Scientific research has big problems, and it's getting worse

Scientific research is being distorted by incentives that reward output over truth, and the damage is spreading—from rare fraud to increasingly organized scams and, most importantly, a system-wide push toward low-value, high-citation work. While misconduct and outright fabrication grab headlines, they remain comparatively uncommon. More consequential is the “paper mill” ecosystem and other coordinated operations that manufacture fake authorships, citations, and even images—sometimes using AI to generate papers and planting fabricated work online to build the appearance of real researchers. These schemes are expanding beyond their earlier geographic strongholds and are likely to accelerate as AI tools improve.

The deeper problem is why the market for fake science exists at all. Buying authorship and citations can produce real career and funding returns: grants, jobs, and institutional prestige can follow from the appearance of credible, frequently cited research. Even when work isn’t outright fake, the same incentive structure nudges researchers toward strategies that maximize citations rather than scientific value. The easiest route to visibility, she argues, is producing “useless garbage” that colleagues can approve and that the public can’t easily evaluate—especially in fields where few outsiders can judge technical claims. This doesn’t require researchers to consciously decide to waste their careers; it can emerge as the rational “winning strategy” inside an evaluation system that rewards publication volume and impact metrics.

That incentive trap helps explain why progress can slow even when effort remains high. Researchers face constant pressure to publish more, in high-impact journals, and to generate citations. Examples include senior faculty steering younger scientists toward crowded topics simply because they are more likely to be cited, department-level reprimands for insufficient publication output, and grant proposals padded with sections that are “nonsense” in practice. The result is a race to the bottom: riskier, uncertain breakthroughs become harder to fund, while safe, incremental, or strategically framed work becomes the default.

Communities often fail to self-correct. In psychology and parts of sociology, flawed approaches to statistical significance persisted for decades despite being widely known; the cost of changing methods—making publication harder—helped keep the status quo in place until public scrutiny forced attention. In physics, the issue can take the form of “mathematical fiction” or “fairytale physics,” where elaborate formalism substitutes for empirical grounding. Similar patterns show up elsewhere: mislabelled cell lines in biomedicine, questionable antibody reagents, and the limited translational value of mouse models—problems that persist because they are cheap, easy, and familiar.

Economists and scholars of research systems point to structural causes: universities rely on PhD students and postdocs as inexpensive labor to generate paper output, and faculty incentives discourage risk. A faculty member’s next grant can hinge on whether proposed work looks like a sure bet, making uncertainty a funding liability. The net effect is declining “return on investment” in science—more papers, less progress.

The proposed remedy is not simply better policing of fraud. It requires deliberate mechanisms that reduce sociological and economic pressure from shaping what gets studied and rewarded. Without consequences for long-running wrong predictions and without changing how success is measured, the system keeps selecting for bad science. The bottom line: fraud matters, scams are growing, but the incentive structure that makes low-value research the rational choice is the central reason scientific progress is stalling—and worsening.