Re: Moving longdesc forward: Recap, updates, consensus

On Fri, May 6, 2011 at 7:11 PM, John Foliot <> wrote:
> The "Hidden Metadata" argument has been continually asserted by those who
> are opposed to reinstating @longdesc as a flaw with the attribute. It isn't,
> and we need to ensure we say so clearly and forcefully.

The "Hidden Metadata Fallacy" does neither.

> We have a number of editing tools that expose the functionality/ability of
> associating a longdesc html file or ID fragment to an image both at original
> authoring time, as well as at editing (change) time. We have a number of
> user agents (both visually based and screen readers) that expose the
> presence of @longdesc to the end user, and accessing the value of that data
> is as fundamental as it gets on the web: follow the link.

I guess your point is that improving @longdesc's discoverability reduces
the costs of it being hidden. I think that's a good argument to make, but
it does not make the hidden metadata argument fallacious, just less damaging.

> Using the Hidden Metadata argument, it can be argued that microdata and RDFa
> data is "hidden", as it is not usually visually apparent on the screen to
> sighted users.

This sounds very similar to your erroneous argument about microformats:

The general idea of both microdata and RDFa is to mark up visible data not to
hide metadata.

Both microdata and RDFa provide features for providing machine-friendly
versions of human-friendly visible data that *could* also be used to create
hidden metadata that has no visible representation. Microdata and RDFa differ
from microformats only in that they do not explicitly prohibit this; it is
not intended to be the typical case.

At a slight tangent, I've been trying to explain this in the rather
different context of annotating historic texts with editorial information
over on the WHATWG list:

> With the recent evolution of touch-screen devices (and Steve Faulkner's
> recent enquiries to browser vendors
> it could
> also be argued that the value of @title is quickly becoming "hidden"
> metadata, as no touch-screen device that I am aware of today expose @title
> values to *any* of their users: using the "Hidden Metadata argument" perhaps
> we should drop @title from the spec too?

The poorer accuracy of hidden metadata is certainly a good argument against
keeping @title conforming.  But, as with @longdesc, @title may have merits that
counterbalance it being hidden metadata.

Benjamin Hawkes-Lewis

Received on Friday, 6 May 2011 20:54:46 UTC