Daypop “word bursts” is an interesting idea. The implementation seems to be rather naive at the moment, though. It would be great to be able to quantify and track various viral phrases. For example, I’ve noticed a resurgence of the adjective “super” around Microsoft; particularly “super” with at least one other adjective (“super slick”, “super cool”, etc.) Another example: after the 2000 presidential election fiasco, the incidence of the phrase “rule of the law” made a stunning statistical jump (and is still way more common than before, as far as I can tell). And has anyone else noticed how widespread the phrase “make no mistake” became after 9/11/2001?
These are all the sort of things that a reasonably smart computer could clue us in to. The computer could recommend that you use a phrase that is at the cusp of becoming popular, or warn you when you use a phrase that is trending hyperbolically to becoming cliche.
This is no different from concerned parents who analyze the past 100 years of census data to select a child’s name that will be most auspicious when the child reaches maturity and to avoid names that are trending toward obsolescence. Most parents do that, right?
Whoa!Did I just seesome guy explaining to Greta van Susteren that “the CIA has done Bayesian analysis to predict with 85% certainty that Saddam will launch a pre-emptive terrorist strike on the U.S.”?I wonder what they used as inputs for their analysis, considering that nobody even knows what Bin Laden or his crew have been doingfor the past year and the weapons inspectors haven’t been able to find any of the weapons that Saddam is hiding. Apparently Bayesian techniques are so powerful that actual information is unnecessary.I wish we had known about this magic technique before! Hurry up!Let’s run a Bayesian analysis to tell us where Bin Laden is!