The report at that Web site mentioned the subject of earthquake forecast failures. My own data indicate to me that earthquake forecast accuracy is often governed by the following:

*** You need to work with a reliable precursor or statistics based program. And some precursors and statistics programs are better than others.

*** If the ground is beginning to shake beneath you, that could probably be regarded as a fairly good sign that an earthquake is imminent. However, with most other precursors and statistical programs, earthquake occurrence times are controlled by probabilities. Few if any forecasts have a 100% probably associated with them. So, some will always fail. The expected earthquake might not occur for months or even years.

*** Strain stored in a fault zone which is about to cause an earthquake to occur might be abruptly reduced as the result of events occurring elsewhere such as another earthquake in a parallel fault zone. And so, even though you are working with a reliable precursor which accurately indicated that an earthquake might be about to occur, those external events ultimately determined how accurate your forecast was, not the quality of the precursor itself.