W3C home > Mailing lists > Public > public-html@w3.org > February 2009

Re: summary="" in HTML5 ISSUE-32

From: Philip Taylor <pjt47@cam.ac.uk>
Date: Thu, 26 Feb 2009 13:50:56 +0000
Message-ID: <49A69E40.204@cam.ac.uk>
To: Ian Hickson <ian@hixie.ch>
CC: Steven Faulkner <faulkner.steve@gmail.com>, HTMLWG <public-html@w3.org>, W3C WAI Protocols & Formats <w3c-wai-pf@w3.org>
Ian Hickson wrote:
> On Thu, 26 Feb 2009, Steven Faulkner wrote:
>> Philip's "study" in his own words:
>>
>> "Philip notes that his thing was not attempting to be a particularly 
>> useful or detailed or well-thought-out survey, it was just scraping some 
>> easily-available information"
> 
> Philip may not consider his data "particularly useful", but since his 
> results show similar proportions to other studies, I believe he is 
> unnecessarily self-deprecating.

I never said it wasn't particularly useful, just that it wasn't 
attempting to be particularly useful :-)
In particular, that data came from (if I remember correctly) a small 
list of pages from Yahoo search results (about 1.5 years ago), which 
included a very small number of pages using summary and is not a sample 
from a well-defined population, so I'm not happy with relying on it much.

So I've now created 
http://canvex.lazyilluminati.com/misc/summary-20090226.html based on 
pages from dmoz.org a month ago, which is more recent, and more 
well-defined and reproducible, and a lot more data. Also I limited it to 
counting one page per domain, to prevent sites like Wikipedia 
overwhelming the results. Obviously it's still significantly biased in 
certain directions (e.g. dmoz.org focuses more on front pages than the 
deep content pages that are more likely to contain data tables, and on 
general-interest pages rather than very specialist pages), but at least 
it's better defined bias now.

(It's still unreasonable to conclude from the data that there aren't 
sites which do use summary usefully and widely, because it can't prove 
non-existence. But it can demonstrate that making summary visible to 
graphical UAs on all tables on all pages would be unacceptable, by 
giving a lot of examples where that would result in text like "Layout 
Table: wraps all of the content on the page" becoming visible and very 
few examples where it would be beneficial.)

-- 
Philip Taylor
pjt47@cam.ac.uk
Received on Thursday, 26 February 2009 13:51:40 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:39:01 UTC