Coordinated Research Programs
When spider webs unite, they can tie up a lion.
— Ethiopian Proverb
The Challenge
In many cases, ambitious R&D problems are not well-suited to individual academic labs, startups, or other existing institutions.
Problems can fall outside the scope of an academic lab for a variety of reasons. Academic labs tend to prioritize specific types of novelty, have modest resources, and focus on specific research domains. In practice, this means that a problem can be too engineering-heavy or require too much coordination between multiple disciplines to be a good fit for an individual academic lab.
While startups can be a natural entity to tackle many cutting-edge problems that fall outside the scope of academic labs, this is only in certain cases. Venture capitalists tend to fund startups to work on technologies that not only address problems with large market caps (~$1 billion+), but which can capture a large portion of that market in a reasonable amount of time (~10 years). In practice, this means startups are generally geared toward solving problems with large commercial markets — and most of the scientific risk often needs to be resolved before fundraising.
Many problems that are wildly important to society are ill-suited to individual academic labs, are not venture-scale, or are laden with scientific risk. Coordinated Research Programs (CRPs) are a natural approach to tackling these problems.
The Play
Coordinated Research Programs (CRPs) are suited to tackling large, cross-disciplinary problems that are not in line with the typical work of a single field’s academic labs. Often, CRP-shaped problems are larger in scope than what an individual academic lab can pursue. ARPA-style programs are one classic example of a CRP. Focused Research Organizations (FROs) — time-bound R&D organizations that assemble startup-style teams to pursue an ambitious technical goal, such as developing improved implants to measure and modify brain activity — are another example. Virtual Institutes, which coordinate distributed R&D teams towards a shared technical vision with multiple leaders (one for each team), are a third example.
Fueled by an ambitious technical goal and active management, CRPs enable funders to allocate funds in ways that are super-additive — compared to funding a smattering of labs separately working on problems related to your goal. In early ARPA history, the CRP approach helped deliver world-changing technologies like the internet and early autonomous vehicles. Today, it is seen as a natural approach to tackling problems like mapping an entire mammalian brain or developing blood substitutes.
The timing must be right to undertake a CRP. If you have a clear goal in mind, the time is right for a CRP when the collective knowledge in relevant technical areas has progressed to the point where a coordinated team of engineers, researchers, and others can make far more progress working as a team than as a set of disjoint labs with their own budgets. This coordinated team working in lockstep to pursue the problem can be housed under the same roof (as with FROs) or in separate organizations contracted to work together towards a shared goal (as with ARPA programs).
It is important that a funder thoroughly assesses whether the time is right before diving into a CRP. If it is an opportune time, a CRP can be an ideal use of funds. But the success of CRP efforts is often built upon the knowledge generated by prior research. If a funder is dedicated to some technical problem and not enough is known about the problem to undertake a CRP, then a more traditional research effort, such as those undertaken at universities, might be more appropriate. If the CRP approach does fit a problem, a medium-sized set of funds can go a long way. CRP efforts in the low eight-figure range have changed the world on more than one occasion.
From a funder’s perspective, there are several differences in the CRP funding playbook when compared to funding university labs. These include:
More Active Management. CRPs, whether they are housed under one roof or several, are actively managed. Whether they are managed by a CEO appointed by the funder or an ARPA-style PM who helps direct the efforts of multiple research teams, these are the individuals chosen by the funder who are charged with setting the vision for the program, finding the right staff, and making adjustments to the program as the work progresses.
A Specific Applied Goal. CRPs generally seek to build something specific — which serves as the North Star of their program. This can be some scientific instrument, a continuous stream of new datasets for a target group of researchers, developing methods to deploy in some sub-area of research, etc. The program’s main goal will help dictate the project’s management, staffing, pivots, etc. in the same way the profit motive does in a startup.
While the specific steps a funder takes to pursue a CRP will vary, some general steps will usually apply. The all-important first step will be to select a small number (one, in many cases) of ARPA PM-style individuals well-suited to scouting a program area — even if they do not end up managing the program themselves. They will then take the lead in undertaking steps like the following set— which were originally put together in a longer form as part of the SpecTech Brains Accelerator by Joshua Elliott and Ben Reinhardt:
Exploration and Field Mapping: Identify what knowledge exists in the field and which people and organizations are working on similar projects. Understand gaps and limitations, and understand the field enough to identify who is excelling and if the program’s vision is feasible. At this point, funders should begin to decide if they want to commit to this program or look for a new program area.
Refining Ideas: Produce a clear thesis of what your program hopes to do and the difference it can make, if successful. In working towards this thesis, it is generally necessary to learn about and articulate what the current state-of-the-art is in the field and why this is an opportune time to pursue this program. This phase might include discussions with industry experts and running workshops. The PM or CEO chosen to lead the program should be chosen by around this point, if that was not done before exploration and field mapping began.
Program Design: Create the program’s structure by considering what work needs to be done, if any new groups should be funded to carry out key pieces of the project (or the entire project), how separate groups might interact, what milestones and phasing are best, and what needs to happen at the end of the program so that it can have its intended impact.
Program Engineering/Execution: Clearly outline who is doing what, which might include soliciting proposals, negotiating contracts, and other administrative tasks. When the project is underway, the program’s leader must facilitate collaborations, make changes where necessary, etc.
Tech Transition: While tech transition technically happens at the end of a program, it should be considered from the program’s earliest planning stages. Given the applied nature of CRPs, the ethos of transitioning the technology successfully is all-important, not an afterthought. Those managing CRPs should identify which individuals, companies, or organizations will continue to utilize, develop, or scale the program’s resulting technology. Those groups should be thought of when making key program decisions and included in necessary conversations throughout program planning and execution.
Each program is different and will require its own set of steps. But the general list above paints a rough picture of how programs usually progress.
When it comes to CRPs, as in business, the doctrine of comparative advantage is vital. CRPs should be spun up only if a program area truly falls outside the abilities or incentive structure of traditional academic labs, startups, and other existing institutions. But if a problem is well-suited to a CRP and the timing is right, the outcomes can be world-changing.
Case Studies
ARPA’s Early Autonomous Vehicle Work
By the mid-1980s, ARPA-funded computer vision researchers had developed promising vision algorithms. But these algorithms could not be taken “off-the-shelf” and used in many applications. ARPA had the practical goal of building autonomous vehicles. This was a project that required too much systems engineering for academic labs and was too cutting-edge for ARPA’s traditional private-sector R&D contractors. The ideal vehicle team would consist of top-tier vision researchers with an applied bent as well as research-minded electrical and mechanical engineers. So, ARPA began to fund a hybrid group at the CMU Robotics Institute. While ARPA continued to fund basic research on component technologies such as algorithms and faster chips, they made continued bets on a systems engineering-focused group at CMU which built and improved test bed vehicles that fueled the early technology improvements of the autonomous vehicle revolution. This most notably resulted in a 1995 vehicle with a neural net-powered steering system that drove across the country 98.5% autonomously.
Forest Neurotech
Forest Neurotech is an FRO started in October 2024 and is developing a minimally-invasive implant capable of measuring and modulating brain-wide activity. This technology has the potential to unlock understanding and treatment of the human brain on a personalized level, which would significantly impact those with neurological and psychiatric disorders. This is an area ripe for a CRP because the technology’s development requires both frontier academic knowledge and high-level engineering staff. Forest’s status as an FRO, a kind of CRP, enables the organization to build and test its technology across a variety of implementation areas. Given the major technical risks involved in the work, if a VC were to fund the effort it may only be for Forest to pursue a single, narrow use case that was considered lucrative.
Carbon to Sea Initiative
Carbon to Sea was launched in June 2023 and seeks to harness the natural process of Ocean Alkalinity Enhancement (OAE) as a solution to carbon pollution. The Carbon to Sea team funds several separate R&D projects looking to better understand and scale approaches for OAE to remove C02 from the atmosphere. The technologies that the initiative seeks to develop are both on the scientific frontier and engineering-heavy — requiring toolkits in ocean modeling, engineering, physics, biology, project management, and more. Given the scientifically and practically ambitious nature of the goal, the initiative simultaneously funds academics undertaking studies that help address known bottlenecks and cost drivers as well as field research sites suited to carrying out aspects of the work related to implementation, testing, and scaling of the methods.
[Learn more]
Learning Engineering Virtual Institute (LEVI)
LEVI is a Virtual Institute. Virtual Institutes (VIs) are a type of CRP with a similar structure to an ARPA program. Like ARPA programs, VIs have clear R&D goals, are time-bound, and coordinate distributed groups of existing researchers toward a shared goal. Unlike ARPA programs, they have multiple program leads (one for each team). LEVI’s ambitious technical goal is to leverage AI tools to double the rate of middle school math progress. LEVI’s network of affiliate contractors are meant to work towards this goal with solutions that, by the end of the program's fifth year, can be scaled to over a million students internationally at a cost of <$1,000 per student. These R&D teams are focused on approaches like AI-powered chatbots, AI teachable agents (where the student teaches AI), and allocation of human and AI tutors based on student need. To complement the work of these R&D teams, LEVI also funds “hubs” that monitor and accelerate progress across all LEVI teams. For example, the Engineering Hub embeds part-time engineers with its research contractors for short-term product sprints, while the Scaling Hub provides operational support and connections to help teams expand their interventions into more school districts.
How we can help
The RenPhil team is dedicated to helping interested funders explore and execute CRPs in program areas of interest. This includes RenPhil assisting in the following ways:
Sourcing field leaders, program managers, and other experts. These individuals can determine the feasibility, upside, and implementation plans for new programs.
Training and Advising Program Managers. Many with the right set of mental tools to be excellent program managers of a CRP need a bit of training or assistance. This might simply take the form of helping PMs network, find the right labs, negotiate contracts, etc. It may also consist of RenPhil giving would-be PMs a crash course in what it means to be a PM, common strategies they can deploy, pitfalls to avoid, etc.
Assistance in Spinning Up Programs. Once a leader for a CRP is found and a funder green lights a project, the legal and operational steps to turn the idea into a real-world initiative with legal agreements, lab space, a recruiting strategy, etc. must be undertaken. RenPhil is capable of taking the lead on these efforts (the most hands-on option) or advising (a more passive option) on all of this work.
Resources
Research Leaders’ Playbook (Speculative Technologies)
ARPA Playbook (FreakTakes)
The ARPA Model: A Reading List (Institute for Progress)