Re: Empty vs no alt attribute (was Re: Baby Steps or Backwards Steps?)

Right. By convention or standard, purely decorative artwork should have 
alt="". Ideally decorative artwork really should be added via CSS. In 
particular CSS enables the use of a sprite map for performance 
optimization and these decorative bits will go away in high contrast 
mode. This is not the case with regular images being used for 
decoration. In some respects I consider images with alt="" a vestigial 
pattern left over from before the semantic web.

CB

Richard Conyard wrote:
> Chris,
>    What about the instance of images in place purely for decoration?  
> If an image holds no value in to the document in a non-visual context 
> then surely supplying an alternative is degrading the non-visual 
> experience.  In regards to your suggestion of calling out "Author did 
> not provide alternative text", this would also degrade user experience 
> would it not?  I can imagine going through a page and having that read 
> out 20 times to be far more annoying than having no alternative read.
>  
> Richard.
>
> ------------------------------------------------------------------------
> *From:* public-html-request@w3.org [mailto:public-html-request@w3.org] 
> *On Behalf Of *Chris Blouch
> *Sent:* 21 August 2007 15:49
> *To:* joshue.oconnor@cfit.ie
> *Cc:* Sander Tekelenburg; HTMLWG; wai-xtech@w3.org; Don Evans
> *Subject:* Re: Empty vs no alt attribute (was Re: Baby Steps or 
> Backwards Steps?)
>
> Two aspects that seem to be overlooked in the discussion.
>
> 1. While there is an expectation that alt should be used for 
> alternative equivalent text descriptions, in practice many browsers 
> also use it for their popup "tooltips". This had lead some site 
> authors to assume the purpose was more supplemental rather than 
> alternative so they have stored additional meta data rather than real 
> descriptions. Hence we see sites putting up "200x100 GIF Image July 
> 12, 2007" as their alt text instead of a real content description. 
> While this is incorrect use of the attribute, it is hardly a 
> surprising assumption based on browser behavior.
>
> 2. The problem with allowing images without alt attributes in some 
> cases is that this essentially writes a loop hole into the standard. 
> This then puts the burden on the standard to sufficiently constrict 
> this loop hole such that it can't be abused. Hardening any loophole in 
> a set of rules is difficult at best. We need look no further than laws 
> of governing bodies for examples. As for the presence of alt tags 
> being provisionally required, it would seem to be simpler to require 
> them in all cases and then move debate to defining the presence and 
> quality of the descriptions themselves, whether good, lousy or absent. 
> At least this makes for any easy minimum test, a test which many tools 
> already implement. Loopholes for poor or missing alt text would be a 
> more fertile ground for work in the standard, but still a quagmire.
>
> I suggest that missing alt text should be called out in a manner that 
> points out the author's mistake (either directly or via their tools). 
> Something like "Author did not provide alternative text" or whatever 
> wordsmithing is desired. This provides a means for discoverability by 
> others to feed back to the source that something is wrong. If Grandma 
> sees no tool tip or one that says just "undefined" she will ignore it. 
> If she sees a message pointing fingers at my failure to do something 
> right I might actually get an email from her. In other words, while 
> the tooltip and non-descriptive errors for alt text may not directly 
> enable users who need alternative content, it can enable other users 
> to seek out resolution to a problem that was previously invisible to many.
>
> CB
>
> joshue.oconnor@cfit.ie wrote:
>> Sander Tekelenburg wrote:
>>   
>>> I cannot imagine how alt="unknown" could be useful to anyone. 
>>>     
>>
>> I am also at a loss to understand how this is useful. Is it expected
>> that screen readers would read output "unknown" when they came across
>> and image that used the attribute in this way?
>>
>> Josh
>>
>>
>>