"> 1. Predictor Id ... Number, name, what?"

Probably a numeric ID of some sort. Leave it to the predictor if they want to publish a name or contact info ...

"Some (Berkland) insist on local time."

The issue I have with that is time zones. What if you have a prediction that spans a geographical area with different time zones? You have AZ that is in daylights savings time denial. Parts of Indiana the daylight savings time used to vary by county. Congress then changes what day daylight savings starts or ends ... Oops, I have to change a bunch of code. I'd also have to keep track of special time zones world wide. Gets pretty ugly pretty fast. UTC is the gold time standard in seismology, I believe for good reason. Using UTC reduces ambiguity and increases the ease for folks who do analysis :-)

"Line? That's pretty specific if you mean exactly on a line. Straight or crooked? Crooked requires multiple corner coordinates."

I do not mean exactly on a line, although a predictor could do that with a range of zero. Crooked is what I mean when I say line segments and yes, this would mean multiple corner coordinates. A line or line segment prediction would have a range parameter, which would mean a distance of x km/mi from some point on the line or line segment.

"> 9. Range for non-polygon predictions ... A circle requires center point lat-lon and radius."

Yes, it certainly does. A circle prediction would result in the circle prediction type flag being set, a range, and 1 coordinate pair.

"I can give you NEIC place names for a given point. The problem is working forward with things like "SSE of San Bernadino"."

What is the resolution, where is this database? I'd be interested in looking at it.

"You thought this would be easy, didn't you? 8-)"

Most things worthwhile are generally not not easy lol ...

Michael