How should the US government stimulate job-creating innovation?

Thomas Friedman excited some discussion with a New York Times column suggesting that the government invest some of the stimulus money in leading venture firms, on standard venture fund terms and conditions. (Specifically, he suggested what's called a 20% "carried interest" for the VCs.) Friedman's premise, as I understand it, is that -- like other financial services outfits -- VC firms are short on capital right now. And if anybody gets more, they should, because VC firms tend to invest wisely, and furthermore have an interest in socially useful technologies such as clean energy.

To a first approximation, whether this is a good idea depends on whether VCs really are short on capital. Fred Wilson argues that venture capitalists actually have too much capital, which is why VC returns have soured. Wilson does have a bias here, and I don't just mean "Please don't give money to my competitors to help them bid against me for deals." He's the prototypical internet-age VC, doling out small amounts of cash to companies that can bootstrap themselves with a couple of pizza-eating young programmer-founders. But he's not alone in his view. For example, a not-at-her-best Sarah Lacy offers a similar opinion, but with more snark and less coherence. Don Dodge, while seemingly agreeing with Wilson, offers an alternate plan -- put government money into a slightly different kind of VC vehicle, namely incubators, thus creating jobs for lots and lots of young pizza-eaters. But that doesn't address the core of Wilson's argument, which is that the best VCs don't need capital, and the rest don't deserve it.

Friedman further suggests that it's good for government to sponsor worthy industries like clean technology. Here the historical evidence is mixed. There have been some huge successes of that kind, such as multiple DARPA projects or the Indian software industry. But there have been many more failures, such as Japan's Fifth Generation Computer project. Generally, a big reason for the failures is that government bets on misguided, over-optimistic ideas -- which, come to think of it, is how a lot of VC investments fail as well.

There's a further problem with Friedman's idea in its simplest form -- who says the jobs created will actually be in the US? For years, VCs have favored outsourcing product development to cheaper countries. That's good business, and it provides boosts to economies that may need them even more than we do here in the richer United States. But enriching India is not exactly a top goal of US public stimulus spending.

An alternative strategy, which Friedman praises too, is subsidy. There are two main arguments for subsidies:

• If a business has positive externalities, such as saving the planet (clean tech), it should be rewarded by appropriate subsidies, thus encouraging it to grow.
• Subsidies for a growing industry might help it get to a "critical mass" -- e.g., economies of scale -- after which it can grow and prosper on its own.

Subsidies can take many forms. Most popular with politicians are tax credits. Popular with economists is a guaranteed market -- e.g., government will buy at least 100,000 carbon-neutral cars, if anybody comes up with a way to make them. On occasions when technology invented for defense/aerospace purposes translated when into the private sector -- e.g., Arpanet -- the latter approach has had some major successes.

This is all a complex and worthy subject. My own leanings are to a hybrid approach. I'll try to work through that more cleanly in followup posts.