Grants Only Go So Far

I’ll be blunt: I’m worried that because of how easy and humble grants are, enabling research through grants will crowd out enabling research through more definite ways.

The optimism embodied in grants and the fact that they are effective puts me in the position of doing a tightrope walk over a pit of moral hazard where, on the margin, I simply justify more inaction or pessimism. Instead I want to convince you that:

1. Grants are good, but they have limits and many good things fall outside of them.
2. The mindset that “research funding” is all that research enabling organizations do is wrong.
3. There are other pragmatic ways to enable research.

The implicit way that many people think of enabling research¹ (especially outside of your own organization) is as a matter of how you get money to people and filter potential projects. Call this the “filtering” or “granting” approach. Funding and filtering are powerful aspects of research management, but they’re just two tools in a much larger toolkit. Overfocusing on grants limits you to enabling a particular set of activities. In other words, grants only go so far.

Grants are great

The idea of grants (especially small grants) as the way to enable more awesome in the world feels like it has exploded in the “tech-adjacent/crypto” world — Emergent Ventures, Fast Grants, Impetus Grants, Gitcoin, Moth Minds, and more. Grants are also a default mode of both government and traditional philanthropic funding as well. Let’s be clear: these programs and institutions are a massive boon for the world. Both new grant programs and older granting institutions have² enabled impactful work that would otherwise not have happened, directed money towards human flourishing, and generally channeled optimism and the desire to enable awesome towards productive ends.

The granting approach has many positive attributes to recommend it. In their purest form, grants have extremely low transactions costs. Someone says “I want to do this” and the grantor says “Ok. Here’s the money to do that.” Put a bookmark here: this core interaction is also where the limits of the granting approach sneak in.

Grants don’t require a huge time commitment or deep expertise on the part of the grantor (it’s pretty straightforward to get an experts take on any particular grant). Granting doesn’t require the person giving out the money to come up with good ideas. In that way it’s very humble. Grants let you build a portfolio of support that, if you believe you don’t have the best ideas, maximizes the chance that you’ll support successful projects. It’s a very indefinite optimistic way of approaching research management.

In their platonic form, grants don’t require any ongoing work — you give someone money and let them rip. In part, we’re seeing so many granting programs because of how they can be done well without a massive time commitment. Most of the most lauded new examples are run as side-projects by extremely busy people. Grants are a tool for some of the most effective people in the world enable research despite many other commitments. In the absence of grants, they might not help at all!

Grants are humble, fast, low time commitment, and maximize the chance that you’ll support something successful — their appeal and popularity makes a lot of sense. But they have limits.

Grants have limits

Grants filter for people who know what the right thing to work on. However, not everybody who could potentially do good work knows the right thing to work on. Especially in a world of increasing specialization people often see problems only through a specific lens or only see one node in a system-level problem. More generally, people don’t know what they don’t know.

Small grants aren’t just miniature versions of big grants — they’re cats and lions, related but entirely different beasts. (What entails “small” and “big” obviously varies, but generally breaks down on whether the grantor feels like the amount is “significant” or not.)³ A small grant can generally buy a piece of equipment or enable a one full-time person project. That sort of work is very different than the multi-person, capital-heavy work that needs a large grant. In addition to enabling very different flavors of work, small and large grants require drastically different amounts of trust from the grantor. This difference is (I suspect) is why so many new, fast, low-reporting-style grants are small and have an upper size limit. Ideally, people could use a small grant to do trust-generating work that they could use to get larger grants. However, the difference in kind between the types of work can make this difficult. “How can I trust that you can coordinate a team when haven’t done it before because you haven’t had the resources to do it?” As a result you see many projects start with low-overhead small grants but either putter along forever or die because the criteria for larger grants are so much more stringent.

Most significantly, grants fail to address one of the biggest impediments to weird new things in the world: the shadow of the future.⁴ You get a grant that pays the rent for a few months, perhaps hire a contractor, buy some equipment, or do work that wouldn’t normally be funded in your university. Then what? You need to either land in an institution or get another grant. Either way, you need to tune work that you did on the grant to set you up for your next steps. You haven’t actually escaped institutional incentives at all. This phenomenon constantly looms over the shoulders of those of us who want to see new institutions and games in the world and deserves a much longer treatment.

Grants could arguably get around the shadow of the future by enabling side projects, but that scale is a significant limitation in itself — for many projects part-time work is different in kind from full-time work, especially with multiple people involved. Additionally, I suspect that many of the people who grants are going to are not being paid by the hour. The idea that a chunk of money can free someone up to work on a project part-time is often a fallacy.

Research management matters

Research management matters. At some point, unless a single person can do the work for a project, someone needs to play a coordinating function. Coordination involves incentivizing people to do things they wouldn’t otherwise do. The actions can be at levels of abstraction and the incentives can involve various levels of coercion, but at the end of the day, effective coordination involves more than just shining a spotlight on problems and potential solutions. Vision alone is not enough. In recent memory, Mark Adreesen’s It’s Time to Build and Patrick Collison+Tyler Cowen’s We Need a New Science of Progress were both inspiring pieces and wasted opportunities. Yes, they galvanized people, but a lot of the energy was diffused without being channeled into change. To a large extent people framed what they were already doing in a new language instead of changing trajectories. Analogously, Vannevar Bush got people excited about the idea of a Memex in 1945 with As We May Think, but it took Licklider, Sutherland, and Taylor deploying capital, nudging+enabling people to start labs, and coordinating work on different pieces of the connected personal computing vision decades later for the vision to become reality. Good research management has many dimensions and can manifest in many different ways depending on the context: from Elon Musk’s involvement in minute engineering details to J.C.R. Licklider’s funding labs and keeping communication channels open between them based on a fairly precise vision.⁵

The importance of research management is probably a “duh” point when spelled out, but if you agree that research management matters, it means that the common perception that “research enablement = research funding” is wrong. To focus only on funding is to implicitly say that filtering-based research enablement is the only option.

The reason DARPA works is not because they are exceptionally good at picking projects to fund, it’s because they use that money to incentivize projects that otherwise would not be considered and coordination between them. It’s not a funding organization, it’s a research coordination organization.

Pure filtering-based research enablement always pushes the coordination function one level lower. Grants defer on having an opinion about what should happen beyond a binary yes/no, saying “well, the person or org receiving the grant will do any necessary coordination.” The trick is that usually the recipients of a grant will have less money, social capital, and reach than a grantor, all of which makes coordination harder. Pushing coordination one level lower also subjects the coordination to the incentives of whatever institutional structure the grant is going to — be it a startup, an academic lab, or an individual out to make a name for themselves. Institutions shape how individuals interact.

In the extreme, if grants are the lauded way to enable things, everybody will end up giving everybody else grants hoping for some coordination to emerge on some distant Discord server. Grants all the way down! Too much focus on grants leads to everybody wandering through the world playing mother-may-I.

Alternatives to grants

It’s easy to point out deficiencies and hard to propose concrete solutions. I’m going to list some potential alternatives to avoid the moral hazard that, on the margin, calling out the limitations of grants might just decrease the money and effort going into them without directing it somewhere productive. You’ll note that some of these are tweaks on grants. Remember, I’m not saying grants are bad, there are just many other things to explore. I beg forgiveness for how sketchy these alternatives are — I plan to make them much more structural later. If any are particularly interesting, let me know.

• Patronage/Fund people not projects. Instead of a one-off grant focused on an objective, pledge to support someone for an indefinite time. You don’t need to be absurdly wealthy to do this — you can pool money with a number of other folks: it takes less than a thousand people each giving $10 a month to give someone an incredibly reasonable stipend. You can go one further and be a “lead patron” giving more money and publicly rallying people to give. A long-term source of funding can start to abolish some shadows of the future. Yes, this is nominally what the Patreon platform is for, but it has a number of limitations — the biggest being that it requires people to hit at least some level of local fame, filtering for people who do flashy work and are willing to ask for money in public. Do the work to find people you think are doing great work and ask what level of funding would make a discrete difference in their ability to pursue their research.

• Give fewer bigger grants, not more small grants. Big grants can be very different from small grants (see above) — they enable entire classes of activity (hiring people, buying high-quality equipment, actually quitting a job or education). It’s true that smaller grants can be incredibly transformative, require less trust, and allow you to build a more diversified portfolio. But if the majority of grants are small, we’re going to see primarily work that can survive or get started on small grants, slicing off a whole chunk of potential work.

• Give money to incentives/Fund organizations, not projects. Instead of seeing grants as a mechanism for filtering specific projects or outcomes, you can think of them as delegating to organizations that encode a process, vision, or theory of research management. This approach is the one-level-higher version of “fund people not projects”. Because institutional structures enable or constrain different activities, supporting the existence of an institution can lead to good work that doesn’t necessarily fit into a grant-shaped bucket. At their core, new research organizations like Convergent Research, Actuate, PARPA or Dynamicland are attempting to shift the incentives around research and enable sets of activities that wouldn’t otherwise happen.⁶

• Create soft landing sites. If you are in a position to do so, you can give talented people an indefinite open offer of a role they would want should they ever decide they want it. An option on a job, if you will. This move addresses shadows of the future by creating a ‘soft landing site’ — the knowledge that their career isn’t completely destroyed if they try something irresponsible for a long time and it crashes and burns. Yes, this creates some risk for you: it’s another way of putting your ‘money’ where your mouth is. The world needs more people willing to stick their necks out for awesome without expectation of direct upside.

• Front status. Many times, status can far more valuable to a project than money. People and organizations lend each other status all the time, though endorsements, introductions, titles, etc. Typically, high status people/orgs want to see a significant chunk of evidence that their status won’t be lost (and in fact will probably get good ROI on it) before they’re willing to lend their status to a person or organization. If you are a high-status individual or org, you can buck the trend and front people status. If you work for a high-status organization, give someone you want to support a title like “fellow” even if they aren’t on payroll — it’s shocking how many doors it can open. Signal boost things you aren’t yet convinced you want to be involved in. Tell your friends about it. Make intros.

• Do tranched funding. Understandably, most people don’t want to give a huge chunk of money to an unproven idea and team. However, the difference between vague noises of support with the potential of money in the future and an extremely clear set of conditions under which a specific amount of money would be available is huge. It not only provides a goal to work for but enables someone to turn around and rally support by saying “if I can just get this much to get to milestone X, I will unlock Y” — it effectively lets them project some fraction of that future money into the present without lying. Many times, a large chunk of clearly-conditioned money in the future can be more valuable than a small chunk of money now. Like so many of these suggestions this one does require work on your part and reduces your optionality. Let’s celebrate people doing work and reducing their optionality to enable awesome things!

• Offer non-monetary support. For someone doing research or running a research organization, time really is money. If you don’t have the money to give significant grants and don’t want to pool your money with others but have valuable skills — legal, accounting, design, engineering, literature coming, internet searching — volunteering them can be far more valuable than a grant. Not only does the person not have to pay someone to do them or spend their own time doing them but they also save the time looking for someone or . The trick here is that you need to lower the transactions costs as much as possible — don’t say “I want to help, use me.” Instead, proactively say “I’ll help you file your taxes” “design you a logo” or “build you a system that does this thing I suspect you need.”

• Proactively find people doing good work and fund them/nudge them to do something different. Grant applications are a demoralizing pain, even if you make your application straightforward. If you’re working on something weird, you face an exhausting stream of people asking you to make it legible and justify what you’re doing, trying to predict which framing will work on whom. You also need to know about a grant in order to apply for it. Instead of putting the work on other people, if you want to enable more awesome things, you can proactively go out and find people/organizations who are trying to do what you want to see more of in the world. This mode takes more work and doesn’t have the same satisfying counterfactual impact as grants (“I enabled this thing that wouldn’t have otherwise existed!”) but can be just as important. You can go one farther and find people who you think could be doing great things but are not and nudge them with money towards the things you want to see more of in the world. Do this with multiple people! Get them to work with each other! Coordination!

• Proactively match other allocators. If you see someone who you think has good taste either supporting or offering to support something, publicly +1 them. This move enables researchers to leverage their fundraising time but more importantly, it could solve the coordination problem where nobody is willing to be the first mover. Matching does happen to some extent but it is usually quiet and the onus is usually on the person raising money to negotiate it. Do we want to filter for researchers who are also master politicians?

• Give people money to work on someone else’s project. Grants tend to encourage a fractal scattering of projects, with everybody working on their own thing. Often, someone can have more impact by joining an existing weird project. This kind of grant exists in the traditional academic system in things like the NSF Graduate Research Program where funding is attached to a grad student, instead of a lab or project. This move also addresses the coordination problem by making it easier for projects to scale without raising additional funding themselves.

The common themes here boil down to proactivity, supporters taking on more risk, being more opinionated about what is good but less opinionated about what is going to succeed, giving up filtering specificity, and planning. Basically the inverse of the limitations of grants.

Conclusion

1. Grants are good, but they have limits — they filter for specific kinds of people and projects, force ambitious projects onto a specific trajectory, encourage portfolio theory and success bias, and continually push responsibility down a level while maintaining control.

2. The mindset that “research funding” is all that research enabling organizations do is wrong. Research management matters.

3. There are many potential alternatives and tweaks that can enable work that falls outside the limits of traditional grants. Nothing is a silver bullet, but on the margin, different approaches to giving money to support research can enable entirely different sets of activities and the human flourishing (and awesome sci-fi shit) that so often comes from trying weird new things.

Acknowledgements

Thanks to Andy Matuschak, Molly Mielke, José Luis Ricón, Michael Nielsen, and Kanjun Qiu for reviewing this essay and comments.

Thank you to all of you who are doing grants and anything to enable good research, frontier opening, and a flourishing future — I know this essay is critical in places.