Tuned out from HTML@W3C working group for a few months. (I was very busy at work, and participation in the working group is entirely a free time activity.) Ran across this long and apparently continuing argument that seems to boil down to whether <img> tags should be required to have an alt attribute to be accepted as HTML5 conformant. One or more of the guys associated with accessibility at their place of work seem to be arguing for mandatory alt usage.

Just to be clear, if the writer is not attempting to make a document accessible, then all we need for HTML is:

    <img src="a.png">

The above is an image with no extra information for accessibility.

All of the following is a waste of time:

    <img src="a.png" noalt>
    <img src="a.png" alt="">
    <img src="a.png" alt=".">
    <img src="a.png" alt="*">
    <img src="a.png" alt="boilerplate-put-in-only-because-it-is-required">
    <img src="a.png" alt="not-very-meaningful-text">
    <img src="a.png" alt="downright-misleading-text">

The above is ritualistic waste - about the only good it will do is give global warning a tiny boost, and perhaps delay the next ice age.

Frankly, I do not care about accessibility. Except when I do.

When I do not care about accessibility, if an HTML5 conformance tester bugs me until I waste time putting in bogus attributes, I am going to regard HTML5 conformance as a somewhat bogus waste of time. The vast majority of web pages have been and will be written from this frame of mind. Any attempt to force accessibility on this population is going to fail. Worse - attributes added for HTML conformance, but lacking meaningful human-semantic value, are likely to make pages a bit less accessible.

When I do care about accessibility, I need help - big time. Somehow I doubt that just filling in a few attributes is going to do the trick. I have never been part of an organization big enough to have folk dedicated to accessibility. That means I am pretty much on my own. Sure, I could go through and mindlessly fill out a bunch of alt attributes. I am sure this will make some bureaucrat happy, but I doubt the result will be very good. What I really need is something that will let me test the page as a whole, and give me some insight into how alternate browsing tools present the page.

Also it always kind of bugged me that the text inside an alt attribute could not be styled. Seems like the text should be treated like other text. Sometimes a little emphasis in the right place goes a long way.

This argument is a waste of time. Requiring alt attributes when the writer is interested in accessibility is a waste of time. Making HTML documents non-conformant because they lack meaningless boilerplate is a waste of time.

A test for accessibility (usable without expertise in accessibility) is a great goal, but pretty near independent of what should be the main concerns of the HTML working group.

Accessibility is a noble goal. You cannot force people to be noble.