As I was saying, Arkansas was my 'Eureka" moment. We had observed that all the earthquakes in southern Ontario were under the lakes. Thus, all our historical seismicity was wrong. It was simple: No water, no earthquakes.
Fracking was going ballistic, and there was something like 2 million gallons of water per hole. All of this had to go some place, and that was deep underground. Arkansas had the advantage of being close to New Madrid, so it was partially covered by a intense seismic monitor array.
I noticed right away that they were getting earthquakes, and they were forming a pattern. The first earthquakes were lining up on a NW line, and then came a very long NE line. This was the exact same pattern as New Madrid! It was also the proper pattern for rock failure along the megathrusts, which have a very strong NE direction in horizontal stress. Yahoo!
There were some injection earthquakes in Ohio, but the lack of seismic monitoring meant no clear patterns.
There had been the M5.6 Oklahoma earthquake, but I was not following the pattern at that time. In retrospect, the reprocessed locations showed a very strong NW line.
Thus, the rock was acting as though it were in a giant rock testing machine. Using the testing machine analogy, all California earthquakes were acting as if there were a very weak, lubricated cut in the rock sample, and the machine turned on. As the rock was compressed, the fault slid.
On the other hand, the earthquakes in ENA were completely the opposite; they acted as though the rock sample was compressed and you were weakening the rock with acid. There was a major difference in the way the surrounding rock acted, and those California guys didn't have a clue.