The energy department once had an Advanced Energy Projects division, which, until it was abolished in 1998, was the incubator for cutting-edge energy research. Started in 1977, it funded what Ryszard Gajewski calls the “infants and orphans” of energy research. Gajewski, a former director of the division, says that the time from proposal to funding was sometimes as little as two months, even though the division relied on peer review of proposals, which DARPA does not do.

Will the DARPA model work at another agency? Critics have serious doubts. For one thing, DARPA has one customer, the Pentagon, which buys the products that result from its innovations; ARPA-E will promote the development of products that would have to be taken up throughout the economy to make a difference. But adoption of its innovations will depend on vagaries of policy-making far beyond its specific purview or even that of the Energy Department itself.

For example, ARPA-E might successfully identify revolutionary fuel-saving technology for vehicles but fail to see it break into the market because Congress or the White House declined to tighten fuel economy standards.

Adequate funding, of course, will be crucial to ARPA-E’s success. Since 2006, various legislative incarnations of the bill authorized annual appropriations anywhere from US $300 million to $1 billion. Where will that money come from? “I would guess that an ARPA-E will have to be funded at the expense of existing DOE programs,” says William Happer, who ran science programs at the DOE in the early 1990s. But King notes that a bill adopted in January would set up a fund for clean energy research—including ARPA-E—using money from discontinued oil and gas subsidies.

As for the question of what specific problems ARPA-E can try to solve, its advocates said dwelling on that would put the cart before the horse. Through July, House Speaker Nancy Pelosi (D‑Calif.) was still concentrating on getting the legislation passed before hammering out a plan for funding the agency or determining how to use its findings. “That’s point C,” said senior Pelosi energy aide Amy Fuerstenau, speaking back in July, “and we’re still trying to get to point A.”

Longtime critic Goldston thinks that this strategy invites long-term failure. “What happens in an agency in the initial year or two sets the culture in that agency for decades to come, and it’s very hard to change,” he says.

A cautionary tale of what could happen to an ill-conceived agency, as critics see it, was in the creation of another DARPA clone at the Department of Homeland Security. HSARPA ended up getting lost in the giant, newly created homeland security bureaucracy, unable to function as the nimble, red‑tape-cutting unit envisioned when it was set up as part of the Homeland Security Act of 2002.

But does ARPA-E have to end up that way? King says HSARPA veterans helped modify the ARPA-E legislation to avoid some of the pitfalls, especially a lack of autonomy, that hobbled the homeland security research agency.

Much of the Energy Department’s research, in particular that undertaken at national laboratories, is universally recognized as essential. But the pace of research is often glacial, and the culture is notoriously risk-averse. High-risk projects demand a certain tolerance for failure, and congressman Gordon says that the DOE has none. Meanwhile, responsibility to shareholders makes private industry cautious, and university research depends mostly on federal funds. But by acting as an independent manager and venture capitalist, ARPA-E could get the three sectors to collaborate.

Consequently, ARPA-E could kick-start projects that no other single entity would dare to support alone. Whereas the DOE advances mainstream biofuels and advanced solar research, for example, ARPA-E could step in to fund promising but orphaned contenders like solar paint or alternative biofuels feedstock, like kudzu. “It takes an integrated program to do biofuels right,” says Steven Koonin, who directs BP’s $500-million-per-year R&D effort. “The DOE has not been able to do that.” An alliance with the Department of Agriculture might be needed for alternative biofuels research, for example.

The ability to reach across departments might be the agency’s best attribute. Former DARPA director Frank Fernandez says that DARPA consistently collaborates with the military services, other DOD agencies, the CIA, and the Department of Justice, among others. “This is because DARPA can only take ideas to the prototype stage,” he says. After that, another entity has to take over. “I would think that this kind of collaboration would be an essential part of ARPA-E’s business plan,” Fernandez says. In fact, Fernandez hopes to see DARPA and ARPA-E work together on projects of mutual interest.

The bill creating ARPA-E, H.R. 2272, has been dubbed COMPETES, for Creating Opportunities to Meaningfully Promote Excellence in Technology, Education and Science. Its more comprehensive provisions include sharp increases of funding for the National Science Foundation, the National Institute of Standards and Technology, and DOE’s Office of Science; more than $33 billion in authorized spending, during the fiscal years 2008–10, by government agencies on science, technology, engineering, and mathematics; and more support for teacher training programs in those areas.