Writing OSS is close to research, especially when it comes to projects like new programming languages. Even then, my point was that government stepping in to deal with cases of market failure is not a radical idea. And the NSF has an answer to all of the questions you raised.
The reason government funds research is because the value provided by basic research cannot be easily captured and sold. This same logic applies to OSS.
Why is it ok for taxpayer dollars to go toward research but not toward building stuff that wouldn't otherwise get built?
new programming languages. Here's one that really gets me.
New programming languages aren't really coming up with anything that's particularly new. They're just solving the same problems in different ways. This is one area where I would be absolutely against taxpayer funding.
Consider if we decided to constantly invent new spoken languages, in order to communicate with each other more concisely. Maybe English 2.0 would come with 50% less ambiguity and 20% less verboseness. Maybe version 3.0 would do away with the letters Z Q X. Would that be a good use of taxpayers money? Shall we also fund all the translation of libraries books to the new language?
Constantly inventing new programming languages (I would say) often does more harm than good.
The issue is that furthering science clearly has benefits to society. Making yet another open source hip language probably doesn't.
People enjoy hacking. Things get built usually because it's enjoyable, or because there's a real need for it. I think the current situation works pretty well.
Oh, ffs. I wasn't suggesting the government sponsor a million programming languages. I was suggesting they sponsor Clojure.
Now, Clojure may or may not be a project worth funding. That is a question worth exploring. But what about say, making better tools for programmers? Building experimental user interfaces? Any of a million other things you or I could be working on that are hard to translate directly into cash, yet could contribute greatly to society.
Imagine being able to work on these things full time, with a steady paycheck; being able to freely share our code and discuss our work with one another, cooperating to make it better. Tell me that doesn't sound nice.
Hobbyists can only do so much in their free time. Full time development only gets done when it forwards a narrow corporate interest. And then you have the free-rider problem. But now I'm repeating myself, so I'll leave it at that.
The reason government funds research is because the value provided by basic research cannot be easily captured and sold. This same logic applies to OSS.
Why is it ok for taxpayer dollars to go toward research but not toward building stuff that wouldn't otherwise get built?