Biz Stone got this directionally correct, it is just not in a format that is usable in real time, yet…
If I were the lead product manager at Twitter, I’d be setting up two primary screens:
1) Keywords across the entire network overall which toggle switches for the frequency compared to normal (it would be cool if this was somehow “normalized” so that less frequent words with huge percentage increases actually got noticed – so that everything wasn’t Angelina Jolie, Barack Obama, Brad Pitt, Michael Arrington, Jason Calacanis, Fred Wilson, etc).
2) Graphs and/or alert spikes of user defined keywords – ie ones that are important to oneself personally or to one’s business or clients. I would dare to say this might actually be a business model that could lead to meaningful monetization – I think alot of web services haven’t thought this through nearly enough. Organizing real-time data for useful decision making as a business model worked out OK for Michael Bloomberg if I recall correctly. Some might say Google Trends does this already from a search perspective, but it doesn’t break down the word clusters to core words with “sidekicks” and is not the leading indicator that Twitter is by an uncertain but definite time margin.
Before I’d get to this though, Twitter would have to become stable and Twitter would have to fix the AIM problem I posted about on June 1 and have not gotten a response to yet. Good luck on getting those items in order first, then feel free to give me a shout Biz.