Skip to main content
Blog Post

Reimagining Federal Education R&D: DARPA for Education

AEIdeas

March 31, 2025

Earlier in my professional life, I was the head of the political science department at Stony Brook University. When the chairs of the arts and science met, someone from the physical or life sciences would inevitably argue that the “hard” natural sciences deserved more support than the “soft” social sciences. My rejoinder was one of my favorite quotes: “God has chosen to give the easy problems to physicists.” I always (and wrongly) attributed this to Richard Feynman (an unimpeachable source). I was never questioned about the quote’s origin, so I kept whipping it out to support my claim that political science was a much harder science than physics. 

If political science was hard, education research focused on finding practices that actually work to improve student achievement is harder.  

I once estimated that only about 15 percent of the math interventions the Institute of Education Science (IES) supported over many years were backed by evidence that they worked. This was depressing—until I started looking at the success rate of other fields. Only about ten percent of clinical trials “work” and the replication crisis that has befallen many sciences suggests that even low estimates of success may be too high.  

I’m not sure if this lack of success was part of the reason that IES faced such severe cuts (around 90 percent of the IES workforce was let go—far higher than most other agencies). I’m also not aware of any administration plans to rebuild IES or move its various centers to other government agencies.  

A few months ago, I suggested moving the National Center for Education Research (NCER) to the National Science Foundation’s STEM education directorate (EDU). EDU is not a strong unit. It invested heavily in DEI; it has not consistently pursued rigorous research; and it does not have a track record of focusing on what works for whom under what conditions (IES’s guiding philosophy).  

I thought that merging the units would introduce IES’s “DNA” into EDU, improving  
EDU while providing a home for the education researchers at NCER. But with only one person left in NCER, there simply won’t be anybody with the right experiences and standing to improve EDU.  

What happens next to NCER is unclear. If we assume that there will be at least some continued government support for education research, I want to lay out a strategy for a new NCER, if rebuilding starts. 

First, it’s clear that the “business model” of NCER has run its course. I describe that model as “three Fs:” five years, five million dollars, and failure (the length of time of its grants, the amount of money usually given out, and the typical outcome). NCER employed a “shotgun” approach to funding research in which last year saw over 40 individual research projects in 12 different topic areas funded to the tune of $100 million. This approach did fund some high-quality research—but far too much of it went to areas driven by niche academic interests and not by the education needs of the nation as a whole. NCER was also late to realize that scaling any successful intervention is essential to improving student achievement in a meaningful way.  

One way to break out of NCER’s stale approach is to adopt the practices pioneered by DARPA, a division of the Defense Department that focuses on breakthrough national-security technologies, but which has also advanced many civilian breakthroughs like the internet, GPS, advanced microprocessors, and Covid-19 vaccines. During the Biden administration, “DARPA envy” was widespread, and ARPA-like units were set up in several agencies, including the Departments of EnergyTransportation, and Health. All aimed to use modern research methods for mission-critical breakthroughs in their respective policy domains. Over the years, people have proposed an ARPA-ED for education, and during the last few years, there was a concerted push to create a new DARPA-like center, the National Center for Advanced Development in Education (NCADE), within IES. Legislation was introduced in Congress but never enacted.  

I have argued for NCADE before. But despite the efforts of many to create the new center, the heavy lift of getting authorization through Congress never came to fruition. In turn, attempts to create more innovative DARPA-like practices within IES were constrained by the habits and practices of NCER and by the 20 or so Center program officers, most of whom were long-term government workers committed to business as usual.  

For better and worse, DOGE cleared out NCER, leaving a green field upon which to rebuild an education research center that employs nimbler and more modern approaches to education research and development. There is now an opportunity to build ARPA-ED within existing legislation without asking Congress to create NCADE.  

This is an opportunity that must be seized.