Before these guys get even more funding to add even more stuff to their models they should debug their existing code first. The level of disagreement between th

Before these guys get even more funding to add even more stuff to their models they should debug their existing code first. The level of disagreement between the existing set of models has been increasing over time, not decreasing. We're told the science is certain but it's been getting less certain as the models get more complex. They also often suffer numerical instability problems in which model runs end up as hot as venus or Earth as an iceball. Rather than fix the bugs these runs are simply discarded and they try again. Models are often full of hacks too where things that should be program-stopping assertions are just wallpapered over by adding a min/max call to the values.