank because, like too many developers, they don't understand (or care?) about screen reader users, or because they genuinely considered it decorative.

In cases of doubt I believe the user, not the author, is the one who should decide if an image provides useful information for them. Some images that many of us (including the original developer) might dismiss as decorative, other sighted users may nevertheless find they provide interesting context. Sighted users can make that decision for themselves.

Blind people should have the same right to decide too - but when we put an empty alt attribute on something, we take away that right to decide from them because they will no longer be told an image even exists! So we need to be quite sure that something really does not provide any information, i.e. is truly just decorative, before we decide to take away that right. I believe that when SC1.1.1 says "the equivalent purpose", it means the purpose of the image as it actually appears on the page, not any nebulous guess at what the author might have intended. So the Understanding document describes the intention of this SC as "The intent of this Success Criterion is to make information conveyed by non-text content accessible through the use of a text alternative." - no reference to the author, only to the actual content as it appears on the page.

In the case of headshots, a lot of people can obtain a lot of useful information from these - how old they are, what gender, what ethic background, and so on. Even if we don't find it easy to include all that in the alt text, blind people can ask a friend what the person in the photo looks like. But if we don't tell them an image is there, they can't.

I recently audited a university website. Many hero images on various pages show some really beautifully designed buildings, but because they are hero images they were all given empty alt texts. There was, in fact, not a single screen reader announcement, anywhere, of what the buildings looked like, although sighted people got to see them all! So I recommended that, in a few of the images showing the best views, they should add good alt text descriptions, even though not particularly relevant to the adjacent text, just so blind people could be given some idea of what these buildings looked like. The buildings may not have been relevant to adjacent text, but they are certainly relevant to anyone interested in the website and the university as a whole.