[uf-new] Use of img in rel-* (with analyzed data)

Benjamin Hawkes-Lewis bhawkeslewis at googlemail.com
Tue Jul 17 01:14:55 PDT 2007

Manu Sporny wrote:

> This analysis required human interaction, thus the sample size is small
> (but still statistically significant). A small GUI displayed an image to
> a person and asked them to select if the image matched the ALT tag. This
> is the first time this data is being presented:

I think your analysis has done a good job of showing that @alt usage is 
better than I at least would have generally assumed, and it's great that 
you actually tested @alt with humans. But I do think there is a 
methodological flaw with how you did this. @alt text does not exist in a 
vacuum, but in the context of a page. @alt does not match image, but the 
use of an image within a given context. For example:

<a href="help"><img src="question-mark.gif" alt="Help">Help</a>

would be better than no @alt, but would still be misguided. In context, 
the correct alternative text would actually be alt="". And such errors 
would matter for microformat parsing, e.g.:

<span class="fn"><img src="benjamin-hawkes-lewis.jpg" alt="Benjamin 
Hawkes-Lewis">Benjamin Hawkes-Lewis</span>

So it would be better to present at least the immediate context of the 
image to human testers, not just the image itself.

(Note I /strongly/ agree that microformat parsers should treat @alt text 
just like other text as per the HTML specification and WCAG; I'm making 
a purely methodological point about your statistical approach here.)

Benjamin Hawkes-Lewis

More information about the microformats-new mailing list