I partly agree. A conceptual breakthrough always rests on a foundation to which many contributed. All of whom, in some sense, contributed. But my reading of history says that the reconceptualization that leads to intellectual breakthroughs themselves usually only involve small numbers of people.
If you've read The Structure of Scientific Revolutions, what I'm saying is that new paradigms are usually created by very small numbers of people. But they have both a foundation and their further success from the contributions of many.
I'm very much not offering an opinion on a great man theory of history in fields outside of science. Your example of the American Revolution is entirely off topic.
I'm also very much not saying that who will contribute what is in any way predictable. At best, the necessary collision of circumstances to make the breakthrough possible is chaotic, and therefore cannot be predicted. Nor did anyone else. The original point a few posts up was that, even if though there might be a haystack of clearly wasted effort, there may still be a needle powerful enough to make up for the rest.
All good points, but remember the claim in question was:
> But if we pay enough of these people to sit in rooms and work on problems, maybe one of them will figure something out.
and the response that you called "very reasonable" was:
> There’s more than enough already. (And, historically, you only need less than a dozen.)
So you were agreeing with someone who said we are paying too many physicists. There are too many people studying this problem. Okay, let's get rid of some then. Which ones?
> I'm also very much not saying that who will contribute what is in any way predictable
Uh oh, then how do we know who to get rid of? Which physicists should we not be paying? The claim that we should fire a bunch of scientists because we "only need less than a dozen" is nonsense, and you called this claim "very reasonable", with more examples. But maybe I should have replied to that person instead. It's a little awkward trying to have an N-way conversation when you can only reply to one response at a time.