longdesc in the wild [was: Re: Is longdesc a good solution? ...]

> *If* there's as much noise in longdesc as http://blog.whatwg.org/the-longdesc-lottery
>  says, it would probably be better not to grandfather it.

Probably not.

It says 96% are in error, but ... none of those are clear errors.

When longdesc first came out, it was often described as a longer alt,
and filled with literals.  <... alt="lake" longdesc="Sunset Lake,
Crystal Falls, Michigan, viewed from the North at 3:30pm, showing the
park on the left, and water lilies on the right." ...>

This clearly isn't an URL, so it contributes to the 96% error rate,
but simply treating it as a literal does provide real value.  I'm
pretty sure that at least some early browsers (lynx?) (previously?)
did, because I remember reading such descriptions at the time.

Other "errors" include longdesc="" or pointing to the whole page --
both of which seem as likely to be legitimate (and harmless) as
alt="".

> see also http://lists.w3.org/Archives/Public/public-html/2007Sep/0350.html)

This is more convincing, but it speaks only to uselessness, not
poisoning.  Several people have said that the newest tools are the
first that do support longdesc well, so it may be that we have ~1 year
of history and ~12 years of magic talismans -- with that study
conducted during the magic talisman period.

-jJ

Received on Tuesday, 9 September 2008 20:31:19 UTC