The Future of Tags

Joshua Schachter was spurred to created Del.icio.us (http://del.icio.us) because it was so difficult to share some of the many Web sites he'd bookmarked. So, he set up a free site that lets anyone record sites she'd want to remember. What distinguished Del.icio.us from other online bookmarking sites was a simple technique Schachter added for sorting through the hundreds or even thousands of Web addresses that someone might accumulate: Tagging. It couldn't be simpler: When a user adds a bookmark to her page at Del.icio.us, she's prompted to enter a word or two. Then, the user can sort her bookmarks using those tags. It's a bit like putting the bookmarks in a folder, but because a bookmark can have multiple tags, it's far more flexible.

Private tagging is useful. But at Del.icio.us, tagging is public, and that blew tagging into the center of an economic, computational and semantic storm. Tagging is public at Del.icio.us in that anyone can see not just everyone else's bookmark list, but can see all the pages that any user has tagged with a particular word. For example, you can see every page anyone has tagged with the word "taxonomy" or "Iraq." The result is a stream of pages, put together by the wisdom of the crowd that hangs out at Del.icio.us.  Then secondary effects kick in. If users can see how others have tagged a page, will that affect how they themselves tag it? Will folksonomies -- bottom-up taxonomies -- emerge? As more people started using -- and talking about -- Del.icio.us, questions arose: How to handle the ambiguity of tags? What sorts of intelligent clustering can be done? Can we cluster tags in multiple languages? Will tag spam be a problem? Will social groups coalesce around shared tag sets? Will tagging scale? Will it matter?