[whatwg] Problems with the definition of <cite>

[James Graham]

Benjamin Hawkes-Lewis wrote:
> I wrote:
> 
>> That's not a fair summary: see the example I gave to Anne van Kesteren
>> of getting back to a Hamlet scene text from <cite>Hamlet, I.ii</cite>
>> with a mere Google query.
> 
> James Graham wrote:
>> Using the <cite> attribute to link to a search page is at best almost-useless 
>> and at worst damaging and confusing. 
> 
> Sorry, I didn't make myself clear. The <cite> in my example does /not/
> link anywhere. My point was that even with the mere knowledge that the
> contents of <cite> are a citation, agents can construct useful web
> queries. Not as good as a <cite> with a URI, but not /useless/.

Oh, you were proposing hidden metadata. Well that's almost worse
because experience suggests hidden metadata in HTML doesn't work. Consider:
* The number of pages that use rel="next"+etc. links, despite some
browser support for these features
* The number of pages that miss out alt-text or have useless alt text
like "Photograph of the author". Indeed the latter seems to be
increasingly common as people get the "you should/must use alt text"
message but don't actually see their page with the alt text in.
* The relative number of pages that use  <abbr> compared to <acronym>

If we accept that we need to mark up citations then I suggest that the
best way of linking to the source of the citation is to provide a link:

"We apply the stochastic model of iron transport developed by <cite><a
href="http://ukads.nottingham.ac.uk/cgi-bin/nph-bib_query?bibcode=2005astro.ph..1141R&db_key=PRE&data_type=HTML&format=&high=454db6506508253">Rebusco 

et al. 2005</a></cite> to the Centaurus cluster"

This way the metadata is visible and useful. Of course you could 
question the added value of the <cite> element given that the real-life 
pages which provide that and similar scientific articles don't seem to 
use <cite>.

> Guaranteed recognition of the <cite> element
> is better than nothing.

Only if the value it adds to the web outweighs the cost of putting it in
the spec. Knowing what the use cases are (real, specific, use cases, not
just hypothetical intelligent agents with undefined function) is
important in deciding how the cost/benefit works out.

> I asked:
>> Deprecating <cite> wouldn't solve any problems, as far as I can see. How
>>  would you connect <q> or <blockquote> to a particular hCite block?
> 
> James Graham replied:
>> Indeed in many cases where citations are important there is no 
>> direct quote to match the citation (almost all scientific papers fit this 
>> model). 
> 
> This might well be an argument for providing ways to explicitly connect
> <cite> with elements other than <q> or <blockquote> (a global attribute
> "citeref" could perhaps do that rather well). Given the existence of the
> humanities (a literary form in which extended series of quotations are
> decorated with footnotes and occasional commentary), I can't see how
> it's in any way an argument against providing ways to explicitly connect
> <cite> with <q> and <blockquote>.

How would that work in the example above and what UA feature could I 
design if I knew which bit of text the citation specifically applied to?

-- 
"The universe doesn't care what you believe. The wonderful thing about
science is that it doesn't ask for your faith, it just asks for your
eyes" --- http://xkcd.com/c154.html