- From: Elliotte Rusty Harold <elharo@metalab.unc.edu>
- Date: Sat, 17 Aug 2002 09:29:36 -0400
- To: <www-tag@w3.org>
- Cc: <www-style@w3.org>, <w3c-css-wg@w3.org>
At 9:02 AM -0400 8/17/02, Didier PH Martin wrote: >>From the social and anthropological point of views you are totally >right, especially if you are referring to the main stream web content >designers. For them an HTML document is simply a document's visual (and >sometimes aural) layout. This also gives us a good clue of the reasons >why we do not see yet a semantic web ;-) > You know, on further thought I don't really agree with this. An HTML document like the Mercury News homepage is not just a visual layout. Nor is the print edition just a visual layout. Looking at the home page, it is really obvious what the headlines are, just as it's really obvious looking at the printed front page at the news stand. No person is confused by this. The visual layout includes semantics. For example, when a paper prints a headline in type that's three times larger than usual, we all know that this means something extraordinary has happened. We don't need a footnote saying this is a "really important headline" instead of just an ordinary, average, headline. Unfortunately, computers are much stupider than humans. Computers have a great deal of trouble associating visual cues with semantics. Computers get even more confused when they're told that sometimes blue means a headline and sometimes it means a link and sometimes it means the designer thought blue text looked cool. Humans, by contrast, have no trouble disambiguating these different uses of the same color. The whole goal of semantic markup -- whether SGML, XML, or something else -- is to get humans to dumb down their writing and design to the point where computers can understand it. Maybe this is necessary. Maybe it isn't (or maybe it won't be one day). Either way, let's not kid ourselves that this is a good thing. It is at best a necessary evil. In the ideal world, computers should adapt to humans, not the the other way around. Computers should learn to speak our language. We should not have to learn to speak theirs. A small percentage of the population on one end of the bell curve can adapt their thought processes to the needs of the machine, but most people can't. Most people are far more comfortable speaking a natural language, including a natural visual language in which presentation carries semantics. Do not be surprised that these people latch onto whatever visual model you give them, and ignore all the nice semantics you've provided. To them, the tagged, explicit semantics range from painful to meaningless. Presentation, by contrast, is easy, obvious, and natural. In hindsight, it's obvious that SGML on the Web was going to fail. It was too far removed from the way most people think. It's also no surprise that it achieved reasonable success as a medium for computer to computer communication in technologies like XML-RPC and SOAP. XML is actually quite well designed for the way computers think. And it should be no surprise that XML worked well for programmers, because they're the people who are most comfortable talking in the computers' language. But I remain extremely skeptical of any claim that says we're going to reach the semantic web by teaching all the Web developers and authors to think like machines. -- +-----------------------+------------------------+-------------------+ | Elliotte Rusty Harold | elharo@metalab.unc.edu | Writer/Programmer | +-----------------------+------------------------+-------------------+ | XML in a Nutshell, 2nd Edition (O'Reilly, 2002) | | http://www.cafeconleche.org/books/xian2/ | | http://www.amazon.com/exec/obidos/ISBN%3D0596002920/cafeaulaitA/ | +----------------------------------+---------------------------------+ | Read Cafe au Lait for Java News: http://www.cafeaulait.org/ | | Read Cafe con Leche for XML News: http://www.cafeconleche.org/ | +----------------------------------+---------------------------------+
Received on Saturday, 17 August 2002 09:36:23 UTC