Imagine you were building the first fission bomb, but doing it the Calico way (which is the way most science is done these days). You'd say: we need enormous progress in a wide variety of very distinct fields: nuclear physics, isotope chemistry, metallurgy, high-speed photography...
So you'd ask PIs in all these fields to send you grant proposals related to the problem of building an atom bomb. They would take their existing grant proposal template, rewrite it to talk about atom bombs, and send it in. You would send them money. They would deliver progress reports.
What would happen? Probably, you'd build an atom bomb. The first test would be in 1986, after 45 years and 500 billion dollars. It would cause major damage to three creosote plants and severely startle a jackrabbit.
Next, imagine Calico's funders said: we will take this same amount of money, and pour it into a single problem -- rejuvenating a dog. Much like Manhattan's single product deliverable.
Moreover, we will not hire existing players to repaper their existing lines of research. We will hire the best new PhDs and tell them to do whatever want to do, so long as it represents one piece of a reasonable strategy for, or step toward, an realistic engineering procedure for rejuvenating dogs. (Or at least mice, if long-lived animals are too hard a first step.)
In this kind of organization -- ie, a hierarchical company-shaped structure with a company-shaped goal, delivering a product -- the standards of relevance and collaboration will be much, much higher. Los Alamos hired the best people in the world, but no one took their existing research and tried to shoehorn it into the Manhattan Project.
No wonder our grandparents could get big things done, and we can't. We actually put a lot of money into science and technology, and the personnel are just as talented as ever. Every PI is an Oppenheimer. But there is no General Groves.