Minor powerlevel but I've been involved in very basic modelling projects for biological cycles (not related to climate).
If you're attempting to model something that's well understood in terms of how it functions then your model will change much more slowly and generally won't change much at all, perhaps some values will be tweaked slightly over time to be more precise etc but fundamentally the model won't change.
On the other hand if you're attempting to model something with many variables and causes such as climate then your models SHOULD be changing constantly to reflect improved understandings in any of the 100's of factors your model might include. Even very basic and broad models should be updated to improve their accuracy - if a basic model proves to be basically accurate then it should be added to so that it becomes more complex and a better reflection of reality. The stuff I was involved in had maybe 7 or 8 variables which weren't constants, tweaking the value of those variables could produce hugely different results depending on what numbers you used - we're not even talking huge tweaks here, changing a value from 1.5 to 1.8 (both perfectly reasonable values depending on which data set you used) could change the functionality of the system completely, from a system experiencing strong positive growth to a system experiencing catastrophic collapse.
Bob definitely doesn't understand this, he thinks there's a big science machine where men in lab coats put in numbers about rain, pollution and heat and get a neat little graph as an output. A graph which should be treated as holy scripture and never questioned to any serious degree.