- From: Karl Dubost <karl@w3.org>
- Date: Thu, 20 Apr 2006 11:06:58 +0900
- To: Ian Hickson <ian@hixie.ch>
- Cc: XHTML-Liste <www-html@w3.org>
:)
Le 06-04-17 à 07:01, Ian Hickson a écrit :
> On Sun, 16 Apr 2006, Laurens Holst wrote:
>>
>> Seems that nowadays some think heuristics are the answer to all
>> problems. Heuristically determine language. Heuristically determine
>> content type. I’m afraid that in practice it will turn out that that
>> doesn’t work either, and certainly isn’t interoperable.
>
> It can be interoperable if the heuristics are well defined.
How to slide semantically to another level or just put the problem a
little bit further. That's very interesting how people
- acknowledge a problem
- create something to deal partially with it
- and then create the paradigm of the new problem.
=> well defined heuristics which is how to create a (hidden) standard
for heuristics.
Anyway :))) The point is
* markup is not a lost cause
* heuristics is useful
* things created by humans are prone to errors.
I know a lot of lost cause if we keep this kind of reasoning that we
hear these days. A few example?
* English expression is a lost cause (look at the number of English
messages written with bad grammar and spelling. We should stop
teaching English and only apply well defined heuristics.)
* Road Code is a lost cause (look at the the number of bad drivers
out there on the roads. We should stop teaching Road Code and only
apply well defined heuristics.)
The fact there are flaws in languages, that humans don't know how to
use them, that people abuse them, doesn't mean that they have to be
removed. Heuristics is necessary it's not a replacement and/or the
solution of everything, that would be as extremist as saying we only
need standards and no heuristics.
> And it might not work perfectly, but it works better than explicit
> labelling. This isn't a hypothetical statement -- for Content-Type
> browsers have unanimously found heuristics (even without a spec!)
> to be
> more reliable than content labelling. An earlier post on this thread
> pointed out the same thing is true for language tagging.
No. :)
Repeat after me. Context.
It's all about context.
--
Karl Dubost - http://www.w3.org/People/karl/
W3C Conformance Manager, QA Activity Lead
QA Weblog - http://www.w3.org/QA/
*** Be Strict To Be Cool ***
Received on Thursday, 20 April 2006 02:07:12 UTC