It's a DARPA-filled Week
A big week if you’re into DARPA, or rather, ARPAs of all kinds.
First, Issues in Science and Technology published several major pieces on the ARPA model of funding research—and taking quite varied points of view. Fascinating stuff:
“No, We Don’t Need Another ARPA,” by John Paschkewitz and Dan Patt.
“Facing economic and technological challenges from China, the United States has reacted by launching ‘ARPAs for everything,’ modeled on the Defense Advanced Research Projects Agency (DARPA). Yet the goal-oriented research at the heart of an ‘ARPA approach’ is not a panacea for all that ails US innovation, argue two former DARPA program managers. Drawing from their interactions with the commercial sector and broader research community, they make the case that the United States has lost sight of its true strategic advantage: its ability to adopt and spread innovations, or ‘diffusion capacity.’”
“Building a Culture of Risk-Taking,” by Jennifer E. Gerbi.
“The job of a risk-taking organization such as an Advanced Research Projects Agency (ARPA) is to break eggs in order to make the proverbial omelet—which can be terrifying in a government culture where many people’s jobs are devoted to keeping eggs whole and in the carton. The former acting director of the ARPA for energy talks about how she inculcated a risk-taking culture in the agency.”
“How I Learned to Stop Worrying and Love Intelligible Failure,” by my good friend Adam Russell.
“With growing worries that the return on investment in science and research is falling, there’s an ever louder call for more organizations modeled on the Advanced Research Projects Agency (ARPA). But how can any ARPA know whether its processes are effective, or even whether the assumptions behind them are valid?”
Russell coins several clever new terms to think about in judging the success of an ARPA, but more importantly, he focuses on the main metascience question itself (one that ARPA-H unfortunately seems to have sidelined): How do we figure out what works?
Second, Good Science Project fellow Eric Gilliam has embarked on an ambitious initiative to do a deep dive on dozens of DARPA projects, both successes and failures. His first installment—analyzing DARPA’s involvement with parallel processing computers in the 1960s and onward—just came out today. It’s as thorough and thoughtful as you would expect if you’re familiar with Eric’s writings, and you can find it at the Good Science Project website, or at Eric’s Substack. The overarching goal in this series is to produce a larger “ARPA Playbook” that will hopefully be useful to funders in ARPA-like organizations.