kyleandkelly.com

Or so it would seem, according to my friend and colleague, Christian Bell. He provides the fitting burial with “An obituary for the semantic web”:http://six27.com/weblogs/christian/christian.2006-08-23.writeback.

First, a quick aside: I would consider the “Semantic Web”:http://en.wikipedia.org/wiki/Semantic_web proper to be the realm of RDF documents and generally far more advanced than what we mere mortals can aspire to understand. Christian and I are more concerned with the practice of using the HTML element that is most semantically appropriate for the content it contains, henceforth known as the “semantic web”.

While I have a great deal of esteem for Christian and his technical perspectives, this issue is likely to be a bone of contention between us. The bulk of his argument rests on the assertions that 1) HTML elements aren’t used in consistent enough manner to be semantically useful and 2) browsers don’t really care what sort of element you use because they don’t really _do_ anything with the elements anyhow (especially in the CSS world where you can make any element look like anything else).

Here is where I take issue: though these statements may be generally true, it doesn’t follow that they will always be generally true. Furthermore, generally true means there are exceptions, and the exceptions in this case are growing.

To begin with: one of the reasons browsers haven’t done anything special with semantically coded web pages is because we haven’t been in doing the HTML/CSS “separate presentation from semantics” thing long enough. Before browsers can start doing interesting things with semantically encoded content, the content has to be there. In order for the content to be there, you have to be freed from coding for presentation, which requires good support for CSS. We needed something like Firefox 1.x just to get started with the creation of the semantic web. Firefox 2.0 is just around the corner, and we’re beginning to see portends of what could be done with a fledling web of well-encoded content. More on that in a bit, because I am guessing the phrase “semantic web” has been used enough to have Christian cringing.

A quick detour to address Christian’s argument on the difficulties of “objective” and “inherent” word meanings: agreed and no argument from me. On the other hand, do we really need perfect, or even good, agreement on word meanings? “Good enough” will do; just enough people using the element in approxiamately the same way to be generally useful.

But useful how?

We’re beginning to see the first hints of how the semantic web might enhance the browsing experience:

* Tooltips on the <acronym>-encoded acronyms scattered throughout this post
* Detection of a related RSS feed via the <link> tag (the little orange icon that gets displayed)
* “Link Widgets”:https://addons.mozilla.org/firefox/2933/, a Firefox extension for enhanced navigation via the <link> tags
* “GreaseRoute”:http://code.highearthorbit.com/greaseroute/index.php a Firefox extension that recognizes the “adr”:http://microformats.org/wiki/adr and “geo”:http://microformats.org/wiki/geo microformats to add mapping shortcuts to any address embedded in a web page

I have to admit that microformats (especially adr) bother me to a certain extent and may actually lend weight to Christian’s point. The typical example for adr is shown as:

<div class="adr">
 <div class="street-address">665 3rd St.</div>
 <div class="extended-address">Suite 207</div>
 <span class="locality">San Francisco</span>,
 <span class="region">CA</span>
 <span class="postal-code">94107</span>
 <div class="country-name">U.S.A.</div>
</div>

As you can see above, the semantic meaning is entirely in the value assigned to the class attribute; the elements themselves are pretty meaningless. In all honesty, I am not sure why, given that the “Tantek Celik”:http://tantek.com/ was the author of the draft. Last I knew, he was a still a big “proponent of semantic XHTML”:http://tantek.com/presentations/2005/03/elementsofxhtml/.

In the end, it is entirely possible the element will become semantically meaningless, as Christian forsees. It is also entirely possible the Detroit Tigers will win the World Series. In both cases, there are many good arguments why they will not come to pass. As web sites continue to embrace CSS for presentation, certain HTML elements will become common enough in their usage patterns to be useful as semantic markers.

Microformats have the potential to further enhance the trend; even if they neuter element names of semantic value, that value has just been transferred to the element attributes. As microformats continue to creep into the mainstream, we will see more support in browsers for doing interesting things with the data. Given the renewed browser wars, they will be looking for anything that might hint at a competitive advantage (”Look! Our browser can recognize addresses and plunk them in your address book!”).

Final words: we’ve been “down this road before”:http://www.thenoodleincident.com/tutorials/box_lesson/why.html and “debated this topic”:http://www.7nights.com/asterisk/archive/2003/07/standards-semantic-markup-distributed-authorship-and-knowledge-management, though perhaps in different terms. In the end, I am most interested in future-proofing my content against uses that I can’t even comprehend. Semantically encoding my web pages in combination with emerging technologies like microformats gives me the best shot at actually doing that.

written by Kyle