The utility of twitter as a profit engine has been a growing concern for Evan Williams. Other social services are quickly adapting to the popularity of microblogging within their own social media structures (Facebook, FriendFeed). While I have posted on the potential of twitter before, the real value is in the datamining application of microblogging.
The Twitter Gold Mine & Beating Google to the Semantic Web by Nick Bilton got me thinking about targeted advertising. What I commented on (and Nick didn't touch on), was that the profit engine in microblogging isn't necessarily the service. Sure you want your microblogging service to have a sizable community. But what you really want to provide is the best set of tools for user intent comprehension. It's likely that the most powerful fully semantic data miner tools will be our first look at Artificial Intelligence.
Imagine a virtual personal assistant that tracked all your activities (microblog entries in this case). It can create pertinent products or activities by analyzing a combination of historic data from your past and collective data from others who share your current status. The "others" that comprise your data group could be people you follow, you are friends with, or are in the same geographical vicinity. For example, you're visiting Columbus Ohio to meet some customers. While chatting after the meeting your virtual assistant conveniently texts your portable media device with the name and address of the most locally popular hibachi grill restaurant (because you love Hibachi while traveling). The algorithms will have to discern meaning from the microblog entries, hence the focus on semantic knowledge and A.I. They can then be processed by any number of estimation algorithms (clustering, matched filtering, other correlative techniques).
The software developers that can best deliver this function will earn titanic profits (from users and advertisers).