Comment by HerbManic

There is also the additional issue of computer models constantly chasing global changes. About 10-15 years back I used to talk with folks that worked on weather modeling and they were in a state of frustration in that as soon as they could make models that could work on older data sets to do reasonable predictions, the global weather patterns had change just subtly enough that it made them just kind of average on forward predictions.

This was right before GPU compute started to become a big thing, I do wonder if they now use machine learning models on these to speed up model iteration? I would hope so, but even then there is the human factor as you said. Eventually someone has to make the call on what the data shows and how to present it to the world.