that's not entirely true, it is to some degree. by convention there have been a few buffer layers between actual grant allocation and naked politics. funding gets allocated to someplace like NSF, NIH, ONR or DARPA. Those organizations have directorates or area concentrations. Each directorate has a program manager (the terms vary based on org) who puts out request for proposals (grant applications).

The PMs are generally chosen from the sciences, and are responsible for authoring RFPs that meet strategic goals, and negotiate with the PIs (grant recipients) about terms and sizes and such.

So there are really two political realms, above the funding agency, and underneath, and its entire function is reconcile those worlds in a pretty vague way with a certain amount of autonomy given to the PM.

This isn't 100% great, but if you have good PM, some good science does get funding. While this seems like a lot of machinery, if you short circuit all of it, and have the presidents direct flunkies make funding decisions, that basically means that almost no real science gets done.