Back in 2016, Vox asked 270 scientists to name the biggest problems facing science. Many of them agreed that the constant search for funding, brought on by the increasingly competitive grant system, serves as one of the biggest barriers to scientific progress.

Even though we have more scientists throwing more time and resources at projects, we seem to be blocked on big questions — like how to help people live healthier for longer — and that has major real-world impacts.

Grants are funds given to researchers by the government or private organizations, ranging from tens to hundreds of thousands of dollars earmarked for a specific project. Most grant applications are very competitive. Only about 20 percent of applications for research project grants at the National Institutes of Health (NIH), which funds the vast majority of biomedical research in the US, are successful.

If you do get a grant, they usually expire after a few years — far less time than it normally takes to make groundbreaking discoveries. And most grants, even the most prestigious ones, don’t provide enough money to keep a lab running on their own.

Between the endless cycle of grant applications and the constant turnover of early-career researchers in labs, pushing science forward is slow at best and Sisyphean at worst.

In other words, science has a short-term memory problem — but there are steps funding agencies can take to make it better.

Principal investigators — often tenure-track university professors — doing academic research in the US are responsible not only for running their own lab, but also for funding it. That includes the costs of running experiments, keeping the lights on, hiring other scientists, and often covering their own salary, too. In this way, investigators are more like entrepreneurs than employees, running their labs like a small-business owner.

In the US, basic science research, studying how the world works for the sake of expanding knowledge, is mostly funded by the federal government. The NIH funds the vast majority of biomedical research, and the National Science Foundation (NSF) funds other sciences, like astrophysics, geology, and genetics. The Advanced Research Projects Agency for Health (ARPA-H) also funds some biomedical research, and the Defense Advanced Research Projects Agency (DARPA) funds technology development for the military, some of which finds uses in the civilian world, like the internet.

The grant application system worked well a few decades ago, when over half of submitted grants were funded. But today, we have more scientists — especially young ones — and less money, once inflation is taken into account. Getting a grant is harder than ever, scientists I spoke with said. What ends up happening is that principal investigators are forced to spend more of their time writing grant applications — which often take dozens of hours each — than actually doing the science they were trained for. Because funding is so competitive, applicants increasingly have to twist their research proposals to align with whoever will give them money. A lab interested in studying how cells communicate with each other, for example, may spin it as a study of cancer, heart disease, or depression to convince the NIH that its project is worth funding.

Federal agencies generally fund specific projects, and require scientists to provide regular progress updates. Some of the best science happens when experiments lead researchers in unexpected directions, but grantees generally need to stick with the specific aims listed in their application or risk having their funding taken away — even if the first few days of an experiment suggest things won’t go as planned.

This system leaves principal investigators constantly scrambling to plug holes in their patchwork of funding. In her first year as a tenure-track professor, Jennifer Garrison, now a reproductive longevity researcher at the Buck Institute, applied for 45 grants to get her lab off the ground. “I’m so highly trained and specialized,” she told me. “The fact that I spend the majority of my time on administrative paperwork is ridiculous.”

For the most part, the principal investigators applying for grants aren’t doing science — their graduate students and postdoctoral fellows are. While professors are teaching, doing administrative paperwork, and managing students, their early-career trainees are the ones who conduct the experiments and analyze data.

Since they do the bulk of the intellectual and physical labor, these younger scientists are usually the lead authors of their lab’s publications. In smaller research groups, a grad student may be the only one who fully understands their project.

In some ways, this system works for universities. With most annual stipends falling short of $40,000, “Young researchers are highly trained but relatively inexpensive sources of labor for faculty,” then-graduate researcher Laura Weingartner told Vox in 2016.

Grad students and postdocs are cheap, but they’re also transient. It takes an average of six years to earn a PhD, with only about three to five of those years devoted to research in a specific lab. This time constraint forces trainees to choose projects that can be wrapped up by the time they graduate, but science, especially groundbreaking science, rarely fits into a three- to five-year window. CRISPR, for instance, was first characterized in the ’90s — 20 years before it was first used for gene editing.

Trainees generally try to publish their findings by the time they leave, or pass ownership along to someone they have trained to take the wheel. The pressure to squeeze exciting, publishable data from a single PhD thesis project forces many inexperienced scientists into roles they can’t realistically fulfill. Many people (admittedly, myself included, as a burnt-out UC Berkeley neuroscience graduate student) wind up leaving a trail of unfinished experiments behind when they leave academia — and have no formal obligation to complete them.

When the bulk of your workforce is underpaid, burning out, and constantly turning over, it creates a continuity problem. When one person leaves, they often take a bunch of institutional knowledge with them. Ideally, research groups would have at least one or two senior scientists — with as much training as a tenured professor — working in the lab to run experiments, mentor newer scientists, and serve as a stable source of expertise as other researchers come and go.

One major barrier here: Paying a highly trained scientist enough to compete with six-figure industry jobs costs far more than a single federal grant can provide. One $250,000/year NIH R01 — the primary grant awarded to scientists for research projects — barely funds one person’s salary and benefits. While the NIH has specialized funding that students, postdocs, junior faculty, and other trainees can apply for to pay their own wages, funding opportunities for senior scientists are limited. “It’s just not feasible to pay for a senior scientist role unless you have an insane amount of other support,” Garrison told me.

Funding scientists themselves, rather than the experiments they say they’ll do, helps — and we already have some evidence to prove it.

The Howard Hughes Medical Institute (HHMI) has a funding model worth replicating. It is driven by a “people, not projects” philosophy, granting scientists many years worth of money, without tying them down to specific projects. Grantees continue working at their home institution, but they — along with their postdocs — become employees of HHMI, which pays their salary and benefits.

HHMI reportedly provides enough funding to operate a small- to medium-sized lab without requiring any extra grants. The idea is that if investigators are simply given enough money to do their jobs, they can redirect all their wasted grant application time toward actually doing science. It’s no coincidence that over 30 HHMI-funded scientists have won Nobel Prizes in the past 50 years.

The Arc Institute, a new nonprofit collaboration among research giants Stanford, UC Berkeley, and UC San Francisco, also provides investigators and their labs with renewable eight-year “no-strings-attached” grants. Arc aims to give scientists the freedom and resources to do the slow, unsexy work of developing better research tools — something crucial to science but unappealing to scientific journals (and scientists who need to publish stuff to earn more funding).

Operating Arc is expensive, and the funding model currently relies on donations from philanthropists and tech billionaires. Arc supports eight labs so far, and hopes to expand to no more than 350 scientists someday — far short of the 50,000-some biomedical researchers applying for grants every year.

For now, institutional experiments like Arc are just that: experiments. They’re betting that scientists who feel invigorated, creative, and unburdened will be better equipped to take the risks required to make big discoveries.

Building brand-new institutions isn’t the only way to break the cycle of short-term, short-sighted projects in biomedical research. Anything that makes it financially easier for investigators to keep their labs running will help. Universities could pay the salaries of their employees directly, rather than making investigators find money for their trainees themselves. Federal funding agencies could also make grants bigger to match the level of inflation — but Congress is unlikely to approve that kind of spending.

Science might also benefit from having fewer, better-paid scientists in long-term positions, rather than relying on the labor of underpaid, under-equipped trainees. “I think it would be better to have fewer scientists doing real, deep work than what we have now,” Garrison said.

It’s not that scientists aren’t capable of creative, exciting, ambitious work — they’ve just been forced to bend to a grant system that favors short, risk-averse projects. And if the grant system changes, odds are science will too.