(1) Making 30 researchers spend time on grants when only 1 wins is a waste of researchers time, regardless of productivity.
It's a system built for bureaucrats who don't know to evaluate work-product from those who do by comparison if not mutual critique. It's a wealth transfer from researchers to taxpayers.
(2) Grants always are on topic, but the topic is mostly selected to avoid political embarrassment. In the 1990's, it was HIV. In 2010 it was Alzheimer's Amyloid Beta. Then computer security. Unlike the internet or jet engines, these have not panned out as foundational research (except perhaps for some of HIV). Again, topics are selected not to deliver value to citizens, but to politicians.
Yes, this is inefficient. Yes, academic bureaucracy sucks and the publication game is easily corrupted and early advances are more relatable than later ones.
But the real problem with science in the U.S. is that there are not enough jobs for scientists, because companies are happy to get research and development done more cheaply elsewhere. Companies have only disincentives to develop local labor pools.
That's the negative incentive that dwarfs everything in academia and publicly-funded research.