The Slow Research Promise Was Never Meant for You

Who gets to slow down?

In spring 2025, a “slow research” initiative circulated among AI scholars. One hundred nineteen teachers and researchers signed it. They pledged not to select students by paper count. The wording stayed modest. It targeted undergraduates. It did not call for sitting on a cold bench. It did not claim moral high ground. It said something practical: a CV stacked with papers can hurt an applicant with teachers who care about taste.

A job seeker saw the news and paused, coffee in hand. He had spent several hundred thousand yuan of his family’s money on a PhD. He was doing good research. Two years after graduation, he had no decent job and no stable income. He asked: Is this the slow research you wanted?

The question points to a structure.

I. The asymmetry of cost

Zak Witkower wrote in Nature Human Behaviour that the slow science movement, despite good intentions, imposes costs only mid- and late-career researchers can afford. Tenure and funding protect them. For early-career scholars, output is survival. Early-career researchers who slow down risk self-sabotage. Their safe mid-career and senior peers can remain undisturbed. They can keep romanticizing a model of scientific life only they can afford.

This passage did not use “truly” or “worth deep reflection.” It stated a fact: slowness is an economic choice. It requires a stable salary, security from proving output, and funding that does not depend on short-term results. Only people past the evaluation threshold have those conditions.

A study in an oceanography laboratory put it more concisely: early-career researchers are screened by project and paper output. Senior researchers hold influence in institutions.

A tenured professor who says, “I’m only publishing seven papers this year,” is running an experiment. A fourth-year PhD student who says, “I’m only publishing one paper this year,” is gambling with graduation and job prospects. Both say “slow.” They do not mean the same thing.

II. The price of timing

“Most work is only valuable if it comes out at that timing. If it doesn’t come out, it’s not valuable.” The sentence is correct. It needs one question: who manufactures timing?

A 2026 OECD report said university ranking data can mislead decision-makers. It promotes narrow, easily manipulated indicators. Those indicators divert time and resources from more meaningful academic work. They are increasingly driven by profit motives that do not always align with science. At the individual level, evaluation usually includes input metrics (funding numbers and amounts) and output metrics (publication and citation counts). Journal impact factors remain widely used in faculty evaluation. At top institutions, publication in high-impact journals is often a key requirement for hiring and promotion.

Price makes this concrete. In China, each Web of Science paper can earn cash rewards from $30 to $165,000. In South Africa, a paper in an accredited journal can bring up to 120,000 rand in government bonus. After the system began in 2005, the country’s research output grew from 4,063 papers to 25,371 in 2018.

When academic output gets institutionally priced, speed becomes a tradable asset. A paper published in a hot window gets more citations, higher visibility, and an edge in the next funding round. Miss the window, and the same work can go unnoticed. The work did not get worse. Attention moved.

A scholar wrote after applying for a talent program in Spain: In the system, a candidate with 50 papers competes with one who has 10. The comparison is cruel. Yet a CV with dozens of publications has little value to me if those papers did not contribute to the field. He continued: When institutions still judge by CV length, slow science is a risky strategy for early-career researchers fighting for stability.

III. The symbolism of one scholar’s output cut

In 2025, Nature published an article. A scholar announced he would cut his annual output in half, from about 14 papers to 7, to double the time he invests in quality. The article cited data: PubMed indexed 1.2 million articles in 2014. By 2024, it indexed more than 1.7 million. This overproduction makes thorough reading or review impossible. It lowers the overall quality of science.

The action was symbolically powerful. It publicly admitted something almost everyone knows and few say outright: more papers do not necessarily mean better science. It said “I refuse to participate” with a number.

The scholar is tenured. His cut is safe. He may gain reputation for courage. People without tenure face different consequences if they imitate him.

A critical article on the LSE blog pointed to a deeper problem. Advocates of slow scholarship demand the abolition, dismantling, and destruction of neoliberalization. Their prescriptions are often individual: publish less, read more, refuse acceleration. There is a logical break. If the problem is structural, personal lifestyle changes cannot solve it.

IV. What institutions can do

Some changes do not stay at the level of declarations.

Shanghai Chuangzhi Institute offers a case. This AI talent-training institution, founded in 2024, does not use conventional exams or past GPAs for doctoral selection. It runs quarterly practical training camps and selects students on a rolling basis. It assesses them in frontline project work. One doctoral graduate called it “a rare deceleration lane.” The institute gave students freedom from utilitarian, short-term metrics.

The case shows that selecting students without counting papers is possible. The premise is that institutions will bear the cost of changing the screening method. They must organize training camps, design evaluation criteria, and train reviewers. The obstacle is resources. Moral commitment is not enough. Chuangzhi Institute has funding and institutional space for an alternative screening mechanism.

It is a special case. It was born under a particular policy window and resource conditions. Most universities cannot copy it. Without fundamental change in the evaluation system, an individual institution’s innovation looks like a carefully maintained experimental zone, not a scalable solution.

China’s “breaking the five onlys” reform offers another observation. Official policy requires no more counting papers, projects, or awards in professional title review. It calls for representative works evaluation and focuses on impact and contribution. Some pilot units have adopted 5- to 10-year evaluation cycles. Universities such as Harbin Institute of Technology are exploring a high-level representative works system.

A 2025 assessment found that research institutions have generally removed explicit “four onlys” provisions from talent evaluation systems. In practice, professional title review, term assessment, and talent recruitment still rely mainly on paper count, project level, and award rank. The reform direction is correct. The conversion from policy text to practice is slow. Representative works evaluation also faces an operational problem: who defines a representative work? Without accepted standards, evaluators’ subjective judgment can be less transparent than quantitative metrics. It can be more easily influenced by networks.

V. What to do in the cracks

For people without a safety net, what is a realistic way out?

There is no way out that does not exact a price. But one can separate survival output from identity work.

Survival output is work that must meet hard thresholds: papers for graduation, results for job hunting, numbers for evaluation. For this part, prioritize efficiency. Do not chase perfection. Stop when done. This is resource management: use the least time and energy to satisfy minimum requirements and preserve space for important work.

Identity work is the part not completely ruled by the evaluation clock. It can be a small question you are curious about, a period when you can read and try things freely, or thinking not aimed at publication. This slowness produces no papers and adds no length to a CV. It protects your core as a researcher from being completely instrumentalized. It does not change the system. It lets you keep something inside the system that the system does not define.

This is a pragmatic strategy for preserving as much academic vitality as possible amid structural injustice. Can we make the market slow down? The answer is no. But we can keep the market from completely defining why we do research.

The scholar who applied for a talent program in Spain wrote at the end of his article: Unless we hit the brakes, the system will derail. Before incentives change, insisting that quality matters more than quantity is still an act of resistance. The danger is that resistance sounds romantic. Without a safety net, choosing quality first can cost you the qualification to keep doing research. Resistance may mean participating at the lowest cost while protecting yourself from being completely swallowed.

Finally, the person who spent hundreds of thousands of his family’s money still has no decent job. He opens his laptop and sees a rejection letter. He closes the email and waters the pothos. Water seeps from the bottom of the pot and drips onto the desk. He wipes it with his sleeve. Then he opens a blank document and starts writing a question no one asked him to write.