alt text on Validator.nu image report (was: Re: alt crazyness)

Robert J Burns wrote:
> On May 3, 2008, at 1:56 PM, Smylers wrote:
> 
>>  1 This webpage conforms to the HTML 5 standard except that it includes
>>    unknown images from external sources for which we are unable to
>>    provide alterternative text.
> 
> This conveys another myth we need to dispel immediately. Few should ever 
> be authoring a page where they do not know why they included an image on 
> the page (perhaps this is part of the same misconception I'm already 
> trying to dispel). All anyone has to do is remove the image from the 
> page, decide whether the page is missing something important in not 
> having that image there. If it is, put the image back and briefly 
> describe what was missing without the image. Perhaps you or someone 
> could point us to a real world example of a page where you think the 
> purpose of the image on the page is inexplicable. That way we could 
> continue the dialog and demonstrate that it's not the case.

http://validator.nu/?doc=http%3A%2F%2Fnews.bbc.co.uk%2F&showimagereport=yes&showsource=yes#imagereport

Those unknown images from external sources on Validator.nu are a 
critical part of the content, and there is usually no available 
equivalent textual alternative for them. (The purpose of these images 
isn't "inexplicable", but they seem to match what Smylers was 
describing, and Henri appears to think alt should be omitted for them.)


My attempt at understanding this specific case:

The site knows:
  * The existence of the image on the page.
  * The image's filename.
  * That the image is critical to the functionality of the page.

The UA knows:
  * The existence of the image.
  * The image's filename.
  * The image's bitmap data.

The user wants to know:
  (a) That the image is critical to the functionality of the page.
  (b) Details about the structure and content of the image.

((a) is important for the user because it means they should spend some 
effort to examine the image in detail. They wouldn't want to spend the 
same amount of effort on most other images on the web, because most 
aren't that critical and it would be a waste of time.)


The UA can decide very little about (a) from the alt attribute being 
absent, present, or set to "", since all those possibilities are abused 
by existing content. So the site cannot unambiguously convey the 
importance to the UA, because the UA cannot trust the site to be correct.

The UA can discover some of (b) (details about the image's content) from 
having downloaded the image data, e.g. reporting its filename and 
dimensions, deciding whether it's iconic (small, few colours, high 
contrast, sharp edges) or photographic, looking at the metadata inside 
JPEGs/PNGs, etc. The site knows nothing about (b), since it doesn't have 
the image data (and it would be too slow and expensive and insecure for 
it to download all the images itself), so it can't help with that at all.


The user can be quite certain about (a) because of context: they know 
this is in Validator.nu's image report output, so they require only that 
the UA indicates the existence of the image, and then they will know it 
is an important image and worth examining in more detail.


So, the goal of the site's markup is to make sure the UA tells the user 
clearly and concisely that there is an image here. It is up to the UA to 
determine as much detail as possible about the image, and it is up to 
the user to decide (based on context) that the image is important and 
that it is worth looking at in detail.


In that case, <img src="..." alt="External image"> seems to achieve the 
goal: as long as UAs do not ignore images with alt text, the user will 
be made aware that the image exists, and can ask their UA for further 
details.

It is also slightly more informative to users than <img src="..."> (no 
alt) could be, since it can say that this is an external image (or some 
other terminology indicating that it's from the reported site, rather 
than being a part of Validator.nu itself), which is information that is 
impossible for the UA to deduce.

Also, a UA that performs some image analysis to optimise the rendering 
of typical (non-conforming) web pages should treat <img src="blank.gif"> 
(where blank.gif is a 1x1-pixel transparent image) as equivalent to <img 
src="blank.gif" alt=""> (i.e. ignore the image entirely), because the 
image typically conveys no information. In the Validator.nu case that 
would be harmful, because the user must still be made aware of the 
image's existence, and so Validator.nu should not use markup that will 
cause this type of problem.

So... Am I missing ways that <img src="..."> better (or equally) 
achieves the goal, or is my determination of the goal incorrect, or is 
this sensible?


(My interpretation of the current HTML5 spec is that alt="External 
image" would be non-conforming, because it does not serve as a 
substitute for the image - it serves only as a marker for the image and 
an indicator that the image is 'external'. The spec should not be so 
strict in what the alt text must be, else it force sites like this to 
compromise their user experience in order to achieve conformance.)

-- 
Philip Taylor
pjt47@cam.ac.uk

Received on Saturday, 3 May 2008 19:51:10 UTC