With regards to the reliability of the various GFS runs, would be very interesting to see the anomaly correlation for days 6,7,8. The day 5 anomaly correlation seems to suggest negligible differences in the verification between the various runs.
We can all cherry pick examples to prove a particular point, but I think these verification charts should take some of the subjectivity out of the debate.
My two pence and probably time to move on from this discussion....
Edit: Just seen the reply from Chio with regards stats from days 8-10.