Hey All,

As I continue to mull over the possibility of building an internet based repository of earthquake predictions, I am trying to determine what are the critical components of a scientifically meaningful earthquake prediction.

As I have for years now, I continue to consider creating a web form or site to accept the critical components of a prediction, and then publishing those predictions via RSS, http, ftp, or by whatever method to anybody that is interested in the data. This would simply be a transparent, reliable data registry and relay service, providing no analysis or commentary whatsoever.

It would seem to me that making predictions available in this manner would solve most of the problems with predictions, their interpretation, and their validity, and in particular, scientific accuracy, timeliness, and the ability to easily analyze predictions or thousands of predictions using automated means.

The bottom line is I would like to see the arena of earthquake prediction standardized and legitimatized, because quite frankly it is currently dysfunctional and useless, and I hate seeing any scientific endeavour in that state.

I would like to pin down what the critical elements of a scientific earthquake prediction are. I swiped the text below from Wikipedia as a starting point. I also looked at the prediction form on this website. Specific Area, Specific Magnitude or Magnitude Range, and Specific Time Window I think are pretty much items that are understood to be required parameters by the scientific community, correct?

A physical basis, as explained by the Wikipedia text below, is not required, but would be useful to the scientific community, I suppose if a predictor showed some promise, others would want to learn more about the method(s).

The one item I cannot get my head around is 'Estimate of probability compared to random chance'. I've always been of the belief that either a prediction worked or it didn't, sort of like you are either pregnant or you not, there is really no gray area. I don't see how it has any place in a earthquake prediction.

What does it mean if a prediction is made with say an 80% probability? How can one tell if a prediction was useful or not if there if there is a percentage attached to it? Seems really pointless and invalidates any scientific validity if a prediction is made with a percent attached to it. Am I missing something here or is saying an earthquake has a 80% chance of happening is just a bunch of B.S.? I would think 'Estimate of probability compared to random chance' would be something that would be left to somebody analyzing a prediction result rather than something that would be provided by a prediction itself. I'm thinking this would be Roger's department lol ...

I understand that seismic experts, Dr. John Rundle comes to mind, have made forecasts, not predictions, with percentages attached to them, and I can say I do not fully understand that either. I am guessing, for example, he is saying an 80% chance that area XYZ is going to have a >5 quake over the next 10 years is based on the long term seismic history of area XYZ, and the 80% would basically be a calculation based on the average recurrence rate for that particular area over a given time frame. Is this an accurate assumption or am I way off base?

This lay person would appreciate any input regarding the use of percent chance with predictions and/or forecasts and their usefulness, if any, with regards to the scientific method.

Thanks,

Michael Tolchard

==============================================================================
http://en.wikipedia.org/wiki/Earthquake_prediction
==============================================================================

A meaningful earthquake prediction must have all the following elements:

* Specific area
* Specific magnitude or magnitude range
* Specific time window
* Estimate of probability compared to random chance
* A physical basis

A meaningful 'forecast' does not require the same accuracy, rather the term refers to prognostications on a longer time scale that are usually probabilistic.

Note that a physical basis is most critical for a prediction to have meaning to the geological community. Any method to make accurate predictions would be welcome and meaningful to many people even if no mechanism were known (much like doctors are now investigating diagnostics through statistical analysis of gene expression microarrays, even when the mechanisms linking particular genes to conditions remain unknown), as long as it was repeatable and the reliability could be quantified.