JimD,” You went off on a tangent.”
No, you have this wishy washy some models are this some are that blah blah. I have this, why is that model high or that model low approach. Models that more closely “get” absolute temperature tend to forecast less warming, models that grossly under-estimate absolute temperature tend to forecast more warming. The reason is that many parameters require an accurate initial temperature. Supposedly if you run the models long enough they will “discover” boundaries even if the initial conditions aren’t close. That may not be true if the parameters are “calibrated” incorrectly. Water freezes at 0C if it is fresh, about -2C with standard salinity and between -2 and -50C in the atmosphere. You are ahead of that game if you initialize temperatures to realistic values instead of hoping the models will discover the right values. That is simple for most folks to understand.
One of the more critical model parameters is convective triggering. Somewhere between 27C and 28C there is a greater potential for deep convection. Deep convection is a beyatch of a parameter because it impacts all levels of the system from sub surface to stratosphere including ozone, stratospheric water vapor, cloud cover, pole ward or wall energy transfer, surface wind velocity, atmospheric relative humidity, precipitation, arctic winter warming, sudden stratospheric warming, sea ice concentration and likely a few more fairly important feedbacks.
The people that should be the most critical of the model performance should be the modelers themselves. Many though have shifted to the circle wagon defensive mode so it is impossible to criticize their “babies”. When “I am sorry old chap, but why is there this discrepancy?” stops work you tend to get a bit more direct, especially if you are a redneck with not a great deal of patience. Squeaky wheel and all that doncha know.