Sometime do you think you might speak to why models diverge so much at times and why we don't seem to know at times how to reconcile those differences? To what extent can they be analyzed after the event in order to drive new and improved models (or even to know better pre-event which lane to pick when issuing any given forecast)? For example, I have read that the GFS and ECMWF models differ on the behavior of the closed low off CA by New Year's as well as an upstream trough in the Pac NW, and are therefore "out of phase." Is there a specific model bias to point to as the cause for this or is it much more complicated determining why? Due to the vast amounts of data and inherent complexities, what are limitations of post-mortem analytics (data) and analysis (logic)?