Hi Ian,
Thank you for the insightful comments and information about parsers.
On Mar 12, 2007, at 16:21 , Ian Hickson wrote:
Why do you think search engine behaviour is more important than
browser
engine behaviour? For what it's worth, search engine engineers I have
spoken to have told me that what browsers do is far more important
than
what a particular version of a search engine does in terms of what the
specification should say, because their results are better when their
algorithms match the browsers' behaviours.
My opinion, which may be wrong, is that the current balance of power
is not the only factor in importance for the design/study of markup
languages.
* browsers mostly determine (largely) whether and how the documents
are presented to the user. Most of it is actually a question of CSS
support, not relevant to this discussion. The rest is a matter of
parsing model and element/attribute support, where the browsers do
have an enormous influence, but even that may shift as more people in
the world move to lightweight browsers on mobile devices than desktop
browsers. If lightweigh browsers with less tolerance of tag soup
carry more weight, the state of the art will be whatever parsing
model is standardized, less so what browser foo or bar does on the
desktop. Ditto for apps/widgets in non-browser environments. There's
more to browsing than the desktop computer.
* Authoring tools and CMSs determine (largely) how documents are
structured, and what features go in documents. If a feature of a
markup language gets no adoption from them, then regardless of
browser support, it will remain in the confidential little world of
web geeks (no disrespect meant, I'm putting myself in this category)
who edit their pages by hand.
* Search engines and their indexing mechanisms determine (largely)
how documents get found. I've seen estimates that content-rich sites
get half of their traffic through search engines. As you aptly point
out, search engines are mostly mimicking browsers' behavior in
parsing HTML documents, but that's not all.
An example: 10 years ago any serious Web page had to have meta
description and keywords information, because that was the key to
being listed in search engines. When search engines started ignoring
those because of spam, usage fell. If today a feature of HTML, or
RDFa, or microformats, caught the fancy of the major search engines
and gave their user a serious visiting boost, the adoption rate would
soar, regardless of browsers support.
* Servers, proxies, cache have their say too, though probably not
much when it comes to markup languages.
All considered, of course I understand your point that desktop
browsers *today* have a considerable influence in defining the state
of the art of the web. But any standardization work, or study of the
web, made under the assumption that other classes of product only
have a minor importance because for the most part they follow this
current balance of power and mimick the desktop browsers, is IMHO
missing a good chunk of the "big picture".
I hope this helps clarify why I was wondering if Henri had considered
other classes of products than browser desktops in his study.
Regards,
--
olivier
