The results are mostly free and open once published. It's incredibly nichey; many scientists dedicate their lives to a very narrow area of study. Nobody, DARPA and the NSF included, is great at predicating what is going to bear fruit years down the line.
And yet science happens.
It's not subject to a crippling "free rider problem", even though the whole world benefits. It's not massively corrupted by bribery.
Are these just what industry folks say when they want to be subsidized by taxpayers? It's ok for core R&D to be publicly funded, but all these problems will emerge suddenly if they bring it to market, so that last mile has to be left to us.. and the profits too?
How do you know how much of something to produce? Sure, science happens. Enough? It's very difficult to say.
With a market, outside of big market failures, you can be reasonably sure that, roughly, enough of the right kind of a given good is produced because of the feedback mechanisms involved: South Korea has abundant food - more than enough to feed everyone. North Korea does not.
If the government were the sole purveyor of software, sure, it would still get written, but I think that not enough would, and that it would turn into a big ugly mess.
Or for that matter, high tech. Oh, you thought high tech isn't a product of a planned state-controlled economy?
It isn't. The government is one actor, and certainly has many contributions - the Internet is a big, obvious one - but there are certainly a lot of private actors that count for something too. Apple, IBM, and Microsoft come to mind. So "product of a planned state-controlled economy" is what I would call either "science fiction" or "wildly inaccurate", just as I would say that "high-tech is entirely the result of the free market" lunacy as well.
http://www.amazon.com/dp/B001C6MQA8/?tag=dedasys-20 is a good book to read for some of the history of Silicon Valley. Private companies figure prominently in it.
> Presumably you don't think modern science is a "big ugly mess"
Actually, my wife does research at a university, and I very much do think there is a lot of messiness in science, and think it likely that it's vastly underfunded - in part due to the fact that it's difficult to bring market mechanisms into play for things like basic research.
> Why would government-sponsored software development be any worse than corporate-sponsored?
Let's see:
* Free rider problem goes global: everyone hopes some other country pays for useful stuff.
* Governments are not going to be as focused or concerned about niche businesses.
* Corruption. It becomes worthwhile to pay off politicians to have them choose your team.
* Calculation problem: http://en.wikipedia.org/wiki/Economic_calculation_problem
* The NSA. Think they're bad now?
The current market has problems, but I just don't see any evidence that moving software production to the government would be much of an improvement, and it might be a whole lot worse.