Photo cred: my dad.
I’ve won grants, I’ve awarded them, and I’ve advised people applying for them. Everything I’ve seen has convinced me this is a terrible way to give away money.
Some of the problems with this system are well known, but we treat them as unavoidable. Other problems are totally overlooked. The result is that virtually all grant money––whether it’s going to fund a lab, sponsor a fellowship, or support a nonprofit––is spent inefficiently and often goes to the wrong people, all while wasting everyone’s time.
Here are five of the worst problems with the way we do grants now. I list them because I believe there’s a better way, and we can only see it when we look squarely at the current system’s shortcomings.
1. Applications are hackable
Grant applications don’t measure who is best for the grant, but who is best at applying for the grant, running headlong into Goodhart’s Law. This rewards people who are good at sopping up credit for projects, marketing themselves, charming interviewers, and convincing fancy people to say nice things about them. Some good people are good at these things, but unfortunately so are many sociopaths.
Even well-meaning people think they’re doing a good thing when they hack grant applications, which I know because a bunch of well-meaning people helped me hack my way to a Rhodes Scholarship. I only applied because I went to Princeton, where students with good grades get connected automatically to a professional advisor who helps you craft your application, sharpen your interview skills, and defuse your freakouts. I got to read detailed reports from previous students who had interviewed in my district. Several Princeton-affiliated former Rhodes Scholars mentored me, one of whom told me, “By the way, they might ask you how you would ‘fight the world’s fight,’ so make sure you have an answer for that.” Sure enough, they asked me that exact question in my interview.
When I returned from Oxford and started a PhD, I became one of those well-meaning grant hackers in exchange for free room and board. The colleagues I joined were world-class hackers themselves: my dorm, which houses ~10% of Harvard seniors, produced 32% of Harvard’s Rhodes and Marshall Scholars over a ten-year period. My predecessors advised six successful Rhodes Scholars in 2013 alone, after which the Rhodes Trust barred applicants from getting help on their personal statements. I’m sure everybody abides by this vague and unenforceable rule.
2. Applications rely on bad proxies for talent
Even if nobody tried to hack grant applications, they’d still be a pretty awful way to pick people.
Which of these is the better way to get to know you?
A) You send me a personal statement, resumé, transcript, and three letters of recommendation. I read these, then I interview you for 20 minutes.
B) I hang out with you for an afternoon.
I’d pick B and I bet you would too, because application materials miss a lot of important stuff. That C+ in organic chemistry doesn’t come with a note that says, “By the way, my mom was diagnosed with cancer right before the final exam.” Your letters may say nice things about you, but the person who supervised your internship may not have appreciated what it felt like to be the only person of color in the office. And interviews are nearly useless for predicting later performance. Applications would be a ridiculous way to pick friends, yet we treat them like a sensible way to award billions of dollars in grants.
For instance, most Rhodes selection committees include a cocktail party as part of their interview process. This is a pretty bad way of judging whether someone is a good person, but it’s a pretty good way of judging whether they are pleasant to talk to at a cocktail party, and so Rhodes Scholars are often charming conversationalists and sometimes bad people (see: Bill Clinton, Bobby Jindal, noted anti-vaxxer Naomi Wolf).
When I interviewed, one of the committee members at the cocktail party asked me to tell a joke. (My CV and personal statement had comedy all over them, so this was not as confrontational as it seemed.)
“Give me any noun and I can give you a pun in ten seconds,” I said. He said “porcupine” and I said, “A great name for a barbecue restaurant: the Pork-U-Pine For.”
The next day, during my actual interview, the same guy asked me my final question. “You did it for me yesterday, now do it for all of us,” he said. “Your word is ‘lasagna.’”
“Okay,” I said, “a guy was struggling to figure out how to make lasagna, and his friend came into the kitchen and said, ‘Dude, just use your noodle.’” (Look, it was ten seconds under pressure, what do you want.) A few hours later they made me a Rhodes Scholar.
As cool and fun as it is to ask 22-year-olds to tell you jokes, maybe it’s not the best way to figure out who should get an all-expenses-paid master’s degree from Oxford.
3. Grants are ineffective because they reward people who are already succeeding
If you’re giving someone a grant, you want to enable them to do something they wouldn’t have been able to do otherwise. But most grants go to people who are already successful––people with good grades, degrees from prestigious institutions, leadership positions, media buzz, previous grants, etc.––and rewarding them makes no difference.
For example, the Rhodes Trust probably hopes that by picking the most accomplished college seniors and giving them a super prestigious prize, they will encourage the youngsters to do lots of brave and risky things. Instead, the most popular destinations for my Rhodes cohort were top-tier medical schools, law schools, and PhD programs (guilty), as well as a handful of consulting companies––exactly where we would have gone if we hadn’t gotten the scholarship.
A similar thing happens in science. For instance, the scientists who just barely win a postdoctoral grant publish as many papers afterward as the scientists who just barely miss it, but the winners are more likely to win a second grant. This is partly driven by the losers choosing not to apply for as many grants later, so grants don’t just encourage people who are already winning––they also discourage people who almost won.
4. Funders undermine the people they fund by telling them what to do
Giving someone a grant means you think they could do something better with the money than you could. That’s a great vote of confidence! Unfortunately, many funders undermine that confidence by telling recipients exactly what to do with the money.
Some grants restrict what the money is for, and others put the paternalism later in the process: you tell the funder what you want to do, and they tell you whether it’s a good idea. Big science funders like the National Science Foundation and the National Institutes of Health work this way. This may seem reasonable––after all, aren’t the NSF and NIH full of experts? But science is extremely specialized, so when a scientist asks you for a grant, they almost certainly know more about the science than you do. It’s no wonder that grant evaluations are extremely noisy, reviewers completely disagree on which projects deserve funding, and the scores applications earn correlate very little or not at all with subsequent citation counts (a flawed measure of scientific importance, but one of the few we have).
My own Rhodes experience was the ultimate example in wasteful, paternalistic grantmaking. Most master’s degrees are scams, but Oxford has perfected the art. A friend of mine was once in a departmental meeting at Oxford where the dean announced some new master’s degrees and a professor balked, “But none of us have the capacity for advising more students!” “Don’t worry,” the dean told them, “these are research students; they don’t take any time.” The dean was right: I didn’t take up any of my advisor’s time. In fact, she met with me so rarely that when she wrote my final evaluation, she congratulated me for finishing a one-year MSc degree when I had actually completed a two-year MPhil degree. I learned nothing in my program, and I produced a garbage master’s thesis that I put in its rightful place afterward:
Photo taken immediately after completing my thesis.
5. Applying for and evaluating grants takes an incredible amount of time, but most applications end up rejected
Rhodes Scholarships: 12,000 applications, 99.3% rejected
National Science Foundation: 11,447 applications, 82.7% rejected
National Institutes of Health: 55,038 applications, 79% rejected
Nonprofit grants: too many to count, but one professional grant writer estimates 90% get rejected
These applications are hefty. Before you even start applying for an NSF grant, you should probably read the 79 pages of instructions. The NIH helpfully provides a 10-part instructional video series. Rhodes Scholarships require a mind-boggling eight letters of recommendation. I tell college seniors to expect fellowship applications to be a six-month part-time job.
All this applying doesn’t just burden applicants. Professors run themselves ragged writing recommendations. The NSF relies on volunteers to complete 240,000 reviews every year. Entire university offices exist just to manage the paperwork that grants generate; universities bill this back to funders in the form of “indirect costs,” which at Harvard go as high as 70% of incoming grant funding. Grant agencies seem not to realize that by making everything about their grants burdensome, they allow universities to spend much of the grant money managing the grant itself!
This overhead might be fine if we picked the very best applications at the end, but all the hacking, bad proxies, rewarding of conventional success, and paternalism make me doubt that we’re doing a good job. I’m not alone: scientists are so disgruntled with grant funding that some seriously suggest just picking the winners out of a hat instead.
Some funders address some of these problems, but nobody addresses all of them. Emergent Ventures has a simple application and is willing to fund unconventional people like anonymous bloggers and “the world’s greatest hitchhiker.” The MacArthur “Genius” Fellowship gives out unrestricted funds and doesn’t accept applications, but still rewards conventional success: a third of their recipients have degrees from Ivy League schools, and the winners include tenured MIT professors, bestselling authors, and Lin-Manuel Miranda (after he won a Tony Award). These people have plenty of money already, and if they really needed more, they could find ways of getting it.
Even smart people who are willing to buck all the trends end up making the same mistakes, as scientist-blogger Scott Alexander lamented after running his own grants giveaway. He found himself buried under hundreds of applications, struggled to pick the right ones, and ended up personally heartbroken. He summarizes:
If you’re wondering whether or not to start a grants program, the most honest answer I can give is “I tried this once, and now I’m hoping to invent an entirely new type of philanthropic institution just to avoid doing it again.”
A proposal for Trust Windfalls
All of the problems with grant funding have something in common: they arise out of imperfect replacements for trust. Funders scrutinize lengthy applications, insist on personal interviews, and reward people who are already successful because they don’t know which applicants they should trust. They give paternalistic grants because even when they trust someone enough to give them money, they don’t trust them enough to let them decide how to use it.
These are valiant efforts, but there’s simply no replacement for trust. If you’re looking for a babysitter, you could force someone to write about how they’re great at babysitting, demand they provide babysitting references, require them to show you their Babysitters School diploma, and get them to sign a contract declaring they will definitely not harm your child, and after all that they wouldn’t be even 10% as trustworthy as your mom. Applications can be hacked, references and credentials are bad proxies for actual skill, and even the tightest contracts can be broken. Your mom, unless you are very unlucky, simply loves you and your child, and she doesn’t need the threat of jail time to make her a good babysitter.
Instead of trying to replace trust, then, grants should take advantage of it.
To avoid being hacked, funders should evaluate candidates when they don’t realize they’re being evaluated.
To avoid relying on bad proxies, funders should know grantees personally.
To avoid rewarding conventional success, funders should reach into distant, untapped networks via a chain of trusted people.
To avoid being paternalistic, funders should give unrestricted grants.
To avoid wasting everybody’s time, funders should dispense with applications altogether.
Put these rules together, and you get what I call Trust Windfalls, which work like this:
The funder identifies a trusted individual who knows good potential grant recipients. This person becomes their anonymous Agent.
The Agent gives unrestricted grants––Trust Windfalls––to people they know well. The recipients don’t know they’re being evaluated.
The Agent then passes the baton to a second Agent, ideally someone on the edge of their social network, who knows a different set of good potential grant recipients.
Repeat Steps 2-4.
Trust Windfalls should have as few rules as possible, but here a three simple and reasonable ones:
Nobody should know they’re being evaluated for a Windfall, but Agents should gather whatever information is necessary to ensure that recipients can handle the opportunities the Windfall would provide.
Windfalls should go to people who Agents believe have good hearts and good ideas, who are somehow held back by lack of money, and who, because of who they are or what they want to do, don’t fit in well with existing institutions.
Windfalls should catalyze work that benefits more than just the recipients themselves.
In this way, Trust Windfalls are kind of like Genius Grants, but with more trust and fewer Tony Award winners.
As examples, here are some of the people I would pick for Trust Windfalls:
My friend J., who translated the memoirs of a Russian gulag theater director (yes, the gulag had theater!) just because he liked doing it and thought it was an important piece of history. Nobody paid him in money or prestige to do it. I think that is beautiful and exactly the kind of gumption that deserves a Trust Windfall––I want to know what J. would do next if he didn’t have to worry about money.
Slime Mold Time Mold, who do mad science. They have a theory of the obesity epidemic that I think is very important and likely true, among lots of other good ideas. I know them well enough to know that they have a list of things they would start doing tomorrow if they got a Trust Windfall.
My friends K. and O., who… look, it’s hard for me to explain why they deserve Windfalls. They’re very good and smart people who make valuable contributions to other people’s lives in unquantifiable ways, and I trust that they would do something brave and interesting if they got a chunk of money. It’s hard to appreciate just how important they are without knowing them well, and that’s precisely the reason why I think a grant should exist that would reward people like them.
Doesn’t this amount to just giving money to my friends? Yes. That’s the whole point. I know these people very well, better than any funder ever could. I know they are kind and smart and brave. I know that a grant could change their lives, and that they’d do something wonderful with it. They’ve done all their good work without any expectation that someone would show up and pay them for it. That’s exactly what a funder should be dying to find.
But isn’t it unfair that a bunch of money should go to my friends? Also yes. That’s why, if I was an Agent, I should only get one turn at awarding Windfalls. Then I’d have to pass on the responsibility to someone very different from me who I trusted to give out the second round. If I did it right, Trust Windfalls would eventually find their ways into corners of the world that conventional grants could never reach. Just a few trusted links away from me might be a Botswanan ichthyologist or a trucker smuggling medical supplies into Kiev––people who may not speak English or know the right things to say on an application or even realize there are grants they could apply for in the first place. Making Agents temporary also prevents the Trust Windfalls from being hacked: once people know you’re an Agent, every interaction with you becomes a grant application.
If people hate conventional grant funding so much, why haven’t they tried something like this? Honestly, I think it’s because trusting people seems a lot scarier than it really is. Funders have to trust Agents. Agents have to trust their grant recipients, and they have to trust the person they nominate as the next Agent. (We should maybe call the organization that oversees all this the Trust Trust.) Anybody could betray the trust put in them, which would be a huge shame and very embarrassing.
Applications, evaluations, and interviews all sound a lot safer, but they’re not. Just think: every terrible boss, coworker, and college classmate you ever had got into your life through applications, evaluations, and interviews. Your best friends got into your life through trust. If trust is good enough for picking the most important people in your life, it’s good enough for picking grant recipients.
As far as I can tell, 0% of money is currently given out this way, meaning this form of altruism is totally unexplored. Trust Windfalls may sound crazy, but what’s really crazy is giving out every grant dollar using exactly the same method. And if Trust Windfalls fail, I’d rather fail because I trusted the goodness of humanity than scrape by because I covered my ass from all directions.