Topsy launched the other day — it’s a search engine for links in Twitter that ranks link importance by the number of their retweets. An interesting idea, but I have a big gripe about this particular use.
There was a good post recently about the overall value of Twitter as metadata and annotations to the web, but Topsy shows how this can backfire. Take a look at this result & related tweet for example.
An update is a rather predictable beast. It can tell us:
- original/reply/forwarded status (e.g, leading with an RT, or having a reply ID)
- associated users (sender, plus @usernames)
- a location (the URL(s))
- some description/comment (remaining text that is not any of the above)
But Topsy doesn’t do anything special with these beyond expanding a URL and getting its title. It repeats the update without any further parsing, so it doesn’t help us focus past the stuff it has decided to format nicely already. A link update is already noisy, but now it’s even noisier as we have to read things twice; Why do I need to see the short URL when the link is right above it?
In fact, with this screenshot, the body text is identical to the linked story’s headline. So as metadata, on the surface, this update is just plain useless. However, the metadata of the update itself is — the date it was posted, the reply thread, the sender, etc.. This information should accompany the result as well, especially if the text is identical to the page title (which then points out that text should be hidden). Why should your user need to expend the effort to determine what’s different in the update when there isn’t anything different about it?
If we’re going to start using Twitter as metadata, we’ve got to start intelligently parsing it and displaying it for users. Otherwise, it’s just another burden on a user to find meaning and context in what’s been given to them — much like retweets themselves.
BTW: If you haven’t guessed yet, I’m a strong objector to RTs.
