Re: Summary of Thursday's IRC conversation about @summary

On Jun 8, 2009, at 15:11, Shelley Powers wrote:

> Henri, your reasoning is a little flawed here.
>
> A statement has been made that the attribute isn't being used, but  
> then you're saying that authors are expending effort on the  
> attribute, which is the same as saying, it is being used. Which is  
> it? Is it being used, or not?

If authors put *something* in the attribute but either AT heuristics  
suppress the attribute or users disregard the attribute, it is being  
used but is not useful.

> And the methodology used to determine that the attribute supposedly  
> isn't being used, is primarily anecdotal, and not particularly  
> comprehensive. Either it's based on Ian's queries in Google's index,  
> which can't be validated because there is no non-Google party access  
> to the raw data used in the queries, or it's based on one movie  
> interviewing one person.

Philip's data seemed to show similar results independently.

> Philip's data has shown that there is a direct (observed, not  
> statistically measured) correlation between incorrect use of the  
> summary attribute, and incorrect use of the table element. Frankly,  
> I would have to assume the same can be said of Ian's own results.

More interesting than that correlation is the apparent absence of  
summaries that look like the kind of summaries put forward as examples  
of what authors should be putting in there as best practice. Even if a  
layout table detection heuristic prunes out some of the bad summaries,  
where are the good summaries?

> Then the argument is given that those who want to keep the summary  
> attribute have to perform research far beyond anything being done  
> with any other aspect of HTML5,

Well, personally, I'd have been convinced by something like asking a  
few screen reader users to find out if they found table summaries to  
be a net benefit. I guess I'll have to trust the 3 opinions Steve  
relayed, although I still don't comprehend where the useful summaries  
are found when the data I can check (Philip's data) shows a mountain  
of bad summaries--not a mixed mountain where the bad could be pruned  
automatically out leaving a pile of good ones. (Unless, of course, a  
summary such as "Calendar" counts as a good summary despite not being  
the kind of summary that a summary is supposed to be per best-practice  
examples.)

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Monday, 8 June 2009 12:50:01 UTC