EQF and all. Long term forecasting has been done all along (xx% chance of quake exceeding 10% g in 30 years, or xx% chance that a particular fault segment will break in 30 years). But, even that effort has had problems because many fault do not show a time predictable behavior (for example, the southern San Andreas fault may be running 100 years late, the San Andreas at Wrightwood shows clusters of ruptures, etc). Government funded short term prediction in the USA probably was pretty limited for a couple of decades until a few years ago, because the whole area of science had gotten a bad name. But, there seems to be an effort the last 5 years ago to see if short term prediction is possible. For example, the Southern California Earthquake Center has an effort for earthquake predictablility. But, prediction efforts that they evaluate must be physics based and are subject to rigourous testing.

I think Roger and others here would agree that no one on this page has shown statistical skill at short term prediction.
Chris