They were wise where they were drawing on and applying the near-consensus of the profession, and unreliable in their particular areas of expertise

From I Really Like Oded Galor’s Very Good & Very Brand-New "The Journey of Humanity: The Origins of Wealth & Inequality" by Brad DeLong.  A book review but incidental to the review is this anecdote.

I remember showing up at the Clinton administration treasury department in the spring of 1993, and having coffee with one of the CEA senior staff economists, and asking him what he thought of Clintons CEA members—Laura Tyson, Joe Stiglitz, and Alan Blinder. He said Laura was absolutely first-class, wonderful—on everything except trade issues, where she was a little weird and had a few hobbyhorses of her own to ride. He said Alan Blinder was truly great—except that on macro issues he expected the government to be able to successfully fine-tune the economy more than was reasonable. And he said that Joe Stiglitz was superb—except that wherever there was some micro issue that hinged on market structure, he had a proposal for a crazy restructuring to rejigger people’s information and decision sets to make everything much better. In short: they were wise where they were drawing on and applying the near-consensus of the profession, and unreliable in their particular areas of expertise where they thought they knew better than the profession.

Philip Tetlock's research is very good at demonstrating and quantifying the limits of expertise.  His finding is that their Achilles heel tends to be when they move beyond their own narrow field of expertise and opine with equal vehemence and equal sense of self-authority on matters beyond their knowledge domain.

His work supports the notion that a small group of equally accomplished (similar IQs, similar range of professional achievement, etc.) generalists will almost always interpret new data and make better forecasts than a single expert.  

DeLong seems to be arguing the opposite, that there can be solid general domain experts but you have to distrust them in their narrow domain.

I am confident Tetlock is right but I feel like there is something to DeLong's observation.  I think it is more than that Tetlock's observation is more universal and DeLong's may only be intermittently true.  There is, I think. some issue of category definition.  

Perhaps the alternative observation to DeLong's might be "The greater the general achievement in a domain of knowledge, the more probable it is that an expert will have a narrow area of eccentricity."  

Of course there is the opposite problem which we have just experienced with Covid-19 when the "expert" community rejected the "near-consensus of the profession" and focused exclusively on eccentric and non-consensus choices.