The entire article is fabulous, but I love this paragraph in particular:

Another common gripe is that advertisers can’t target segments of users on Twitter in quite the same way as on Facebook, which typically knows a member’s age, work and education history, and hometown. People don’t volunteer the same data to Twitter, and its targeting tools are not as refined.

Costolo says he’s not concerned because there’s so much promise in targeting users based on an “interest graph -- the set of characteristics that can be gleaned from users’ followers and what they click on. He gets animated talking about a treasure trove of data that’s both revealing and noninvasive.

“These accounts I follow paint a very compelling picture of the kind of person I am, even if it doesn’t paint a picture of exactly which uniquely identifiable individual in the world I am,” he says. “I think that allows us to deliver powerful value to advertisers, and powerful value to those who want to speak freely.”

"Revealing and noninvasive" are key.

Sarah Perez write an analysis of this article on TechCrunch, How Twitter is Pairing Its Interest Graph with Ads:

[Costolo] also admitted that Twitter’s targeting tools aren’t as refined as Facebook’s. He’s not concerned, however, because he thinks he’s built the next big thing in terms of ad targeting. The interest graph. Over time, the algorithms behind Twitter’s interest graph could grow more refined, as it learns what we click, what we re-tweet, and maybe even things like the sentiment behind our words. Someone may tweet about Obama, but that doesn’t mean they support them. And since one of Twitter’s latest additions is a tool that lets politicians talk to Twitter’s users through Promoted Searches (search results that stay stuck to the top of the page), it would be incredibly helpful if, say, a candidate could go after supporters, non-supporters and on-the-fence sitters in different ways.

For now, though, the system is much more simple. Search for a candidate’s name, see his Promoted Search result. There’s a lot of potential here to make that system more refined, more exact, and generally more intelligent. If anything ever has a chance at competing with Facebook, it’s not going to be yet another social network mining the same demographic data in the same ways, it’s going to be a network that figures out how to mine new data in new ways – data that defines who people are by their activity, interactions and behavior. It’s going to be the network that figures out the interest graph. And Twitter is off to a good start.

Twitter is off to a great start.

For a more balanced view of Twitter as of March 2012, see Gawker's follow-up to this story.