Let’s build a fleet and change the world
Abandon Big Ship, get on a Little Ship
I write a lot about how science works and how it doesn't work so well. So I talk to a lot of people who would like to make science better: “What if we changed the incentives?” “What if we ranked scientists on how good they are?” “What if we used AI to, uh, you know, do something?”
To these folks, I usually say: go for it! Let a thousand flowers bloom.
But something feels off about these conversations. It's like we're all aboard a big research ship exploring the ocean and we’ve spotted an iceberg on the horizon, and now we're all shouting about what we should do, except everyone disagrees about where the iceberg is and how to avoid it, or whether there's an iceberg at all, or whether we're going to it hit soon or we already hit it in 1971, or if hitting the iceberg would actually be a good thing because it would discourage people from burning fossil fuels in order to explore the ocean, or if the iceberg is technically an ice mass because "iceberg" refers to a specific kind of ice formation with sheer sides of greater than 100 meters, and which only appear in the Antarctic.
We get so caught up in trying to steer the ship that we don’t even ask: why are we all on the same ship in the first place?
After all, I didn’t come aboard to administer the ship but to explore the ocean. And the best way to explore is to have many ships going in all different directions, not one ship that can only go in one direction. Some of those ships will end up going in circles or crashing into reefs or getting blown over in storms. People on one ship will often think the people on the other ships are wasting their time: “You’re heading to the Gulf of Aden? You fools, there’s nothing to learn there!” But if one ship discovers something—a new island, a new kind of fish, a new passage between continents—and sends out a signal, now everybody knows about it. That’s what strong-link science looks like.
This is the fundamental problem with most attempts to reform science: everybody’s trying to turn the Big Ship, nobody’s launching Little Ships.
I’ve got a design for my own Little Ship, my literal dream boat. But first I have to explain two things that happened to me that changed me completely and forever, because they’re the reason I want to launch a Little Ship in the first place, and they form its founding principles.
First, a man with a goatee told me to murder a bunch of ideas, so I did. When I showed up to do a PhD with the social psychologist Dan Gilbert, he began our first meeting by saying: “Let’s just talk about interesting stuff.” And then we did that for the next five years. We didn’t even run any studies for the first six months; we must have talked about 500 different ideas and thrown away 497, which I painstakingly moved to a file on my laptop labeled “THE IDEAS GRAVEYARD.”
I was anxious to produce papers and I thought we were wasting time. I only realized much later that Dan was honing my intuitions: enlarging my olfactory bulb so I can sniff out bullshit, sharpening my retinal cells so I can spot flaws in my own reasoning, tightening my eardrums so I can hear when a sentence sounds right. The only way to do that was to dissect hundreds of ideas with someone who knew dissection better than I did.
But it was more than just me ‘n’ Dan. He’d get together with all of his students and we’d have group dissection sessions: someone presents, we all crack their project open, inspect its guts, remove the organs that aren’t working, transplant some new ones, and stitch it back together again. I got to watch people make mistakes and disagree, including with the Big Man himself. Older members of the lab would inspect my code and show me where I went wrong, or they’d field one of my half-baked ideas in the hallway and set me on the right track.
This is why I’m always going on about how you can't reach the brain through the ears, how you gotta get the vibes right, how the woo is irreducible. Nearly nothing I learned in my PhD was taught to me explicitly, and I can’t explain it to you here—how do you compress years of conversations and observations? I steeped like a human-sized tea bag in a big mug of science-juice, and it permeated every gyrus and sulcus of my dumb little brain and made me a different person. There was no way to speed up that steeping process, not even if you dunked me a bunch of times. That’s why I’m so obsessed with understanding how to produce that juice and, more importantly, how to fill the mug for the next generation.
I thought this was just how all of academia worked. It ain’t. Most PhD students are lucky to get even an hour with their advisor every week; some get little more than an email every few weeks. They often end up running with their first publishable idea, which is sometimes provided for them by their advisor. Dissecting each other’s work requires a rare amount of trust, time, and ability—people are often happy to offer superficial advice, but you have to be pretty tight with someone before they’ll listen to you for 90 minutes and then gently explain why they don’t think your project is worthwhile. If there’s one thing Gilbert Lab alumni will agree on after we’ve gone out into the world, it’s that we’ve never found another Gilbert Lab.
The second thing that happened was that I started writing this Substack. For years, I thought I just wasn’t that good at writing about ideas. I’d sit down to write a journal article and the words would come out all stilted and wrong, or they wouldn’t come out at all, and I thought that was my fault. Then I started writing in my own voice, in public, in a way meant to be read by anyone, and it felt like someone snipped off the 50-pound weights that had been hanging from my earlobes. It didn’t just get easier to express my thoughts—I started having new thoughts entirely.
What really completed this transformation was publishing Things could be better, a real scientific paper written in normal words, with my friend Ethan. When it got more views than my previous two journal articles combined, I went, “oh.” Someone tweeted something like, “I read this to my eight-year-old this morning and she understood it” and I went, “ohhh.” Other people started publishing accessible, public research and I went, “ohhhhhhhh.” It sounds silly now, but I didn’t realize any of that could happen, that you could write about your research in your own voice, be honest, post it as a PDF, and people would actually pay attention.
It’s historically weird and pretty dumb that paywalled, peer-reviewed journals have a monopoly on scientific communication. The worst part of this isn’t the time and money it wastes, nor how it makes papers boring—although that’s bad too—it’s how this system kills ideas in people’s minds before they’re even born. As I wrote before:
Just knowing that your ideas won’t count for anything unless peer reviewers like them makes you worse at thinking. It’s like being a teenager again: before you do anything, you ask yourself, “BUT WILL PEOPLE THINK I’M COOL?” When getting and keeping a job depends on producing popular ideas, you can get very good at thought-policing yourself into never entertaining anything weird or unpopular at all. That means we end up with fewer revolutionary ideas, and unless you think everything’s pretty much perfect right now, we need revolutionary ideas real bad.
Every Little Ship should have its own founding ideas, and these are mine:
Learning how to do science is a mysterious, ineffable process where you must spend a lot of time hanging out with someone who knows the ropes and other people who also want to know the ropes. It is only through this process that you can hone your intuitions, your most precious research abilities.
Science must be expressed honestly, in the voice of the person who did the work, in words that any interested person can understand, and in a place where they can access it. It is only through this freedom that you can fully cultivate your ideas.
So what’s the most efficient way to build a Little Ship where you can do that?
My answer is: buy a house. Your biggest expense is rent for the places where people live and work, so you save a lot of money if you can bundle those costs.
Here’s what it would look like. Trainees live upstairs and they do their work in a lab space downstairs. Their mentor (in this case, me) lives nearby, and we all hang out and do science together. The students stay for a fixed term—say, four years—and they're staggered so that there's always an older student on the way out and a newer student on the way in. There are additional slots for a rotating cast of visiting scholars, so the students always have more than one mentor. Think Institute for Advanced Study meets the Royal Society, except for weird upstarts instead of rich farts. I call these Science Houses, and mine in particular would be a Psychology House, or, as my friends suggested when I ran it by them: “Professor M’s School for Troubled Mammals.”
Living in a Science House is much like doing a PhD, but without the worst parts. The students of Science House will not experience mind-crippling anxiety about whether they'll get a tenure-track job; they are guaranteed not to.1 They won't have to convince themselves that it's cool and noble to produce boring, paywalled journal articles that no one reads; they will instead produce open-access research that's meant to be read by anyone, like this. And they won't spend months stressing out about general exams that turn out not to matter at all; they will instead do...literally anything else.
Science House would always be small. When things get too big, bureaucratic psychosis sets in. Instead of expanding Science House, I'd rather build more Science Houses that other mentors can run independently (and I’ve got a few folks in mind). The students of different houses can do projects together and have barbecues and what not, but otherwise each Science House should flourish on its own—the whole point is to have many Little Ships going in different directions. Eventually, you could have a whole fleet of Science Houses doing their own thing: a Psychology House next to a Theoretical Physics House, a Weird Biology House down the street, and a Botany House around the corner.
Those houses won't just be places where students live and work—they're the home for a scene. We're at the beginning of a moment right now, where more and more people are doing science outside of academia. We've got people running obesity studies on the internet, surveying hundreds of thousands of people about their kinks, building biology labs in their basements, and starting open-access journals, all without a connection to a university. Folks are leaving their academic posts to do their work in the wild. This movement mainly lives online right now2, and it needs a portal where it can enter the physical world, a place where people can visit and give talks and discuss ideas. Just like every artistic movement had a bar or a cafe where its key figures hung out, this new science needs a house.
Science Houses aren’t for everybody or everything. Sometimes you have to build a Large Hadron Collider, or you need to spend millions on petri dishes or expeditions to the Arctic Circle. But lots of science isn’t like that; what you really need is people hanging out with each other, thinking long and hard, and running some experiments. “For everything else, there’s Science House.”
(Science Houses also shouldn’t be run by just anybody. Some people don’t have good taste, some people go bonkers when you give them a bit of authority, and some people are more interested in birthing their own ideas than midwifing other people’s ideas. They should go do other stuff. This is my dream in particular because I am a big dopey resident advisor at heart and I love talking to students, which is why I spent a big chunk of my PhD just hanging out with research assistants in the lab when I “should” have been writing papers.)
By my calculations, you could endow a Science House in perpetuity—as in, buy the house, pay the students and the mentor and the visiting scholars, give everybody health insurance, fund the research, do all the legal and accounting stuff—for $15 million. Let’s say you spend a million on the house itself, turn the rest into an endowment and live off the ~4% interest, which gives you $560,000 per year to spend on personnel, research, house repairs, and the legal and accounting stuff necessary to stay operational.
That’s a big chunk of money in absolute terms, but it's nothin’ in academia. In 2022, Harvard spent about $15 million on postage alone. That is, for the same amount of money Harvard University spends every year to mail rejection letters, you could fund a new scientific institution forever.
In fact, let me show you just how much bang for your buck you can get with a Science House vs. the usual way we fund science. ‘Cause right now, we waste most of that money.
Say you've got $100 million to put towards research. You might think that you just start cutting checks to professors so they can hire postdocs and buy lasers and stuff. But you can’t do that. Those professors work at universities which, much like mafia dons, get a cut of all the money that passes through their turf; they call ‘em “indirect costs.” Johns Hopkins, for instance, levies a 63.75% tax on the federal funds that its researchers bring in. So for every dollar you pay a Johns Hopkins professor, you have to pay Johns Hopkins University an extra 63.75 cents.
Universities would point out that they provide many essential services with those indirect costs. They pay to keep the lights on and the toilets clean. They pay for insurance premiums so that if your lab burns down, you can build another one. They pay for libraries so you can look stuff up. Somebody has to foot that bill, or else you can’t do your research.
(Indirect costs also sometimes "accidentally" pay for other things, like depreciation on Stanford University’s yacht. Rather than building new Little Ships, we are literally paying for the cost of a Big Ship becoming less valuable.)
So it’s fair to have some indirect costs, but the only people who think that 63.75% is a reasonable amount are the people getting their yacht depreciation paid for. If every dollar you spend on actual research stuff (salaries, health insurance, materials, equipment, travel) requires you to spend another 63.75 cents on non-research stuff (administration, accounting, utilities), you're doing a bad job. By the way, did you know that Yale, for example, has almost as many administrators as undergraduates?
(To be fair, these grants require a small army of administrators to fill out lots of paperwork, and so part of what grants pay for is the annoyance of getting a grant. As you squander your grant dollars on bureaucracy, remember that you are partly to blame.)
THE MOST EXPENSIVE FORMS IN THE WORLD
So you’ve blown about a third of your budget on administrators and toilets. The rest goes to science, right?
Not exactly. Professors need grants to do their jobs, and applying for grants takes a long time. One Australian study found that researchers spent an average of 34 days per grant proposal, and the Office of Research and Innovation at UT Dallas claims you need "at least" 120 hours to prepare a standard government grant application. But ~80% of those applications will be denied, meaning the researcher will have to keep applying or else they can't pay for their studies, and they might eventually lose their job. That's why the cost of applying for and evaluating grants takes up somewhere between 10% and 35% of the grant budget itself. (Nobody puts “grant application time” as a line item in their budget, of course, but those working hours have to come from somewhere.)
You've now spent ~$30 million of your $100 million on indirect costs and ~$22 million on grant applications. Over half your budget is gone and nobody's actually done any science. And there’s no guarantee the rest actually ends up funding worthwhile research. Government grant applications are peer-reviewed by grant committees that notoriously disagree with each other, penalize risky and interdisciplinary research, maybe toss a few grants to their friends, and fail to predict which projects ultimately end up being useful. Something like 10-30% of the papers that come out of your funding won't be cited within five years, or possibly ever.
Maybe it's not crazy to distribute some science funding this way, but it's definitely crazy to distribute all of it this way. With your $100 million, you could have created six Science Houses permanently and still had over $600k left.
KEEP THE WEIRD ALIVE
So a big advantage of Science House is that you can get more science for less money. But what really gets me really revved up about Science House is imagining what some talented young people could do if you sorted out their housing and gave them a bit of cash, some guidance, and a lot of freedom.
Right now, pretty much everyone doing science is a product of the academic-industrial complex, the kind of person who had to do everything right at every step to make it through the ever-narrowing pipeline into tenured positions at rich universities. But the greatest scientists often don't look like that kind of person at all. They're weird, they spend years working on things that go nowhere, they piss off powerful people, they worship puppets and learn arithmetic by smell. Nobel laureates keep telling us that they would never survive in today's academia; we shrug and soldier on.
I’m watching this unfold in real time, right now. Year by year, my most interesting colleagues leave academia. They leave because they know they’ll never get an academic job, because they’re tired of playing the science game, because they’ve been ground down inside the gears of a giant bureaucracy that says all the right things and yet does all the wrong ones.
Even worse is when I meet a talented student, someone just starting out who’s got that spark, that precious and indescribable potential to do something interesting. Inevitably, we have The Talk: they kinda like all this science stuff, so what should they do after they graduate? I have to tell them they have two bad options.
One, you can attempt to join the Big Ship and go get a PhD. Some people like the Big Ship a lot, but many people have a terrible time. The pay is crummy, some of the advisors are abusive, and they will most likely toss you overboard once you’re done with your program, because although it’s a Big Ship, it’s already too crowded. There are things you can do to increase the chances you can stay on the Big Ship forever, but most of them will make you feel ashamed: publish at all costs, limit your personal commitments so you can move anywhere, don’t do anything too risky or weird. People will tell you that you can stop doing all this once you get tenure, but if you spend the next 10-15 years of your life doing something, that’s just who you are.
Or two, you can jump into the ocean and try to swim by yourself. You’ll do exactly the work you want to do, but you’ll do it without any support. Act like you’re trying to make it in music or art: find a way to make a living, then do your thing on nights and weekends, knowing you might never be able to ditch your day job. But, uh, there’s a Discord!
It’s sad enough when someone doesn’t get to live out their full potential. But for society, for all of us who never get to benefit from the discoveries these people would make, it’s tragic. We are abandoning a generation of great scientists, and we will pay the price for it.
One day, though, I hope to tell that student there’s another option: “Have you ever heard of Science House?”
Experimental History is made possible by those brave enough to keep the weird alive
Sometimes when I describe Science House to friends in academia, they worry about what Science House alumni will do afterward, since they won't be competitive for academic jobs. I find this hilarious because most PhD students will not get academic jobs either. You'll often find PhD students from my field doing a crash course in data science at the end of their program so they can get a job at a tech company because their education didn't make them very useful for anything other than being a professor. Getting good at academia is like getting good at Scrabble: it's just a made-up game, so many of the skills you acquire while playing it are useless anywhere else. Turns out not a lot of places need someone who knows exactly how to score a “revise and resubmit” at the Journal of Personality and Social Psychology. On the other hand, getting good at identifying interesting problems, gathering data on them, and writing up your findings in accessible language is a pretty marketable and widely applicable skill, and that's the business of a Science House. These students will have to be weird and ambitious to join up in the first place, so I think they'll do just fine.
As a reminder, I host a little corner of this movement in a Discord. If you'd like to join, email firstname.lastname@example.org.