For over two months now I've been attempting to get the posting privileges at James Berkland's "SyzygyJOB" website that were promised to me as a benefit of paying the subscription fee for his newsletter. E-mails to each of the addresses listed at the website, and snail mails to the two addresses provided there and within the newsletter have all gone unanswered despite my repeated and polite requests for admittance via a required "invite".

Consequently I do not feel kindly disposed toward him. I have noticed that he is prone to playing fast and loose with facts as it serves his purpose, and that he is shamelessly self-promoting. I write today to point out just one case of significant variances between the facts as they exist and as JOB portrays them. At his website today, subsequent to a moderate quake 160 miles offshore from Oregon, he claimed a hit - stating that his "Seismic Window of Nov 11-18,2008 gave 80% for a quake of 3.5-6.0M with an Oregon or Washington address." Only problem is, I have his written "seismic window" from his Nov. newsletter in front of me,and it does not provide any probability, its window extends from Nov. 13 (not the 11th) through the 20th (with no time standard provided), and has a broader range than stated by him today at his website (3.5 to 6.5, not 6.0). I leave it to the reader to decide whether an epicenter 160 miles offshore from Oregon counts as an Oregon "address".

I'm inclined to give him a hit despite his rather wiggly parameters, figuring that with all his predictions he is bound to get one right once in a while.

But what really offends my inchoate mathematical sensibilities is his methodology for computing his "success (rate?)". Maybe Roger can advise whether or not this makes any sense: Berkland assesses his October success at 67.5% by giving himself zero for one prediction, 70% for another prediction wherein the fulfilling quake was below his lower magnitude threshold but within his time and location parameters, and 100% each for two additional predictions apparently fulfilled (I did not independently verify any of the time/space/magnitudes provided by Berkland). Now here is where it gets particularly dicey. Berkland then adds those four percentages together to get 270 then divides that figure by 4 (the number of predictions) to get the result of 67.5%. He claims a long term success rate of 75-80%. Does this figure actually mean anything at all?

Mike W.