According to the  Information Technology and Innovation Foundation, President Obama’s budget has nearly $13 billion set aside for energy-related spending–if you look just at the amount allocated for key R&D programs at the Department of Energy along with spending on tax incentives (there’s more if, for example, you include funding for Department of Defense related programs). Most of that money–$7.5 billion—is going to tax breaks of one sort of another. That is, money that goes to deploying technology we already have. The rest—about $5 billion–is for R&D and demonstration of new technology.

Does it make sense to be spending substantially more on deploying existing technology than on developing new technology? And when it comes to cutting the budget down, as will probably happen since the numbers don’t factor in the sequester, where do you cut? What’s the best use of government money? The nearly $2 billion in tax credits allotted for supporting the installation of existing wind turbines could fund an awful lot of R&D for developing new ones.

There’s no way a blog post can settle these questions. But I want to raise two issues.

First, the longer I report about energy, the more amazed I am by how much incremental technical improvements can lower cost and improve performance. Over the past decade, the cost of wind power has dropped by something like 60 percent, to the point that it’s competitive with fossil fuels in some places. The cost of solar panels has dropped by a similar amount since 2011.  But solar panels and wind turbines look much like they did a decade ago. The Toyota Prius has become profitable for Toyota, even though Toyota hasn’t switched from the old, nickel-metal hydride battery chemistry it started with—it’s just gradually made it and other components of its hybrid better (see “Carmakers Find Ways to Make Cheaper Hybrids”).

While support for existing technologies probably won’t prompt radical innovation, the cumulative incremental improvements it does support can change a technology that’s not profitable into one that is. Industry can be very good at identifying problems with existing technology and then inventing solutions that can actually be applied at the very large scales needed for energy. Researchers in academic and government labs often invent eye-popping technologies that can’t be manufacturered and end up having no impact.

The question is whether those incremental improvements can happen without government funding. I don’t know the answer to that. I only know that much of the improvement has indeed happened in the context of government tax credits and other incentives.

That was all issue one. Here’s issue two. There may very well be applications that require radical new technologies to compete. Someone, at some point, has to invent the transistor, has to invent the silicon solar panel or the cadmium telluride solar panel or the nickel metal hydride or lithium ion battery that industry can then go to work on improving. And there doesn’t seem to be an efficient way to invent those things. There’s a lot of trial and error, a lot of money spent on projects that some researcher think are great and other researchers think are stupid. Projects that a practical supervisor at a company with limited R&D funds might sneer at and decline to fund. But there’s always a chance that the conventional wisdom is wrong, that a radical new idea might actually work. 

I’ll go further. There may be technologies that established industries don’t want to develop because they could be disruptive to their existing businesses. Where’s the money for that kind of R&D going to come from if not from government?

The question is, how much money do crazy inventors need? The answer is it could be more than you think. Investments of just a few million dollars can go a long way to developing a new material or prototype—the sort of thing, for example, that the Department of Energy Office of Science or Advanced Research Projects Agency for Energy fund now. But it can take hundreds of millions of dollars to know whether that technology can reach the scale and cost that give it a chance of success in the market, hundreds of millions before a major company or bank will finance a new factory or biorefinery and start the technology down that path of incremental improvement that could really make it a commercial a success. New energy technology has to compete with entrenched, very low cost incumbents, and it can take a lot of investment to get it developed enough to know if it can.

Indeed, it might even take Solyndra-level investments on the order of hundreds of millions or even a billion dollars to demonstrate a new technology.

It’s easy to find people who were warning about the solar cell-maker Solyndra long before Solyndra became a code-word for government failure. I’ve talked to dozens of experts who say they knew all along that the technology wouldn’t work. And they were right.

Yet there have also been naysayers about the electric-car-maker Tesla Motors, another recipient of government largesse. And Tesla seems to be doing far better than Solyndra (or its fellow carmaker Fisker)—it recently said that it will soon post its first real profits (see “Why Tesla Survived and Fisker Won’t”). People warned that Tesla’s plan for its electric car batteries was stupid. It had decided to use the lithium ion batteries used in laptops, the kind of batteries that in rare cases cause laptops to catch fire. It wanted to wire together thousands of tiny cells in a complex design that critics said would be prone to failure, and that could lead to catastrophic and deadly fires. The major automakers have taken a different route, choosing safer types of lithium ion battery cells that are much larger, so you’d only need to wire together a couple of hundred battery cells rather than thousands.

But Tesla’s plan seems to be working. It’s batteries are performing well, and they cost far less than the batteries the major automakers are using. “Tesla’s lithium-ion battery pack technology is five to 10 years ahead of competitors when it comes to a passenger electric vehicle application, as measured by performance and cost to manufacture,” says Andrea James, an analyst for Dougherty. “Tesla’s battery lead allows it to produce a better vehicle at more affordable price.”

It’s too early to declare Tesla a success—it’s still got plenty of challenges ahead, including continuing to bring down costs. The early critics could still be proved right. But Tesla might just have the crazy idea that works, the one that the established industries wouldn’t have developed, the idea that makes it possible to get to low-cost, affordable electric vehicles, and that other automakers will adopt after Tesla proves it. It’s also technology that may never have been demonstrated without a half-billion dollar loan from the government.