I just checked the website referred to in this question before jotting my answer. It is a nice attempt but success count will be speculative only and not scientific. The logic is simple to understand :
1. To predict any earthquake, we must get its precursors (micro-events) rightly
2. Even with precursors, we must get the spatio-temporal pattern rightly.
3. Even with spatio-temporal pattern, we must get the processes involved rightly
Since none of them are possible today, to talk of prediction is non-scientific. To illustrate with example, if we understand the potential zones of earthquakes, we will set up sensors all around to pick up precursory micro-tremors from that area. If we analyse the pattern of these micro-tremors and compare with the trend of past earthquakes (if available), they may match but still there will be no earthquake forthcoming because the earthquake sequence need not follow the same causative mechanism (process) each time. Hence, we are never successful.
Secondly the biggest handicap in the whole episode is picking up the stress build-up whose release results in manifestation of earthquake. There is no mechanism today to pick up the stress build-up across deep-seated fault planes. Even if we try one with deep-drilling and placing sensors, we cannot say with surety as to which section of the fault plane it is loading and what will be the level to which the fault plane can bear the stress. Even if we succeed in that, there is no guarantee that it will be released in that manner, it might trigger a new rupture and release stresses in the form of earthquake. With so much of uncertainty at the theoretical formulation stage itself, experimenting with earthquake prediction is a futile exercise. We have a long way to go in making a vast database of all types of permutation and combinations for earthquake processes and their precursory behaviour.
Had it been so easy, we would have succeeded in predicting man-made earthquakes – typically classified as mining-induced-seismicity and reservoir-induced-seismicity where we have much better control and a small area to monitor. But there also we are grappling with similar uncertainty except a few cases (<10%) where a set pattern is followed in repeated seismicity.
To conclude, I with agree with fellow writer Klein that “anyone who comes up with an accurate way to predict earthquakes should be awarded with a Nobel Prize”.