This is incorrect, and as a counter-example I point the fact that millions of dollars every year are "invested" towards perpetual motion machines.
There are three large organizations with enough funding to construct, say, a super-collider; government, church, and corporations in order of influence and funding. So since the pot is bigger, it is simpler for someone to apply to NSF for funding than petition MegaCorp. But yet many large corporations fund ongoing basic research. New directions are been explored in networking theory, mineral exploration, and medicine every day.
Perhaps the main difference is that governments spend significantly towards weapons development, and as a corollary protection from new weapons. This has produced innovations such as nuclear power, biological engineering, ARPANET, and rocketry. Corporations are generally excluded from weapons research unless granted specific permission from a government, and hence were excluded from first developing these innovations.
And if you focus in on technologies championed by politicians instead of those going on deep inside government labs to address real strategic needs, you see even less success.
Government organizations can and do innovate, but like you said, its usually for defense purposes. Which is fine with me. Like many people, I don't enjoy invasions or nuclear strikes.
When APRANET was developed, I think it was more of a side project than a campaign issue.
The folklore that I know is that ARPANET was created to provide communication channels across the US in the event of nuclear war. I should find a reference for this.
No, it just means less imaginative research. And more research by people good at filling out paperwork and kissing ass. These are not the iconoclasts for the most part.
If we do that, we only end up with things that can reasonably be expected to work.
And things that can reasonably be expected to work rarely change the world.