As someone who both reviews and writes grants to the NIH, I can say that most researchers would completely agree with this. There is no way to distinguish a grant in the 10th percentile from one in the 30th percentile. It’s just not possible to predict the outcomes of research to any higher precision. Further, the most interesting, risky, exciting and innovative grants invariably fall in this range. The highest ranked grants are “known knowns”, projects that have already proven their importance and are simply continuing to mine the already discovered vein.
The bottom line is that the scientific system in the U.S. simply breaks down when the paylines get this low. It’s now a combination of luck, politics, power and prestige.
A lottery system would be more honest, but I suspect that grant writers would feel like they are losing some control over their fate, because at least now you have the illusion of control.