I don't know about you, but one of the biggest banes of a web wonk's life is the meta tag, that strange, metaphysical entity we all seem to rely upon but never quite trust in first identifying and later finding our precious Web content.
An article published this week by Danny Sullivan, the Editor of The Search Engine Report, really got me thinking about the value of meta tags.
In his article, Danny calls for the dissolution of meta tags, mostly because only one search engine (Inktomi) actually uses the incredulous things in building its indices. While I agree with Danny that meta tags really have a limited use when it comes to positioning your content within a winning search engine, I contend that meta tags must continue, not for external but for internal search engines.
We presently use a host of meta tags in our content, each designed to provide a crucial bit of data regarding our content, such as the date published, the technologies discussed, the author's name, and of course, the story's title. We do this not so much for external engines but for our internal Verity engine, which is tuned to identify and then prioritize specific tags.
For example, we use a tag called "article type" which helps Verity build sub-collections of data for quicker searching and listing. Here's an example:
<meta name="article type" content="Feature"/>
We also use a tag called "FirstPage" which tells verity that a given page is the first page of a multi-page story. We do this simply because we want to exclude a series of search results that appear to be duplicates. Clearly, without custom meta tags like these, our internal search engine would operate at serious disadvantage. However, that's not what I worry about with meta tags.
For the past few years, I've grown increasingly ill at ease with these critters, knowing that such tags fail for one simple reason: people.
How can you rely upon the validity and consistency of tags that are so often created by individuals, who, through no fault of their own, cannot count on or be sure of their ability to enter a simple date the same way each time. Have you ever seen the date encoding? Here's an example right from the WWW Consortium's HTML 4.0 specification:
<META name="date" content="1994-11-06T08:49:37+00:00">
Let's see you type that three times fast. :)
What we need, and what we've only recently been able to taste here at Network Computing, is a content delivery system that generates meta tags programmatically, putting an end to erroneous and inconsistent meta tag entries. Our editors simply select the story type, publication date, author, etc. from a UI that can verify and correct their entries. We hope this mechanism will eventually lead to a very efficient historical record of all NWC content, one which you can search with confidence.
As far as external engines like Google.com are concerned, we love them and we say, "let them eat links."
Posted by Brad Shimmin at 11:19:09 PM
|
|
