Are you talking about assertion #3?
I wasn't but I accept #3 as responsive and concede the point.
On the other hand he basically says, "Why yes, we do use human intuition to tune the models every year rather than let the math rule." For a system that's novel and investigatory I'd be inclined to let that slide, but for a system being used to justify trillions of dollars in economic change, I find the models' slipperiness concerning.
scientific consensus arises naturally
There are a couple of meanings that get applied to the word "consensus." One of them is "general agreement," as in no participant objects. Another is "the judgement arrived at by most of those concerned" which is a slippery version of majority vote. There are even variants like "strong consensus" meaning explicitly 100% concur and "rough consensus" meaning that the remaining holdouts can like it or lump it.
Which is your version of "scientific consensus"? My version is that if you have to prove that consensus has been attained then it hasn't. Because actual consensus is just that obvious.