Maybe pay people who's only job it is is to talk to the researchers and write the proposal for them? Someone smart enough to get stuff explained to them, but with the communication skills to boil that down into something the money people can understand?

It's a pretty common position in software engineering because programmers and business people are pretty bad at communicating with each other.

No, it only takes a long time because there's so little to go around. Do you think defense funding takes months and years to award grants? No.

Create gov science centers for each major branch of science, provide funding. Allow them to delegate within their narrower and narrower fields with loose requirements such as x-y% is salary a-b% is resources, and maybe something like each new study can get no less than $z and no More than $r.

I'm not saying this is perfect but spending more money towards it in general and allowing some branch delegation of funding would hopefully at least resolve the grant writing part and ensure salary. Though I'm not sure how one would ensure that they are being productive and not doing frivolous things on purpose. Perhaps q amount of hours a year must go to a gov decided research project and the rest is up to the researcher.

Maybe funding for a project is aquired through hours contributed to projects the gov deems with a standard for high social benefit? I.E. You help with the research on this new hydro electric tech (regardless of outcome because we feel it's an important study topic) and we pay ($p per hour spent on hydro tech) towards a study of your choice.