Ah, models (of the mathematical/projection kind). Wonderful things. Entirely dependent on their inputs and assumptions.
I like what both of you are saying.
But results of a model need not be any guide at all. The 'results' will be entirely defined by their inputs (which is a point you made LTO). I wouldn't discount that modelers, guided by 'experts' have made assumptions 'conservative' in their nature and if there are multiple conservative assumptions, then the 'average case' will be ultra-conservative.
Like LTO, I've constructed models for a living, for a while - financial models of companies. I could make them crash, or do better than Apple by a few subtle tweaks which, on examination, it would be bloody hard to detect!
I'm happy for the experts to tell us what we should be doing in the pandemic. But I reserve the right to be highly skeptical when they quote 'models' - just as skeptical of 'models' in a certain-other-discipline which I won't mention here