The scoring was essentially identical to Canie's form. The 6 points break down into 4 points for date, 1 for location and one for size. A vague prediction like "a quake in California this year"
would get 1 point for year, 1 point for location and 1 for size, but that doesn't translate to a 50% hit. Instead, the seismicity of California for an average year was used for a quake of any size, making the prediction essentially worthless.

The program had definitions for many geographic areas; states, continents, countries, etc etc. and could compute seismicity based on the NEIC database.

So almost any prediction could be accomodated with a bit of guesswork on my part sometimes. I just wish I had the code or even a good description of how it worked, but alas, it all died with the program. The really important part of it was that I had a handle on the significance of a prediction, above and beyond simply stating the chance probability involved.

Roger