The problem seems to have occurred in finance and business planning. When the mobile spectrum auctions were announced in Europe, all the bidders started to construct models to tell them how much they could bid.
The general view was that the more detailed the model the better. Makes sense, right? The more detailed and specific your information, the more accurate your predictions must be, and the more sure you must be that you are paying the right amount and really will be able to make the return you think.
Well no. They all ended up with 100s of pages of Excel and predictions whose basis was totally unclear to anyone except the modellers, and maybe not even them, and they ended up bidding largely on emotion, and in a couple of cases they nearly bankrupted themselves.
Whereas the real drivers of return would have fitted on one A4 – but of course, without specifying the values of all the assumptions in enormous detail. When you do that, what you end up with is something that can be sanity tested by looking at ranges of values for variables in a discussion.
What they ended up with was a complete inability to argue about whether the assumptions were reasonable because the model had put them out of reach.
We seem to be in the same situation with climate. Endless detail is not a marker for accuracy, still less usefulness. Multiplying detail does not usually lead to any different predictions, nor to greater certainty, than very simple models. It just makes the process more obscure and less reliable.