Note: this article is sponsored by and cross-posted to the Good Science Project. They also write a fair bit, and their articles were essential reading for writing this essay!

Also, this article would not be possible without the hours of discussion/editing help I’ve had with several people from these institutions, and a few outside of them. Huge shout-out to all of them!

Introduction

Arcadia Science, Speculative Technologies, FutureHouse, Arc, and Convergent.

All of these are a new form of scientific institute.

Most are funded entirely by a few billionaires. Most are non-profits. Most of them focus on the life-sciences. Most of them have sprung up in just the last few years.

They do all also have one common thread: a grand statement. We are an experiment in a new way to do science.

And they are! Traditionally, research is conducted in academic or private industry labs — dependent on NIH grants in the former and markets in the latter. Given the (often singular) sources of no-strings-attached funding, these new institutions need not satisfy either the NIH or the markets, allowing them to conduct research in a unique fashion.

In one sense, the experimental aspect of these institutions revolves around the focus of the research itself, addressing fields or using methods that the founders — correctly or not — view as underserved/underutilized. But, on a more subtle level, the experimental aspect could be more closely tied to the culture of these organizations. Institutions like Arcadia, FutureHouse, and the rest could be viewed as the production of auteurs — a term from filmmaking for films with such a heavy sense of the director’s personal taste that the film is inseparable from the director.

This is where the novelty within these institutions primarily lie, in how the founders of the institute wish science was conducted. And wielding billions of dollars, thousands of hours of work, and hundreds of scientists as a means to test whether their theories are correct.

Of course, nothing under the sun is truly new. There is an age-old history of scientist dissatisfaction with how ‘things are traditionally done’, and confidently building new institutions to solve the problems they’ve seen. Many of these are now household names amongst researchers: Broad Institute, Whitehead Institute, Max Planck Society, Howard Hughes Medical Institute (HHMI), and so on. Each of these were started with similar contrarian mentalities as the current era of institutions.

Some of these were more experimental than others, most notably HHMI, which prized itself on its focus on interdisciplinary research above all else. But all were experiments, many of them extraordinarily successful. Yet, the current iteration of new research institutes is still arguably more experimental than its ancestors. While the last generation of institutes was typically tied directly to universities, the current era of ones (outside of Arc) are independent, allowing them a larger sense of opinionation on how science should be done.

But, despite this experimentation, there is relatively little information out there on what’s going on inside them. Not in terms of science, but more-so the vibes. While aspects of these organizations have been written about previously, such as in articles in The Atlantic and Endpoints, they aren’t assessing vibes! These other articles are, first and foremost, news-pieces; valuable, but lack any opinionated observations on the inner-workings of the institutions. Nadia Asparouhova’s essay on the subject comes closest to this regarding the history of these institutions, but still few details on how they practically function. This essay attempts to discuss that missing set of vibes.

To do this, I’ve interviewed several people either involved in these institutions or knowledgeable about them, learning about how they view its internals. I’ve taken notes from our discussions, tried to build up common threads amongst all of their unique experiences and thoughts, and compiled them here.

Yet, in many ways, this essay is written too early. Many of these organizations are still in their infancy, only a few years old, still figuring out how to best enact their visions. Despite this, the aforementioned common threads did exist, and I believe they would make for an interesting essay.

I’ll quickly note two things:

1. Nothing that is written here is an expose, controversial, or secret. Those familiar with metascience, the study of how science is best done, may find my observations obvious. That is expected! This article is not intended to put forward brand new theories or say anything particularly inflammatory. It is purely meant to satisfy a curiosity I, and others, have had about this particular sector of research since it first arose.
2. Not all of these observations equally apply to everybody. There are inklings of every topic here across all institutions, but some of them apply more deeply to some than others.

That’s it! Let’s move on. 

Background

Sorry, a few more things actually. Before moving on, it may be helpful for readers new to metascience to get a bit of the lay of the land. Feel free to skip this section if you’re already well-aware of these institutions and metascience in general.

First, the institutions themselves. Let’s go through them and what makes them unique. There may be others that should be included here, but these five represent the most well-known institutions.

Arc Institute, launched in 2021, studies life-sciences. Their angle is in no-strings-attached grant funding to fund whatever a PI desires to study. While this is unique from an academic angle, it isn’t unique historically; the industrial labs of old (Bell Labs, Xerox PARC, etc) operated in similar ways. Arc hopes to replicate the same extraordinary success they had using the same approach. This is likely the most ‘normal’ amongst the list, especially given that they are the only ones with standing university affiliations, specifically to Stanford, UCSF, and UC Berkeley. A very close parallel to Arc would be the Broad Institute, though, crucially, the Broad doesn’t provide internal funding in the same way Arc does!

Convergent, founded in 2022, studies basically anything, though their current projects are in life-sciences and math. They have a strong focus on so-called ‘FROs’, or Focused Research Organizations. They internally spin up scientifically ambitious projects, assign leaders, offer them no-strings-attached funding for 5-7 years, and then largely leave them alone to hire + work. After the time period is up, money is withdrawn, and the FRO either disbands or elevates itself to something larger. In their words, FROs are meant for projects that are ‘bigger than an academic lab can undertake, more coordinated than a loose consortium or themed department, and not directly profitable enough to be a venture-backed startup or industrial R&D project.’.

Arcadia Science, founded in 2021, studies life-sciences. They have an explicit focus on two things. One, the biology of diverse organisms for which relatively few tools and resources exist, especially non-model organisms. Two, the more unique part, a strong commitment to open science; all research they release is publicly hosted only on their website, never in journal publications. It is also, curiously enough, the only for-profit institute on this list, with a deep focus on commercialization. All employees receive equity in whatever venture is spun out of it. 

Speculative Technologies, founded in 2022, studies materials and manufacturing. They focus heavily on blue-sky research that is likely to fail, but would fundamentally change the world if successful. The unique part about them is their explicit focus on using the “ARPA model”. What is the ARPA model? We’ll get into that in just a few paragraphs. If curious, Speculative Technologies also has a great newsletter on what’s going on inside of it!

Finally, FutureHouse, founded in 2023, studies life-sciences and AI. This seems to be the latest comer to the game, so there are few details on its cultural ethos. What we do know is that their end-goal is to create AI that can automate the scientific method in biology, and to offer as much financial freedom necessary for researchers to discover how to do this. 

I mentioned the ARPA model in the Speculative Technologies blurb. What is that? Pretty simple: it’s an organizational approach to how research should be conducted. And there’s good reason to believe it’s a decent model for exactly that. After all, DARPA, a defense agency using the ARPA model, orchestrated some of the greatest inventions of the 20th century, most notably an early precursor to the internet. 

But what actually is the model? It’s easy to overcomplicate it, but, generally, the ARPA model is a combination of giving large amounts of agency to project leaders, low bureaucracy, and hiring people with high intrinsic motivation. If you provide all of these, magic seems to happen — though with high variance. There are many other aspects to a true ARPA model, but these essential characteristics seem to pop up over and over again as being the most important.

Why did I wait to explain this? Because, while Speculative Technologies has explicitly tried to follow the ARPA model, almost every organization listed here takes inspiration after it.

Amongst the founders of these institutions and metascience enthusiasts in general, there is a deep appreciation for the ARPA model, its history, all that it entails, and a hope to replicate its success. The acronym pops up all over the place when looking into these institutions! If curious to read more about ARPA specifically and how it has been historically applied, Eric Gilliam’s posts about it are incredible.

Still though, it’s important to not consider the ARPA model as always a good thing. 

One person I interviewed posited that the personal autonomy given to project leaders — done to reduce friction — leads to a huge amount of waste pursuing dead-end ideas. As in, generously speaking, the ARPA model works well for closed-ended, focused research endeavors. But, the more ‘blue-sky’ a project is, the more failure modes appear, and thus the greater need for high-touch oversight to prevent billions spent on useless research.

Another person believed that the ARPA model was successful in DARPA’s case only because a very wealthy military was the singular customer. In the eyes of this person, without a pre-agreed-upon, mission-focused customer with deep pockets, spiritual successors to ARPA are unlikely to pay off. I found at least one academic paper — written by four people closely involved with governmental ARPA programs — that argued something similar.

And, with this necessary background information, we’re ready to move onto the actual observations I’ve had.

Observations

Selection pressures will always exist

Each of these institutions makes a strong internal attempt to combat whatever parts of traditional academia they have chosen to make a stand against; funding pressure, low-ambition projects, and so on. But when we say ‘strong internal attempts’ to go against the tide of some part of academia, what are we referring to?

In a sense, culture. The organizational structure, how it communicates science internally and externally, what projects it chooses to focus on, and so on. But there isn’t a free lunch when it comes to culture.

Ruxandra Teslo recently published an essay called ‘The Weird Nerd comes with trade-offs’. In it, she argues that the aspects of genius embodied in the 'Weird Nerd' rarely align with the culture of academia. The politicking, charisma, and focus on 'achievable' projects needed to succeed as a university PI often clash with the intellectual courage and ambition we'd ideally want from our leading scientists. While overlap sometimes occurs, the false negative rate is undoubtedly high.

These newer scientific institutions are addressing some of the issues that have historically hindered a 'Weird Nerd's' academic career. However, in doing so, they've introduced a new set of filters.

Because of this, a fair number of incredibly talented people will be turned away because they don’t fit the ethos of these institutions. Such individuals may enjoy publishing in journals, desire to focus on a specific subfield instead of being generalists, or dislike computational problems. None of these are explicitly bad things, just personal characteristics that these institutions don’t align with. 

Human capital, as Ruxandra puts it, is deprioritized in favor of the culture that the institution hopes to maintain.

This is a lesson that undoubtedly many places — research and non-research alike — have learned. No matter how well-intentioned the culture behind an institution is, there will be plenty of false negatives amongst those it rejects. It is best to come to peace with this early on and recognize it as an unfortunate consequence of any attempt to organize humans towards a common goal.

But this may immediately seem arrogant. It intuitively feels deeply short-sighted to hold onto a vision of how science should be done at the expense of recognizable genius.

These institutes’ prioritization makes sense given some extra context: the focus on culture over talent was reactive. This leads us to the next point.

Organizational unity is extremely important

In the initial hiring phase for some of these groups, talent was enough to overrule bad culture fits. Being able to hold onto a particularly prodigious and ambitious scientist was worth it! Even if the scientist in question had a personality that ran counter to the ideals of the institute.

But what they found is that bad culture fits are corrosive. The desire to ‘make things work’ with these talented — but ideologically opposed — researchers simply did not work. The issue wasn't just a matter of interpersonal friction, though that certainly played a role. More fundamentally, these misaligned researchers often pursued projects or approaches that diverged from the institute's core mission and methodology — often eating up resources or manpower that, in the eyes of the institution, could be better used elsewhere.

The corrosive aspect of this all manifested via a sort of cognitive dissonance within the organization. Here you have an institute founded on certain principles — say, a focus on a certain approach to computational problems — suddenly harboring individuals who fundamentally disagree with those approaches. This dissonance doesn't just stay contained to that individual; it ripples outward, causing other team members to question the correctness of the institute's value system.

And new cultures are fundamentally fragile. If threatened, people may simply revert to using the same heuristics that the traditional academic system taught them. And then, slowly, the institute is no longer an experiment, but just another branch of traditional academia, with no semblance of the original vision that originally brought everyone together.

One lesson here: hire carefully (and fire quickly) to maintain organizational cohesion. But there’s another second lesson here: don’t grow quickly. Not only is culture fragile, it also struggles to scale quickly. Multiple people I talked to cited scientific organizations that had clear visions at the start, hired a hundred people within the span of a few months, and extreme cultural fragmentation was the end result.

This is a problem that feels more unique to this new age of scientific institutes, given that they are given massive lump sum cash infusions at the very start of their creation. In this respect, all of the institutions I’ve talked to are doing quite well, growing slowly and carefully to ensure that their shared ambitions are maintained.

Importance of scientific marketing

This will be a point obvious to many: creating a splash is important. Not exactly for the laymen's perception of an organization — which, for the most part, isn't all that useful — but more for attracting talent. Scientists and engineers alike are ultimately human, and the outward appearance of Big And Important things happening at a research institution goes a long way in piquing their interest.

It should happen fast too!

It was surprising to learn that several of these institutions are spun up with several projects mostly already completed. For example, if an institute is announced, and a mere two months later they release an incredible paper/result, it is usually the case that the project had been completed long ago. Obvious if you think about it for a second — science takes time, even with perfect culture and perfect funding. But it helps drive the momentum behind the launch of the company and serves their message: “Our way of doing science has yielded interesting results. Work with us!”

Of course, the flashiness should be supported by something more substantial; other ongoing projects, funding support, and so on. Any employee who understands their value will do their due diligence on that front. But marketing leads to far more curious eyes peeking into the organization to see if it’s worth their time.

There’s another side benefit to marketing somewhat related to the above two points: it helps people self-select into where they best belong. A very not-risk-averse person who enjoys academia probably shouldn’t end up at these institutions, and it is ideal if potential employees are aware of that beforehand. Marketing helps teach this!

Interestingly, multiple people mentioned Calico, an Alphabet-run lab focused on the biology of aging, as an example of a failure mode they were hoping to avoid. When the organization was first announced in 2013, it promised to fundamentally change the aging field, it was one of Google’s ‘moonshots’ after all. Even I remember reading about it when I was in high school! But, outside of that first initial news cycle, most people never heard about them again.

This was intentional. Calico chose to be incredibly secretive. They never talked about their ongoing work, what they discovered, and never even revealed their research focuses. This has changed a bit, as they now actively publish, but it is still somewhat shrouded in mystery.

It was such a bizarre strategic decision on their end that even the Wikipedia article for Calico mentions it:

When Calico was formed, Google did not disclose many details, such as whether the company would focus on biology or information technology. The company issued press releases about research partnerships, but not details regarding the results of its research or the specifics of what it was working on. This led to frustration by researchers regarding Calico's secrecy and questions as to whether Calico had produced any useful scientific advancements. Calico said the business' purpose was to focus on long-term science not expected to garner results for 10 or more years, leaving nothing to report on in its first five years.

Calico was started in 2013, so it has been more than 10 years. Has it lived up to its promise? In my limited knowledge, not really, it feels like relatively little of the most publicly exciting work in longevity is coming out of them. As per 2021, Aubrey De Gray seems to agree.

Would having been more public with their results helped? It’s hard to tell. The fundamental problem behind Calico seems to have much more to do with their incrementalist research focuses than anything else. But it may have also been the case that, had they been less secretive, there would’ve been a more public outcry about how their work should be more ambitious.

Or maybe not.

Either way, most other newer scientific institutes have gone the opposite way, adopting a ‘build in public’ mantra. They continuously release updates and progress reports on how their research missions are coming along, leaving relatively little behind closed doors. And it seems to have worked well! Not only are scientists I talked to strongly aware of the promising projects amongst each of the institutions, they also have a decent sense of which ones they’d want to join and which ones they wouldn’t, purely based on each ones culture.

Incentive structures are challenging to pin down

Academia and industry, the two traditional pillars of scientific research, have long relied on different sets of incentives to drive progress. In academia, publications and grants. For industry, patents and products. On paper, this makes sense — academics want to push the boundaries of human knowledge, while companies need to turn a profit.

But these incentives aren't the end goals. They're just proxies, stand-ins for what we truly want. Academics aren't writing papers for the sake of writing papers, they want to make a tangible impact on science and society. And companies aren't just after cash alone, they're aiming to create long-term sustainable value so they can stay afloat. There are second-order downsides to these proxies.

In academia, it’s a glut of papers that make tiny, incremental advances but rarely move the needle in any significant way. Researchers play it safe, churning out "publishable" work rather than swinging for the fences.

On the industry side, the obsession with profits means that a lot of crucial research areas get left by the wayside. If there's no clear path to commercialization, many companies won't touch it, even if the long-term potential is enormous.

These institutes are trying to alter this. They want high-impact science, the type that academia is supposed to do. But they're also deeply aware of how powerful industry incentives are — milestones, equity, and the allure of creating something tangible. They're attempting to thread the needle between these two worlds. In other words, they're after high-impact science in areas that lack immediate commercialization, but would benefit from the same type of usually-only-found-in-industry hyper-focus.

And…it’s hard.

That’s the basic result of the conversations I’ve had. I’ve tried to squeeze something consistent out of my notes, but ‘it's hard’ summarizes a lot of it. Nobody has a good answer for this as of yet.

In many ways, it felt like the conversations I had implied that many research topics are best left to the private sector. Not all problems could fit neatly underneath these organizations. Things that are technically within their scientific purview, but simply too hard to align incentives for. Some of these institutions understood this at the start, others grew to accept it as they floundered to find a place for it. 

Concluding thoughts

When I first started writing this piece, I assumed there would be a clear winner amongst these institutions. One whose cultural experiments would yield an improved way of conducting science, and would be triumphant amongst its peers.

But the more I read into it, the more I realize there will be no such thing.

No institution will suddenly jump out and say "We've figured out how to best do science!" and have every other research lab scramble to copy them. Instead, it's more likely that we'll see a slow osmosis of ideas. The experiments that work well might get adopted elsewhere. The failures will be quietly discarded or morphed into something new. 

Moreover, the successful ideas will likely be domain-specific. A consistent belief amongst many I talked to is that different fields, problems, and practical applications will require different organizational approaches. 

And, retrospectively, that makes sense. Science isn't a monolith, and neither should the institutions that support it be. Pursuing a better electromagnetic stimulation device for epilepsy treatment likely requires a completely different organizational approach than a project dedicated to mapping out a mouse’s brain. The former has the benefit of a clearly addressable clinical market, the latter is blue-sky research with no immediate payoff besides unlocking new questions. It feels obvious, looking back, that the ideal structures to support each type of work will dramatically differ, despite both of them technically being underneath the moniker of ‘research’.

What's exciting is the willingness to try new things. Even if these specific institutions don't become the new paradigm, they're pushing the conversation forward. They're making people think critically about how we organize and incentivize scientific progress. Though the ARPA similarity remains, these institutions are a spin on it, not a wholesale replica. But…are these institutions worth the billions being put into it?

I am a historical advocate that well-meaning people doing scientific research are always worth giving money to. So, in my opinion, it is an obvious yes. But, I think the most correct answer here is ‘nobody knows’. It’s all an experiment, and still in progress. There’s a lot of bullish people and a lot of bearish people. Time will tell who is correct.

One last thing: these new institutions should set themselves up such that the outcomes of their experimental nature are actually measurable. How? One way could be by keeping detailed records of everything — from their entire applicant pool to decision-making processes and project outcomes, such that it could be studied in the future. For example, 10 years from now, a researcher could compare the outcomes of rejected applicants to the outcomes from the institute's accepted employees.

Understandably, it’s challenging to do this in every dimension. Most metascience questions will remain underdetermined for a very long time, but even partial data could be incredibly valuable in learning what the incorrect answers are. 

And that’s all I have to say about this. Thank you for reading!