Are earthquakes predictable?

The answer to the above question depends on the definition of earthquake prediction. We discuss several definitions and possible classifications of earthquake prediction methods. We also consider various measures of prediction efficiency, review several recent examples of earthquake prediction, and describe the methods that can be used to verify prediction schemes. We conclude that an empirical search for earthquake precursors that forecast the size of an impending earthquake has been fruitless. Despite considerable effort in several countries, no statistically rigorous validation of proposed precursory phenomena is available: therefore, reported cases of precursors can be explained by random noise or by chance coincidence. We present evidence that earthquakes are non-linear, chaotic, scale-invariant phenomena. The most probable consequence of earthquake self-similarity is a lack of earthquake predictability as popularly defined, that is a forecast of a specific individual earthquake. Many small earthquakes occur throughout any seismic zone, demonstrating that the critical conditions for earthquake nucleation are satisfied almost everywhere. Apparently, any small shock can grow into a large event. Thus, it is likely that an earthquake has no preparatory stage. This sceptical view of current earthquake prediction efforts should not be interpreted as a statement that any further attempts to mitigate the destructive effects of earthquakes are futile. The seismic-moment conservation principle, when combined with geodetic deformation data, offers a new way to evaluate the seismic hazard, not only for tectonic plate boundaries, but also for areas of low seismicity, that is the interiors of continents. Earthquake clustering with a power-law temporal decay (Omori's law) can be used to estimate the rate of future earthquake ocurrence. Realtime seismology can facilitate relief efforts after large earthquakes and eventually provide an immediate warning of severe shaking a few seconds or tens of seconds before the shaking starts.