[Roger Hunter]

(10-10-2016, 12:55 AM)Skywise Wrote: This post is not directed at anyone in particular. It applies equally to current posters and any future posters and is based on what I and others have argued for years past.

I would like to advise caution to anyone who does quake predictions. Posting a bunch of predictions doesn't make you a predictor. It's tantamount to throwing a bunch of darts while blindfolded - eventually you'll hit the bulls-eye - but you're not going to make it in a professional competition.

A few high quality predictions backed by a reasoned methodology is far more useful than what on the other end might just be a bunch of random guesses.

It's simple statistics and probability. And I don't mean a probability that the predictor places on the prediction. In reality it means nothing unless that probability is actually derived from a statistical analysis of quake history.

A quality prediction has at a minimum the following information:

• a well defined time window - quite simply, a precise start and end time.
  
• a well defined location - one should be able to unambiguously draw the prediction region on a map with no argument.
  

• a well defined magnitude range - at least a minimum magnitude and optionally a maximum or "greater" can be used.
  

Predictions should also be useful. To use an extreme example, saying there will be a magnitude 2.5 or greater quake in a region covering the west coast of the United States for a time period of several weeks is NOT useful. For one thing it's pretty much guaranteed to happen. Further, what good is it to anybody? Unless you are literally sitting right on top of it no one will even feel it. (I've felt quakes as small as 1.2 within miles of me)

On the other hand, predicting a magnitude 5 or greater in a circular region only 300 miles across centered on a specific latitude/longitude with a short time frame such as one week is useful. First off, mag 5 quakes and bigger don't happen very often, even in very active areas (minus aftershock sequences from even larger quakes). There is an historically low probability of a successful prediction in such a case. Continued success with such predictions implies a valid methodology and warrants further investigation. Keep in mind that even if you do have a hit on such a prediction, one time can still be dumb luck.

Another issue to address is the methodology. If the way one comes up with the prediction is, to be blunt, just plain silly, it's not useful and likely to garner harsh criticism and possibly ridicule.  Sorry, but we don't accept tea leaf reading or crystal balls as a valid prediction methodology. How one determines the prediction should be unambiguously definable such that another person reading those instructions would come up with the same prediction.

And secret methods are not acceptable. What good is it if no one else can use the method to protect themselves from a potentially damaging quake?

"Work in progress" methods are fine so long as the predictor is open to changing their method when it is shown to have errors. That's the "progress" part. Otherwise it's just being closed minded and a waste of time.

I would like to continue to have this prediction forum, but I'd like for there to be quality content.

Brian

Hi Brian;

I'd like to suggest two changes.

First, dump Richter. I've tested him before and he's just throwing darts blindfolded.

Second, change the description of the Prediction thread, omitting the wild guesses part.

Roger