Flutterby™! : Smart Tags 2001-06-08 16:22:11+00

Smart Tags

2001-06-08 16:22:11+00 by Dan Lyke 2 comments

There's a big outcry over Microsoft's "Smart Tags" features in the XP browser (aside: has anyone figured out what XP means besides "extra pricey"?), Dave Winer collected people's fears yesterday. I don't know for certain, but from the descriptions I've seen there's a solution: Richly link the hell out of your pages. Larry's convinced me that I need to go CSS, and that'd let me have different classes of links, which means that I could link nearly every phrase in a document (with some tool rewriting). That's what the promise of hypertext means to me, if Microsoft pushes us to making our content richer with better generation tools that doesn't sound all bad to me.

Just like NT meant "New Technology", whoops, meant nothing at all, XP means "eXPerience", oops, nothing at all. It was introduced at the Experience Music Project mostly-Jimi-Hendrix rock museum.

Dan, doesn't the F'by perl do automatic linking to certain things? Manila does the same thing, you can define objects that when "named and quoted" generate links to themselves. If, and this is a big if, Smart Tags were something that were able to be tightly controlled by the website owner, it would be a cool way to generate a personal lexicon. If you're always mentioning Mouthorgan or Burning Man, say, it would always be optionally linked to this definition/new browser window dialog.

I'm more than a little worried, though, that instead it's going to be Third Voice as a monetizing feature, sort of the theoretical inversion of framing somebody else's site. Only this time the links to the advertiser would be right inside your site.

Flutterby's glossary capability has actually been degrading over the years, I used to have a really extensive one that grabbed pretty much every bit of linked text as it was added, but that still had to be underscore-quoted to be recognized, and when I went to database-backed, because pages were getting regenerated later I had to drop much of it.

My vision for the next revision is that I'll have some rudimentary pattern matching to look for parts of speech and capitalization patterns to check some grander databases. There's no reason I shouldn't attempt to match every song title against FreeDB, every movie and person name against IMDB, every book and potential author against the Bowker databases as published by the big online bookstores (BookSense/Amazon/Barnes & Noble/Borders), and for that matter I should take everything that doesn't fit those, append the usual domain names, and try linking it that way.

The issue, besides the time to implement this, is that this kind-of requires putting workflow into Flutterby's CMS so that there's a conscious editing step in adding messages. I don't think my users would stand for it yet.

But content management has slipped way

down on my list recently, so that'll probably wait for a bit.