In 1906 in California on the seismogenic fault of San Andreas seismic disaster occurred that destroyed San Francisco. When the earthquake occurred significant progress along the fault – up to 6 meters. After that earthquake were to re-triangulation measurements over a wide area, blocking its epicentral area. Analysis of these measurements showed that displacement of geodetic points are not chaotic, and regularity – the maximum displacement appeared geodetic points located near the fault, and the removal from it offset the items quickly and regularly (exponentially) decreases. These are the survey data were the basis for the proposed GF Reid «theory of elastic recoil earthquake» [10]. Reid found that in the preparation of the earthquake deformation direct signs appear on the earth’s surface, and that they can directly monitor the method of repeated geodetic measurements, ie, predict the place of training centers of forthcoming earthquakes. American surveyors have attempted to test these considerations Reid. However, when analyzing the results of test measurements was made a gross error. They estimate the accuracy of the measurement results on the basis of a nondeformable applied to the measurements performed on the deformed base. That is what led them to conclude the error of «theory of elastic recoil». There is reason to believe that this was the beginning of the crisis in the problem of earthquake prediction.