(2017-12-20, 11:34 AM)DaveB Wrote: I wonder if it would help if someone compiled a list of events that must eventually come true. For example, elderly scientists involved with the GCP, deaths of other named celebrities, earthquakes is places where cities are built close to fault zones, etc.
It would then be possible to track the performance of the GCP without the constant niggling doubt about whether the effects are due to cherry picking.
That's more or less what the GCP did in the period 1998-2015, with its "Registry of Formal Hypotheses and Specifications":
http://noosphere.princeton.edu/pred_formal.html
They didn't try to define the events before they happened, but they did define them before looking at any of the data, which is the important thing. (Peter Bancel excluded a relatively small number of the events from analysis because they weren't satisfactorily pre-specified.) The final list contained 500 events, and produced a Z value of 7.31, associated with a p value of 1.333x10^-13. At that point they ended the formal experiment, believing they had sufficiently proved that there was a real effect.
They are still analysing individual events, and if I understand correctly the hypothesis for Robert Jahn's death was pre-specified in the same way as those in the formal series. But admittedly the difference is that they are no longer committed to publishing the result in every case. (That's as far as I know, and I may be wrong about it.)