RE: Another summary of alt="" issues and why the spec says what it says

Hixie wrote:
 
"But as pointed out on that thread and in others, sometimes there simply is
no alternative text available. Examples I can think of that have been
mentioned so far include:
[several use cases]"

Actually, as may not have been obvious, I was agreeing with all these use case and thereby agreeing with your decision not to require "alt", albeit in a roundabout way. Requiring an html-author to provide an "alt" in lieu of the image-artist's provided metadata could open liability exposure to browser manufacturers. That is, the naive html author is forced to improvise about imagery data,  which to them may be invisible, but which to software is definitely not invisible.
 
Ultimately there are at least three sources of data pertaining to what an image is:
1. the pixels or vectors themselves
2. any descriptions about those pixels or vectors (or any subsets thereof) made by the author/copyright holder
3. any descriptions about those pixels or vectors (or any subsets thereof) made by the html-author (presumably licensed or otherwise legally empowered) 
(of course, then,  we also have the excitation of the bipolar and ganglion cells in the retina of the perceiver -- but those are generally viewed as outside the scope of HTML)
 
One can easily argue (and it is not clear to me that US case law since Kelly v Arriba resolves otherwise) that inappropriate metadata provided by those other than the copyright or trademark holder could result in some form of legal exposure to the html-author. By compelling html-authors to expose themselves to such liability, the browser makers would possibly participate in contributory infringement or in contributory dilution (if there is such a thing). I note that Google did not historically enter into image harvesting until after ditto.com's legal issues had been (more or less) favorably resolved by the above precedent.
 
Maybe we need two different alt tags: one for the data provided in-context by the html-author-who-uses-the-image and another for creator-and-putative-copyright-holder-of-the-image. While the various disabilties-related acts within the world might complain if we do not make some valid data (of whatever origin) available to the end user, the various IP laws within the world might sound alarms if we make improper data emanating from non-original sources available. This is what leads to my suggestion that browsers should (maybe at some point in a foreseeable future) read existing metadata that may be embedded within the data. Until we are able to read and utilize "authoritative" data from the primary author, it seems risky to compel the User Agent to reject pages that have no alt tags, since those very alt tags might contain spurious secondary-author-provided provenience that is at odds with primary provenience.
 
In the US, the CONFU and CONTU hearings had, as I recall, numerous debates on these issues, alas, to no resolution.
 
regards,
David
________________________________

Received on Thursday, 17 April 2008 03:01:34 UTC