To start off:
there's politics, and then there's politics and then also politics.
On the positive side: we do need some kind of administration to protect the interest of the polity, and represent the sentiment (for want of a better word; the prevailing mind-set) of the polity.
Sometimes that means nobody's allowed to look inside a dead human, but it's perfectly all right to dissect a live dog. Today, some of us believe that's the wrong way around, but it was the attitude of most people at the time. In other times or cultures, it may have been acceptable to test chemical weapons on subject peoples, or it may be forbidden to use airborne chemical toxins in any circumstances.
All the same, the practices of science must have legal limits, regulation and oversight. Science can't be allowed to run amok; we can't have curious persons blowing up buildings or poisoning the water or abducting random experimental subjects.
Laws vary and we may disagree with them - but we
must have influence in the making of them.
In order to ensure that, we have to have a transparent, coherent and comprehensive public policy regarding the sciences, and particularly scientific research. Only governments can make and enforce such a policy.
On the fence, there is the matter of funding.
Many projects are too big and important to rely on private enterprise - and too potentially harmful or beneficial to let private enterprise have sole control of them. Some projects, like the space exploration, epidemiology and climate research, are so far-reaching in their scope that only government-sized organizations can support them.
But governments tend to invest more of our money in weapons research than in medical research - not necessarily in our best interest.
Again, this makes it crucial that the population has influence over funding policies.
On the negative side, partisan politics - like commercial enterprise - can play all kinds of havoc with the orderly acquisition of knowledge and development of technology.
The solution is obvious: to have open referenda, rather than a legal minority representation, express the people's will on policy-making, and arms-length agencies, guided by independent experts, rather than political appointees, allocate funds for research.
(And, fps, put a limit on military research spending already! No matter how much money you give him, Dr. Feynman
cannot make tank fuel out of sand!)
Here is a good start for thinking about government policy:
https://io9.gizmodo.com/10-science-policies-we-wish-the-government-would-enforc-5887189