Carl Bialik (Wall Street Journal) writes in Timing Twitter about the dangers of reading too much into tweet statistics and then says:
She [Twitter spokeswoman Elaine Filadelfo] noted that the company is being conservative in its counting, and that the true counts likely are higher than the ones reported by Twitter. For instance, the company didn’t include “Ryan” in its search tersm for the Republican convention, to avoid picking up tweets about, say, Ryan Gosling rather than those about Republican vice-presidential candidate Paul Ryan. And it has no way to catch tweets such as “beautiful dress” that are referring to presenters’ outfits during the Emmy Awards telecast. “You follow me during the Emmys, and you know I’m talking about the Emmys,” Filadelfo said of the hypothetical “beautiful dress” tweet. But Twitter doesn’t know that and doesn’t count that tweet.
Twitter may not “know” about the Emmys (they need to get out more) but certainly followers on Twitter did.
Followers probably bright enough to know which presenter was being identified in the tweet.
Imagine a crowd sourced twitter application where you follow particular people and add semantics to their tweets.
Might not return big bucks for the people adding semantics but if they were donating their time to an organization or group, could reach commercial mass.
We can keep waiting for computers to become dumb, at least, or we can pitch in to cover the semantic gap.
What do you think?