The answer to “where is the control over direction and goals?” lies on a spectrum between performers - the people actually doing the building - and directors - the people who run the innovation organization, be it a VC firm, or DARPA.
On the performer control end of the spectrum you have have organizations like Howard Hughes Medical Institute (HHMI) that basically say “ok performers - here’s a bunch of money - go for it!”
On the other end of spectrum, Director control looks like NASA’s manned space program where the President has direct input into its direction. (Which is why NASA comes up with a 10 year plan every 8 years.)
Both ends of the spectrum are good for certain situations. Performer control seems good for true exploratory research where you just need someone with good intuition to play around. Director control seems good for something like the Apollo program, where you need a massive amount of resources and alignment from different orgs (congress, etc.) When is push or pull appropriate for innovation orgs?.
DARPA does multiple levels of top-down problem generation and bottom-up solution generation PMs exert this control both by having only one official checkpoint before launching the programs and their ability to move money around quickly. It is relatively easy for DARPA PMs to re-deploy funding while it’s hard for someone to pull funding from the program manager once they’ve passed through the tech council. (Every program at DARPA is intensely technically scrutinized by the tech council.)
They do get informal feedback during the program design process. What role does informal feedback play?.
In addition to concentrating control, program managers also concentrate risk. In the Anna Goldstein Podcast she points out that the risk is accepted at the program level and then the grants that the program manager gives out should be relatively likely to succeed. In practice, it means that programs have a high failure rate, but any given project within a program is likely to succeed. DARPA has a 5-10 percent program success rate
DARPA PMs use seedling programs to ‘acid test’ the riskiest pieces of a program idea. So going into the main program, they should be pretty confident that what they’re paying performers to do is possible. (The DARPA execution framework boils down to showing that thing is not impossible, showing that thing is possible, and then making that thing possible.) So at the end of the day, the risk is assumed when the tech council approves a program (Every program at DARPA is intensely technically scrutinized by the tech council), not when the program manager gives out a grant.
How well does this process work? An Empirical Description of Risk Management in Public Research Funding says that 10% of ARPA-E projects are terminated early. Each ‘project’ is equivalent to a single grant so in comparison to (<NSF NUMBER>)
The ARPA approach to risk is also different from a VC portfolio, where the risk is at the level of each individual company and the VCs create a portfolio.
Concentrating power in program managers can potentially lead to moral hazards as blatant as embezzlement (in the case of William Godel) or more subtle, like funneling money to your buddies. (The ARPA Model is an unbuffered system)