HW (heartland wife) had already downloaded that paper in "Science" along with a news article about it by Kerr in the same issue: I just read them both. The problem is that the seismometers "trigger" to record only on earthquakes, and the only reason this precursor is known is because a single seismometer (or should I call is "seismograph?") was triggered. Even on that one, only part of the 44 minutes leading up to the major quake was recorded.

So, my question is whether places like Parkfield have continuous recording (my guess is "yes"). The problem is likely that you don't know where the next quake will nucleate and if many of the seismometers need to trigger to record, then perhaps tiny precusors will not be recorded (need opinion of John V on this).

I guess it is fair game to take a couple of interesting quotes out of the Kerr article:

"Researchers are impressed, up to a point.
Bouchon “has wrung as much as anyone
possibly could out of such cra*py data,”
says Stein."

("* added because the earthwaves censor was activated")

"The Izmit observations are both tantalizing
and frustrating, says seismologist
Gregory Beroza of Stanford University in
Palo Alto, California. Tantalizing, he says,
because “this is the sort of thing you’d expect
to see” based on theory. “The frustrating
part is they don’t have enough data to show
exactly what’s going on. But this is going
to cause people to redouble their efforts to
closely examine the beginning of other earthquakes.”
The problem with that, he notes, will
be snaring big quakes in dense nets of geophysical
monitors, because “we don’t know
where earthquakes are going to start.”

This bears on posts by EQF: seismologists are indeed trying very hard to understand earthquakes, and are interested in earthquake prediction but are not able to issue useful predictions yet.

So, the same fault, the northern strand of the North Anatolian fault, passes close to Istanbul. But, it is under Marmara Sea. You may need Ocean Bottom Seismometers (OBS) to record close enough to the fault to record weak precursors (although near Istanbul the Princess Islands are rather close). OBS, I think, have to be recovered to get the data, so is not "real time". The French and Turks are planning a sea floor observatory, which would be presumably cabled to shore so "real time" . But, the future quake could initiate 100 km away and propagate towards Istanbul. But, let's say that they did record in real time an accelerating series of quakes on the same small patch of the fault. Theye don't really know if that will just lead to another M4, of which there have been a half dozen or more in the last decade on the main fault.

Finally, seismologists are not logically going to be interested in prediction schemes that do not make sense with physics, or that are secret, or whose credibility is based entirely on issued predictions that do not beat chance.

Chris