geo's nostracodus software...

is not predictive. He is doing what we call 'propagation studies**' without knowing it. His software is using a 'following agent' approach that ONLY picks up conscious discussion and counts the words. The ONLY reason that his results show earthquake related words, today, on May 18, is due to the discussions generated by (mainly my) recent interviews and wujo's upon the very very very narrowly focused internet sites chosen for selection.

Igor ran our propagation analysis tool this morning and located 8 forum sites that had nearly all the words shown on 'nostracodus' results. A single site met 93+ percent of the language. So, no, i do not find that there is value in nostracodus approach as it will always, and ONLY, ever follow discussions via counting words already displayed on the net....it is NOT forecasting the appearance of either language or events in the future...merely reflects the momentary interests and paranoia of yesterday and the day previous from the limited sites chosen.

 

** a propagation study is where you create a grouping of words, seed them to various forum sites and then track how far the language spreads on the internet, in how little time, and what the 'dissolution pattern' is upon break up...