The paper and the recent posts about how Twitter data can be used to predict the first-weekend box office revenues of movies are quite interesting.
Basically, provided sufficient data about a topic and clever ways to process the data, the analysis of Twitter data could potentially indicate if people are going to buy a product, watch a show, vote for someone, go somewhere or think in a given way, etc.

Now I’m wondering if the simple corollary to this is not that Twitter is making the future. After all, Twitter can also be seen as a massive network of influences, and its real-time property makes it much more dynamic than blogs, Facebook, newspapers, or TV. Twitter sphere’s of influence goes also beyond Twitter.com as Twitter feeds are embedded in many other websites and applications.

Compared to other types of data, twitter’s data is easily accessible, is better structured for text analysis (short, you don’t need to parse and find the interesting piece of text in a page long text), as well as for other data or time-based analysis (all tweets are real time, all stamped with the same time/clock system which is a big deal, they cannot be edited, they can all be tracked to their original authors and those who retweeted), and finally there is a lot of it.

Wouldn’t there be a causality link here too? Twitter chatter would not just reflect what people are thinking or doing, it also influences them, like any other viral media channel.
Instead of just investing into methods to know what the chatter says about a product or someone, companies and people are going to invest to increase the amount of chatter related to that product or someone. Like they did with forums, and blogs.

Twitter is ultimately an excellent platform to experiment with prediction and real-time making of the future. Build a system that adds to the Twitter chatter in the direction you want (to promote something for instance) and watch real-time if it is going to influence and affect the outcome.