They're not actually predicting the earthquake - just detecting it as soon as it happens, and transmitting a signal out faster than the earthquake waves travel. So to beat that requires actually predicting when it will happen beforehand, which nobody has made any real headway on in a long time.
It's not a prediction if the earthquake has already begun. The linked article explains that they were able to give ten seconds' warning only because a sensor near the epicenter was able to send an alert at the speed of light, thus beating the seismic waves, which travel at a much slower speed.
About prediction research, so far there are too few ways to gather relevant data for processes deep within the earth, and too little theoretical knowledge. This should be an attractive field for someone with enough physical and scientific training and a willingness to do difficult research.