Over the years your research group has been developing a new approach to the earthquake prediction. Do you work primarily with other scientists or with non-scientific organizations as well?

Our group includes mathematicians, physicists, and geoscientists, as well as the end users of predictions – disaster managers from administration and NGOs. This group can hardly be called mine: it has brought together cutting edge experts from about 15 institutions worldwide. That diversity of expertise is not a smorgasbord of a sweeping blind search. Our collaboration is structured according to the complex origin of earthquakes. Their occurrence is governed by a host of intertwined non-linear mechanisms, such as mechanical deformations, transformations of minerals, rocks-fluids interactions, electrochemical impacts, and perturbations of Earth rotations, to name just a few. These mechanisms act in the time scales from geological to fractions of seconds, and in spatial scales from global to crystals of minerals. Even a grain of rock may act simultaneously as a material point, a visco-elastic body, an aggregate of crystals, a source or absorber of energy, fluids, volume, with its body and surface involved in different processes. Except for the very special circumstances, none of these mechanisms prevails so that the others can be neglected. Those mechanisms create the “organized chaos”, turning the seismically active part of the Earth into a hierarchical complex system.

Typically for complex systems there is no fundamental theory that would unambiguously determine a methodology for earthquakes prediction. We have to supplement the theoretical research with exploratory data analysis and numerical modeling and take into account how predictions might be used. Hence the composition of our group.

What is distinctly innovative in the approach developed by your group?

Prediction is targeted at the individual strong earthquakes. Prediction algorithms detect precursory patterns in the permanent background seismicity or other relevant observed fields. These patterns might be either perpetrators contributing to triggering a strong earthquake, or witnesses merely signaling that the region became unstable, ripe for such an earthquake. A simple example of witnesses is reported by Chinese scientists: they observed that snakes crawl out of their holes before the earthquakes. This has a quite reasonable physical explanation – premonitory compression of ground and/or rise of ground water level. So, although snakes obviously don’t trigger earthquakes, it would be a mistake to meet such observations with supercilious lambent patina.

I would also mention among innovations our large-scale experiments in predicting future earthquakes. Regarding the difference from other approaches: ours is complementary to the earthquake forecasting which gives continuous extrapolation of seismicity, counting together strong and more frequent weaker earthquakes.

How reliable are your earthquake predictions then?

Each prediction is formulated by an alarm indicating the time interval and area where a strong earthquake will occur. Like for proverbial pudding, the only decisive test of a prediction method is predicting future strong earthquakes. For about 20 years we are routinely filing our alarms (in advance, of course) on special websites and email them to more than 100 experts worldwide. [Please see sidebar for examples of the group’s algorithms.-Ed]

One can encounter sometimes a mistaken belief that only a precise short-term prediction is practically useful, since it would justify extraordinary measures of red alert kind, like evacuation, shutdowns, enacting martial law etc. Actually arsenal of disaster management includes a wide range of preparedness measures. It takes different time, from years to seconds to undertake different measures; having different cost they can be maintained for different time periods; and they should be spread over different territories – from selected sites to large regions. What additional measures to undertake depends on specific situation in the area of alarm. Prediction is useful (allows choosing the best way to act), if its accuracy is known but not necessarily high. This is the standard practice in preparedness to all disasters, war included. Many indispensable actions do not require high accuracy of prediction.

You asked [earlier] about “the next big one” in California. By existing regulations the newly issued alarms should be distributed only among professional community, lest they trigger disruptive anxiety of population, profiteering and crime. Their further release is at the discretion of public safety authorities; leaks do occur. Our web sites open to public each alarm after it expires, whether it proved correct or false. This information is important as quality control both for predictions and disaster preparedness.

Your group extended this approach to socio-economic predictions (like in the J. S. MacDonnell Foundation project “Understanding and Prediction of Critical Transitions in Complex Systems”). What makes that approach so widely applicable? And who’s going to win in 2008?

Natural and human-made complex systems comprising the global village have many common (“universal”) features. Among them is persistent reoccurrence of extreme events, also known as disasters, crises and critical phenomena. Our group has found partly universal precursors to such calamities.

Besides the earthquakes we have considered prediction of the following events in the USA: Electoral changes of an incumbent party (by Presidential and mid-term Senatorial elections); economic recessions; surges of the homicides in Los Angeles; surges of unemployment. The latter was considered also for France, Germany, and Italy. The reliability of prediction is already established for elections. Prediction of future elections had started in 1994. Since then correct predictions have been made for 128 out of 150 elections to Senate and for all six Presidential elections have been correct. (Our algorithm predicts popular vote and does not consider whether it will be reversed by the Electoral College). For other events only few predictions are made so far. Statistics is yet insufficient albeit encouraging.

Regarding 2008, (our) prediction will be issued in August.

What are the further possibilities of prediction, and what is most important for success: Funding? Efficient governance? Intellectual or technical resources?

It is encouraging, albeit regretful, that many available and highly relevant data, models, and theories are still left untapped. Using them we have a fair chance to raise prediction accuracy by factor 3 to 5, and to predict a wider set of disasters. Of course, (as Mark Twain said), “…prediction is difficult, particularly prediction of the future.”

You [may] ask how realistic this is. That might be irrelevant when we talk on disasters, commonly recognized among the major threats to survival and sustainability of our civilization. Furthermore predictive understanding of complex systems remains a major unsolved problem of modern science, tantamount to transforming our understanding of the world.

Finally you ask what is most important for success? I would put character before these usual suspects (or scapegoats). Quoting E. Burdick, “…of course, things are complicated… But in the end every situation can be reduced to a simple question: Do we act or not? If yes, in what way?”