Breakthrough innovations such as internet protocols, the transistor, cellular networks, lasers, photovoltaic cells, and the graphical user interface have fundamentally shaped the modern world. Each of these technologies were first developed by a cohort of renowned mid-20th century innovation programs– DARPA, Bell Labs, and Xerox PARC among them.

The field of economic development might learn from the history of these programs and the tradition of mission-driven science. One of the key lessons: build deliberate institutional architecture.

DARPA (Defense Advanced Research Projects Agency) and similar institutions operated with a clear mission and stable funding, insulated from short-term political and financial pressure. Each program was built around a clearly defined goal — specific enough to measure progress, yet flexible enough to allow diverse approaches. 

The current funding architecture tends to reward legibility over ambition, channeling resources toward interventions whose effects are measurable but modest while larger systemic challenges compound.

And each program anchored innovation around a specific type of leader such as the DARPA program manager and the Bell Labs systems engineer. These individuals were technically-trained generalists who mapped entire systems, identified where intervention was most promising, coordinated across disciplines, guided ideas from research to real-world use, and steered around dead-ends and other pitfalls. They were not administrators; they were active navigators, holding complex research and development missions together.

To be sure, the most pressing challenges in economic development — entrenched inequality, biodiversity loss, adaptation to climate change, and the housing crisis among them — are not simply technical problems. They are social and institutional ones, embedded in the structures, incentives, and power relations that govern how societies organize themselves.

As such, DARPA’s model cannot be replicated exactly. But building a comparable model — designed specifically for social and institutional innovation — is possible. The social and ecological crises of our time demand big and bold responses. Instead of targeting discrete goals and focusing on isolated issues, development programs need missions that can shift entire systems.

Barriers to an ARPA for Development

The barriers to building an ARPA for Development are both structural and organizational.

Structurally, the vast majority of economic development strategies funded today are those that can be measured — discrete, bounded, and comparable to an alternative outcome. The rise of randomized controlled trials (RCTs) and analogous evaluation practices in the development sector was a necessary corrective to the ideology-driven prescriptions of earlier decades. The current funding architecture tends to reward legibility over ambition, channeling resources toward interventions whose effects are measurable but modest while larger systemic challenges compound. This selects powerfully against directed exploration. You cannot run an RCT on a market that does not yet exist, or evaluate a financial instrument before it has been invented. 

Institutional structure compounds the problem. Development actors organize around isolated program areas— agriculture, financial inclusion, and gender. The result is a field that funds projects rather than systems, and mistakes a collection of projects for a coherent strategy.

Though far from being the norm, development has, in fact, demonstrated the capacity for nurturing breakthrough socio-technical innovation.

Donor control reinforces both challenges. Funders retain substantial authority over the strategies they support, and they manage their reputations by pushing consistently toward approaches that are tried, vetted, and unlikely to fail visibly. 

The same dynamic leads major science funders like the National Science Foundation and National Institutes of Health to underinvest in highly innovative research. Grant committees select for consensus and legibility, not for the riskier, unconventional ideas that actually lead to breakthroughs. 

Beyond structural barriers, few organizations are designed to hold complex innovation programs, nor do we have established playbooks for the work itself. How to define transformative development objectives, map bottlenecks, design learning and iteration processes, and sequence capital and activity over time has not been codified into a practice.

Finally, there is no clearly defined role for the most important actor: the systems engineer equivalent — the individual capable of reading an entire social-ecological system, identifying where intervention is ripe, coordinating across sectors and disciplines, and navigating toward an end-state that cannot be fully specified in advance. 

To the extent that such individuals, roles, and capabilities exist across the field, they are chronically under-resourced, poorly recognized, and unsupported by shared professional infrastructure.

How to Build Institutions for High-Risk, High-Leverage Work

Other fields have confronted analogous problems directly and responded by building tailor-made institutions for high-risk, high-leverage work. 

Venture capital built an ecosystem for breakthrough commercial innovation, with deal structures, risk norms, and a professional culture organized around non-consensus bets. 

In science, Warren Weaver’s Natural Sciences Division at the Rockefeller Foundation helped establish the field of molecular biology through an opinionated, decade-long funding agenda that no peer-review committee would have approved. 

More recently, a new generation of mission-driven agencies — the UK’s ARIA, Germany’s SPRIND, and Focused Research Organizations in the United States — have been established on an explicit premise: that some of the most consequential scientific challenges are too bold, too cross-disciplinary, and too long-horizon for normal science, and too early or too uncertain for private capital. Each represents a deliberate institutional response to that gap, and provides models for what is possible in the development sector.

Development Can Innovate Too

Though far from being the norm, development has, in fact, demonstrated the capacity for nurturing breakthrough socio-technical innovation. And the conditions that produce it bear a closer resemblance to mission-driven science than is commonly recognized. 

M-Pesa — the mobile money platform that fundamentally reshaped the East African economy and drove significant reductions in poverty — emerged from a public-private R&D partnership between the UK’s Department for International Development (DFID) and Vodafone, built around several features that closely mirror the institutional logic of mission-driven science. 

DFID’s Financial Deepening Challenge Fund provided matched funding — £1 million from each party — designed to protect longer-term, non-core projects from internal pressure to prioritize quicker, safer returns. Critically, the program was organized not around a specific product but around a problem: improving access to financial services for unbanked populations. 

The team had, in the words of former M-PESA lead Nick Hughes, “no roadmap, but created solutions as we went.” When users during the pilot began repurposing the loan-repayment platform to transfer money directly to friends and family — behavior its designers had not anticipated — the problem-oriented mandate gave the team the freedom to follow the evidence and pivot. A dedicated project lead, embedded on the ground in Kenya, provided the connective tissue that held the effort together. 

DARPA, Bell Labs, and others were designed to ensure that breakthrough technical innovation was not left to chance. That same determination is needed to meet the large-scale social and environmental challenges of today.

Beyond M-Pesa, there are a wide range of other impactful socio-technical innovations whose R&D trajectories we can learn from. 

For example, Blue Forest’s forest resilience bonds create new capabilities to finance wildfire management and forest conservation. Enduring Earth’s project finance vehicles create the protected governance models that long-horizon landscape restoration programs depend on. TransCap’s financial backbones coordinate diverse capital types around a shared purpose. Integral’s collectively owned strategies and RIS3CAT’s shared agendas construct common problem definitions across fragmented actor networks. 

These are novel financial, institutional, and social technologies that unlock new capabilities for social innovation. However, their development has depended on visionary founders, fortuitous donor relationships, and the particular conditions that made each one possible. They are exceptions that prove the rule. 

An ARPA for Development will need to build the architecture that turns these exceptions into a replicable process. Certainly, building this architecture is a field-level challenge. It will require philanthropies, multilateral institutions, and governments to invest deliberately in the organizational forms, the playbooks, and the human capital capable of incubating and steering breakthrough innovation for complex development challenges. 

Renowned institutions like DARPA, Bell Labs, and others were designed to ensure that breakthrough technical innovation was not left to chance. That same determination is needed to meet the large-scale social and environmental challenges of today.