The holy wars of the humble alt
Sunday, 7. September 2008, 11:29:19
- Requiring ALT attributes hurts accessibility when tools or people insert empty or silly ALT attributes for significant images to pass validation.
- Requiring ALT attributes helps accessibility when people educate themselves due to validation warnings and add useful ALT-contents, and when companies have to use ALT correctly due to legal threats.
Which effect is stronger and more valuable? This - in a nutshell - is what the ALT warriors should be discussing and researching. For example, there is some research here that indicates even low quality ALT text can be more helpful than none.
There is an incredible amount of time being spent on largely irrelevant arguments - from both sides. For example
- Mass uploading images (e.g. Flickr) means people don't have time to write ALT texts, so Flickr can't require users to do so.
This is just a matter of how conscientious the user is. Flickr should have an optional "description for blind people" box, users who did not fill in details would cause Flickr to output invalid HTML without ALT. So non-conscientious users create invalid HTML - to me, that sounds like a fact of life, not a spec bug.
- Future software may be able to analyse images automatically and provide alternative text if ALT is omitted.
Yeah, I'll believe in this intelligent software when I see it - and in particular I find it hard to believe that such hyper-intelligent software can't be configured to generate ALT-text for ALL images regardless of supplied ALT.
- Requiring ALT means blind users can not upload images they don't know the contents of.
Um, and uploading images you don't know the contents of is a major use case? I sort of can't see the motivation of the user for doing that.
Then there is a lot of hot air and big egos or at least claims that other people have big egos, claims that whoever disagrees with you is arrogant and doesn't listen etc.. I admit that the quality of discussion on the public-html W3C list disappoints me.
The current spec text seems pretty good. I think that in the interest of getting rid of invisible meta data and user-targeted text in hidden attributes it might also say that ALT can be omitted if aria-describedby points to some element in the page.
Now please move on to the next topic..








