The article neither says, nor offers any reason to believe, that Twitter can predict any of those things.
(Predicting earthquakes is a very difficult problem and any system that could do it well enough would be tremendously valuable. What Twitter can do is to detect earthquakes after they happen, by watching for people tweeting about them. That's got some value too, but it's not remotely at the same level of impressiveness and importance that an earthquake predictor would have.)
what would happen is such a system was sabotaged by someone who decides to raise false alarms? doesn't it then mean that the best this can do is spread the word of something already occurred, like gjm11 said already?
am not a seismology expert, but maybe attaching enabling seismic sensors that actually detect S and P waves or other stuff, and then automatically tweet about a possible earthquake.
This i think would be a more trustworthy prediction, and would even have more value.
(Predicting earthquakes is a very difficult problem and any system that could do it well enough would be tremendously valuable. What Twitter can do is to detect earthquakes after they happen, by watching for people tweeting about them. That's got some value too, but it's not remotely at the same level of impressiveness and importance that an earthquake predictor would have.)