Perhaps most surprising are the numbers they come up with on fossil fuel and nuclear power subsidies by the US federal government - a total of $500 billion over the past 50 years, and $35 billion for the future as proposed in the 2004 version of the energy bill. These seem to be solid numbers, coming from a publication "Federal Incentives for the Energy Industries. 1998." by "Management Information Services". Past R&D spending on fossil and nuclear is quoted at $100 billion, compared to only $25 billion on efficiency and renewables.
More interesting, but perhaps more debatable, are the economic analyses in the report. Their modeling finds that just imposing a "20% by 2020" standard would lead to job losses compared to the energy policy currently before congress; however by diverting the $35 billion in future incentives for coal and nuclear power instead to renewables and energy efficiency, they find hundreds of thousands of jobs created and tens of billions of dollars in annual economic benefit.
Oddly the "20% by 2020" standard that loses jobs provides even more economic benefit to GDP in their models - suggesting the models may not be well-enough refined to be terribly meaningful. Nevertheless, this sort of specific policy proposal is very useful as a reference point. We can do better, I believe, but it's not a bad start.