The Fight For Slow And Boring Research

Since the middle of the twentieth century, the National Institutes of Health and the National Science Foundation have embodied an imperfect social contract: Federal agencies would fund basic research at scale, and in return, that research would serve the public good through medical advances, technological progress, and economic growth. 

For scientists, this system created a reliable pathway: Do good work, write strong grants, and federal agencies would keep your lab running. It was never a perfectly fair system, but it was predictable enough that you could build a life around it. If your work was solid and your grants were strong, the system would fund you.

That implicit deal is breaking down.

The erosion began in 2024, when Congress passed an appropriations package that left the NSF with $9.06 billion in funding, roughly 8% below its 2023 level. Then, in early 2025, a new administration froze reviews and payments at the NSF and the NIH, canceled more than a thousand active NSF grants worth $739 million, and proposed a 50%-56% cut to NSF’s budget in 2026.

Technically speaking, some of the damage has been rolled back. The most extreme budget cuts have been rejected by Congress. Since summer 2025, lawsuits have been filed, budget drafts have been revised, and some grants were resurrected. But the repair has been incomplete. The NIH’s funding for new and competitive renewal grants dropped from $10.5 billion (FY 2024) to $8.6 billion (FY 2025) — an 18% decline. The damage is more concrete at the institutional level: Projects remain frozen and universities have paused or reduced admissions to graduate programs and hiring of postdocs. According to states suing the NSF, the funding cuts have already halted efforts to train the next generation of scientists in several fields.

And trust does not snap back like a line item.

Scientists will still apply for federal grants, but the money will feel contingent instead of given. And so, in the coming years, they will start to look sideways — toward private philanthropies, private investors, and other sources of support. To do so, they’ll have to adapt to a landscape in which the ability to explain their work clearly is as important as the work itself. 

Most tenured faculty will not leave the university — but many of them are already starting to diversify their sources of funding. 

According to a 2025 survey of over 300 researchers, 86% of principal investigators and research administrators are actively exploring nonfederal funding, including from pharmaceutical partnerships and philanthropic foundations. In 2023, nonfederal sources — institutional funds, nonprofits, businesses, and state governments — accounted for roughly 45% of all academic R&D spending.

Universities are also responding: In 2025, Harvard committed $250 million using contingency reserves — not its endowment — to bridge federal gaps. Other universities, such as Northwestern and Johns Hopkins, similarly committed to self-funding affected research from their own reserves. The University of Michigan launched the Michigan University Innovation Capital Fund and Consortium in 2024 to support startups across the state’s public research universities, while MIT spin-out The Engine raised $398 million in Fund III in June 2024, explicitly attracting labs that no longer wanted to rely on federal grants alone.

Philanthropic funders have been expanding their commitments in response. The Chan Zuckerberg Initiative announced in 2025 that it would more than double its science spending, committing at least $10 billion over the next decade — up from $4 billion in its first decade. The Gates Foundation is accelerating its annual budget from $6 billion to $9 billion by 2026, a 50% increase. Many smaller foundations have doubled or tripled specific program budgets in response to federal cuts, according to Mindful Philanthropy, which tracks mental health and science giving.

These increases signal a meaningful shift in the landscape of science funding. Philanthropic organizations now provide roughly $30 billion per year to science and health research, a volume that rivals major federal streams like the NIH. ¹

Venture capital firms are also intensifying their pursuit of academic talent in direct response to federal cuts. The years 2024 and 2025 saw multiple new university-focused funds: Catalio Capital Management raised over $400 million for its Nexus Fund IV to partner directly with academic labs, Pack Ventures raised $30 million specifically for University of Washington spin-outs, and universities from Arkansas and Utah to Cornell launched or announced dedicated investment vehicles. The House Fund launched a fund to invest in UC Berkeley research in artificial intelligence as a direct result of federal cuts. Deep Science Ventures launched the Venture Science Doctorate — a PhD program where students build startups as their dissertations. Lux Capital announced a $100M “helpline” for American scientists. What’s new isn’t that VCs fund academic spin-outs — they always have — but that historically later-stage investors are now creating dedicated programs to court researchers earlier in their careers, viewing constrained federal pipelines as an opportunity to capture academic talent upstream.

These are not insignificant investments. This convergence reflects a structural shift in how American science is financed. While federal funding currently remains the largest single source, its dominance is declining as institutions increasingly distribute risk across industry partnerships, international collaborations, and philanthropic grants.

When a lab relies on a single federal pillar, its only real audience is peer review — study sections, editorial boards, and journals. But the moment a lab pursues multiple funding streams, it signs up for more than one kind of evaluator and more than one way of being judged. 

In other words, funding becomes a serious incentive for academics to invest in their science communication.

A growing share of nonfederal funders have made clarity and public legibility formal parts of their evaluation criteria. The Howard Hughes Medical Institute’s Freeman Hrabowski Scholars Program, which commits up to $8.6 million per investigator over a decade, assesses applicants partly on their abstract, which “should be easily understood by scientists not expert in your scientific discipline.” Schmidt Futures’ AI in Science program, a $148 million initiative across nine universities, advises fellows to develop intelligible proposals for nonspecialists as part of their deliverables. 

Philanthropic and hybrid funders operate under very different constraints from federal study sections. Program officers at foundations spend much of their time preparing concise, defensible narratives for trustees who are not domain experts. A project that is technically sound but illegible is difficult to justify in a half-hour board meeting; a project with clear questions, visible progress markers, and intelligible downstream impact is far easier to defend. This is why funders increasingly require public impact statements, plain language summaries, reproducible code, and documented datasets. These are not marketing artifacts; they are the materials that let a program officer justify that basic research still yields returns. 

That is just the philanthropic side. Startups and companies — especially those building tools, platforms, and infrastructure for scientific work — look for academic collaborators with  well-documented datasets, code, and analyses that are reproducible by outsiders, and terms that make it clear whether a lab can plug their systems in and move a field forward in ways that others will notice and copy.

Labs and institutions that successfully attract diverse funding sources tend to invest significantly in communication. It’s difficult to prove that communication directly leads to funding success — there’s a long path between first contact and a signed check — but the pattern is hard to ignore. Research on philanthropic giving patterns shows that donors give more when they can see how their funds will be used, and local philanthropy operates through “relationships and outreach, rather than extensive proposals.” David Baker’s Institute for Protein Design at the University of Washington has attracted funding from HHMI, Open Philanthropy (now called Coefficient Giving), and The Audacious Project, as well as federal agencies — successes that followed his early investment in public engagement. In 2008, Baker launched Foldit, a protein folding game, and built open-source tools like Rosetta@home years before major philanthropic grants arrived. By the time Open Philanthropy committed $11 million in 2018, his lab’s work was already accessible to over a million citizen scientists. CASP, the nonprofit protein folding competition that put AlphaFold on the map, operates on a lean budget and has never been a commercial entity, but its public leaderboards and transparent evaluation framework made it an anchor for the field of protein structure prediction  — so much so that when NIH funding lapsed in 2025, DeepMind stepped in as a sponsor to keep the contest running. These programs represent hundreds of millions of dollars flowing through evaluation processes in which communication is an increasingly critical criterion. 

A lab that cannot explain its trajectory in plain language is at a disadvantage long before peer review even begins.

Universities are good at something almost no startup can afford: deep, slow, and obscure work. Long-term cohort studies that follow people for decades. Experiments on obscure model organisms. Foundational theory. Messy longitudinal datasets that make sense only on the third or fourth reanalysis. These are exactly the things that most alternative research institutions depend on but are rarely incentivized to host themselves because they cannot be justified on a quarterly road map or as a near-term feature.

Suppose universities accept that the funding landscape is shifting, that slow, basic work will not sell itself, and that their audience now extends well beyond grant panels and journals. Suppose they are willing to integrate novel forms of communication as part of the research workflow.

How do we keep this kind of slow, basic work funded and visible without pretending it belongs in the same category as a wearable or health app?

Commercial startups like DeepMind are building products — they're not the right model for academic research. But they demonstrate something valuable: how to make technical work legible to nonspecialists. DeepMind’s AlphaFold was built on decades of academic structural biology, but it shipped as a named model with a public database, clear benchmarks, and demos that made the advance immediately accessible. Academic groups were publishing comparable advances through peer-reviewed papers — the format optimized for scientific credibility, not public accessibility. When labs relied solely on federal funding, that made sense. In a mixed funding landscape, it’s worth borrowing some of the accessibility that commercial incentives produce. Labs don’t need to become startups, but they can learn from how startups package work so outsiders can understand it quickly.

Focused Research Organizations and independent research institutes (collectively, alternative research organizations) offer a closer model for academic labs to follow — they do noncommercial basic research, just packaged differently. Arc Institute, Convergent Research’s FROs, and Arcadia Science are examples of organizations that pursue foundational questions about disease mechanisms, evolutionary biology, and neural circuits. They’re not building products. But they present their work through clear project pages, named programs, and accessible narratives that help program officers and philanthropic boards understand what they’re funding.

These institutions do not pretend to operate like traditional academic departments. They present themselves not as collections of individual labs pursuing separate grants but as unified organizations with shared infrastructure, centralized operations, and coherent communication styles. Arc Institute, for instance, launched with more than $650 million in committed philanthropic funding and a remit to “understand the root causes of complex diseases.” It organizes its work into five named technology centers — Multi-Omics, Genome Engineering, Cellular Models, Mammalian Models, and Computational — each with defined scientific goals and shared infrastructure. Investigators’ work is presented within these unified technology platforms, rather than as independent labs pursuing separate grants.

Arcadia Science is a $500 million “evolutionary biology company” founded by former UCSF and Dartmouth faculty to support long-horizon basic research outside the grant cycle. In 2025, Arcadia’s CEO, Seemay Chou, announced that the company would no longer fund traditional journal publications. Instead, Arcadia’s scientists publish their research as modular outputs, including partial results, methods, negative findings, and works-in-progress, complete with datasets, methods, and commentary on what worked and what didn’t. 

Convergent Research’s FROs typically receive $20-$50 million each to fund full-time teams spending three to seven years trying to solve a specific scientific bottleneck as coordinated projects with milestone-based road maps, rather than collections of independent labs. One of these FROs, E11 Bio, is developing technologies to map neural circuits at scale. Academic neuroscience labs pursue the same goal: understanding how neural circuits give rise to behavior and cognition. E11 packages this work as a technology platform with clear milestones and open datasets. They have explicitly stated they will release their methods for beta versions of their tools that other brain researchers can use in their own labs. The science itself isn’t fundamentally different from what happens in university neuroscience departments. What changes is how the work is organized, documented, and made accessible.

Arc Institute publicly states a crisp mission and model and organizes its work into clearly named programs and technology centers that outsiders can scan in a few minutes. Arcadia Science maintains a central research hub where ideas, data, methods, and even “on ice” projects are published as modular, browsable outputs, with meta pieces reflecting on how well the system works. They are careful, at their best, to distinguish between what is solid and what is speculative. 

A select number of academic labs have spent years building communications infrastructure and attracted remarkably diverse funding as a result. The Church Lab at Harvard Medical School maintains a detailed online register showing support from venture capital firms, corporate partnerships, and philanthropies. George Church attributes this success partly to his “radical open source” philosophy, stating that “by giving stuff away you actually make a better product, and have a better reputation.” This transparent, public-facing approach — including a tech transfer page and active media presence — has clearly helped attract investors and partners who value openness and collaborative innovation.

Jennifer Doudna’s Innovative Genomics Institute provides another example. Founded in 2015, IGI has since secured $70 million from The Audacious Project — a coalition including the Gates Foundation, MacKenzie Scott, and the Skoll Foundation. The institute employs a dedicated communications director and emphasizes public understanding through clear mission statements, a public-facing website, and TED presentations. This investment in communication infrastructure has helped IGI articulate its vision to funders beyond traditional academic grant systems.

The Broad Institute represents the most well-resourced academic example. It began with significant philanthropic commitments and has since expanded that base considerably, receiving over $1.5 billion in philanthropic commitments and cultivated partnerships with DeepMind, Google Research, Microsoft, and major pharmaceutical companies. Notably, they support a data sciences platform that develops tools, including domain-specific visualizations to communicate research findings. Their communication infrastructure — including public project pages, partnership pages, and donor impact stories — helps potential funders quickly understand and connect with their work. While most university labs won’t have access to Broad-scale resources, the principle scales: Treating communication as infrastructure that helps funders understand your work applies whether your budget is $1.5 billion or $150,000.

Unlike most university departments, these actors typically budget for communication as part of their core operations. They hire people whose job is to sit with researchers, understand the work, and help it show up in the right form for the right audiences. They bring in designers, editors, illustrators, and producers as colleagues. Convergent Research employs a dedicated head of communications and an events manager. Arcadia Science has an entire publishing team — including a science illustrator, director of publishing, and project manager — and created a free library of high-quality organism illustrations specifically for science communication. Arcadia also works with professional designers to create its brand identity and visual language. Academic labs have done this, too: The Broad Institute’s Pattern team has brought together designers and software engineers to create visualizations and interactive tools that make research data more accessible and interpretable. The Whitehead Institute has employed a director of strategic communications. The Salk Institute has a head of communications leading an award-winning communications team. Adam Marblestone, the CEO of Convergent Research, and Patrick Hsu, co-founder of Arc Institute, have been featured on numerous long-form podcasts. These are treatments of public engagement as part of the work, not a distraction from it. 

The obvious question: How do universities fund communication infrastructure when they're already stretched thin? The answer: If the investment pays off, they will find a way. Several major research institutions have begun expanding communication infrastructure as a direct response to demand from corporate partners, philanthropic funders, and translational programs. UC San Diego opened a Research Communications program after receiving a $225,000 grant to improve public-facing research translation. MIT developed its Communication Lab to support pitches, translational summaries, and project pages for groups seeking collaboration with The Engine or industry partners. The Alan Alda Center for Communicating Science at Stony Brook University was created specifically to help scientists develop skills in communicating to a range of audiences. On the philanthropic side, the European Research Council’s Public Engagement with Research Award, Wellcome’s Research Enrichment: Public Engagement grants, and the Sloan Foundation’s program for the Public Understanding of Science & Technology all provide supplemental funding specifically to researchers who engage “audiences outside of their domain,” “use public insights to develop their research,” and “advance public understanding” of scientific and technical work. 

If universities treat this shift as background noise, they effectively push their labs into a world where legibility matters — but they have no support in making themselves understood. Some PIs might assemble their own ad hoc communication stack: a half-finished lab website here, a well-produced talk there, an undergrad summer intern who happens to be good at diagrams. Others will do almost none of it and quietly miss out on opportunities that would have suited their work. A few charismatic labs can become de facto spokespeople for entire fields, not because the community agreed they should speak for everyone but because they are the only ones whose work arrives in a form outsiders can easily understand.

The appetite and capital exist for basic research outside federal agencies. Alternative research institutions attract this capital partly through how they package their work. If academic labs adopt similar communication practices, they can compete for these growing pools. But without some shared infrastructure, their ability to survive will depend more on individual improvisation and personality than on the quality or importance of the work itself. 

The infrastructure need not be elaborate. Universities can adopt a culture in which explanation is taken as seriously as experiment design, even with a smaller head count: A modest reallocation of funds at the level of a department or a school might support one or two translators whose job is to keep track of major projects, help researchers maintain a coherent public description of what they are doing, and coordinate with existing media offices when something consequential happens. 

Translators are to science communication what grant offices and institutional review boards are to administration. Nobody expects each PI to personally manage all compliance, contracts, and ethics paperwork for every project; those functions are treated as common infrastructure that makes research possible. Legibility deserves the same status.

Most labs already have the raw materials to explain their work: aims documents, slide decks, methods write-ups, figures. The challenge is assembling what exists into something coherent and findable. In practice, this may mean a persistent URL where the project lives, a plain-language description of the research question and approach, two to three key visuals that convey progress, and clear indicators of whether the work is exploratory, ongoing, or published. If a program officer, potential collaborator, or journalist can find and understand your work in five minutes, you've likely cleared the bar.

Finally, there is the question of training. For the last few decades, most graduate programs have treated science communication as either an elective hobby or something an academic learns on the job after tenure. In a system where multiple evaluators matter, that is increasingly a mismatch. Folding communication into graduate education does not mean turning PhD students into influencers. It can be as simple as expecting them, at some point in their training, to produce a clear, honest explanation of their project that someone outside their subfield can understand, and giving them access to the same tools and translators the institution uses elsewhere. 

None of this asks universities to abandon their basic research mandate. What it challenges is the reflex to treat communication as a threat to rigour. In practice, good translation work is often an extension of rigour: It forces academics to state their questions, methods, and uncertainties plainly, and it makes it easier for others to check, reuse, or contest what a lab has done. Money spent on clearer descriptions, simple tools, or a small translation team is not a dilution of “the real science,” so much as it is a way of deciding how the work can be strategically repackaged to reach a new layer of students, collaborators, and nonfederal funders. 

The choice is not, and has never been, between purity and PR. It is between treating legibility as part of the shared infrastructure that keeps basic and long-term research alive, and leaving it to whoever already has the resources to define how the rest of the system looks from the outside.

In one, universities continue treating communication as ornamental and watch flexible capital, talent, and narrative authority drift toward institutions that package their work legibly — leaving slow, basic research precarious and invisible. 

In the other, they accept that they operate in a mixed funding ecosystem and build enough communicative infrastructure to stay legible alongside FROs, AI labs, and research companies, without flattening their work into products. The wobble in federal funding exposed what it means to stand an entire research system on a single pillar. Diversification is already happening. 

Universities now need to decide whether careful work will remain visible and fundable, or if it will quietly recede into disciplinary pockets that potential funders can neither see nor support.