The counter-counter-argument is that if you don't decouple the "doing of the scientific research" from "influencing human choices", you might compromise the endeavor of scientific research altogether.
The same institutions don't need to do both, and that they are separate is usually because "is" questions and "ought" questions are answered via different mechanisms.
> The counter-counter-argument is that if you don't decouple the "doing of the scientific research" from "influencing human choices", you might compromise the endeavor of scientific research altogether.
I don't disagree with you. Put more concretely, I think we need a blend of:
1. Allowing scientific concensus to influence the choices we make today. We should act on our current best knowledge of the world.
2. Allowing politics and our understanding of what humans need today to influence the way we allocate scientific resources. Critical problems and promising solutions should be given some priority because they have the greatest chance of improving people's lives.
3. Funding "pure" scientific research where we have no idea if it will pan out or not. 2 is biased because we don't have a perfect understanding of what research will pan out. Without some random resource allocation we will only hill-climb to the nearest local maxima of things we know are improvements. Sprinkling some fraction of our resources outside of obvious priorities allows paradigm shifts and serendipity, which are vital for progress on longer timescales.
I think a lot of science nerds, especially here, focus on 3, and I agree that it's important and undervalued. But I think it's highly inefficient to only do that. Smart resource allocation does give greater weight to areas with greater probability of success, and to areas that provide the most benefit.
The same institutions don't need to do both, and that they are separate is usually because "is" questions and "ought" questions are answered via different mechanisms.