By Pierre-Alain Bruchez - September 7, 2025

The crisis in science that undermines research is widely underestimated, largely because irreproducible results, ideological bias, conflicts of interest and fraud are typically discussed in isolation - without recognizing their cumulative impact and shared roots.

Scientists alone cannot resolve this. Citizen scrutiny is essential. But first, citizens should be informed.

SCIENTIFIC FRAUD HAS BECOME INDUSTRIALIZED

Fraud is, by nature, elusive. Although improved detection tools (e.g. image duplication analysis) may struggle to catch current fraud, given how quickly fraudsters adapt, they nonetheless provide valuable insights into past misconduct.

Particularly troubling is the fact that fraud is no longer confined to isolated individuals, but is increasingly perpetrated by organized networks (see Richardson et al., The entities enabling scientific fraud at scale are large, resilient, and growing rapidly). The presence of fraudsters should not discredit an entire profession, but it remains the duty of each profession to expose and expel them.

THE REPLICATION CRISIS

Many published results cannot be reproduced: this is the replication crisis. This is not necessarily due to fraud. In many fields, results are statistical: they can also be due to chance. For example, if you want to know whether a die is rigged, you roll it many times. If one face appears disproportionately often, you conclude it is biased.

However, it is not impossible that the die is fair and the result merely random. Typically, a result is accepted if the probability that it occurred by chance is less than an arbitrarily chosen threshold of 5% (although in some fields, such as particle physics, the threshold is set much lower).

Thus, in principle, one would expect 5% of statistical results to be false. In reality, it is much higher, especially due to publication bias. Spectacular results are more likely to be published, although they are also more likely to be statistical flukes.

As early as 2005, John Ioannidis demonstrated in his landmark paper Why Most Published Research Findings Are False that the proportion of false statistical results is far greater than 5%. A large-scale replication project in psychology confirmed that only a minority of results could be replicated. Oncology and biomedical research also show high replication failure rates. Surprisingly, no meta-study compares replication failure rates across disciplines. Why not launch a massive replication project across all fields?

The replication crisis has been recognized for years and is still not overcome. Yet, in principle, it could be drastically reduced quickly. Solutions exist. Journals must demand transparency: full data and methodology disclosure to enable replication. Methods and hypotheses should be preregistered to prevent post hoc hypothesis fishing. Articles should be accepted based on the relevance of the question and methodological rigor, not the results.

This reduces the incentive and ability to chase statistically spurious findings. The Center for Open Science offers tools to support this, but they are used in only a minority of publications.

Universities should replicate more studies, starting with the most important ones (to test the foundations of the discipline) and randomly among newly published results (to encourage researchers to be more rigorous by increasing the risk that their study will be checked). Students would gain valuable experience while providing a highly useful service. Replication is a powerful pedagogical tool.

Initiatives like those from the Center for Open Science promote replication but still operate on a modest scale compared to global research output. Replication status should be readily available when consulting a study, and journalists should systematically report it. Safeguards must also be implemented to prevent collusive validation fraud, where researchers complacently reproduce each other’s findings. All of this should be put into effect swiftly.

It is encouraging to see the growing number of initiatives aimed at tackling the replication crisis. Beyond the Center for Open Science mentioned earlier, notable examples include the Institute for Replication, Open Science NL, and the NIH’s Replication Initiative. Yet the impact of these initiatives remains modest compared to the magnitude of the replication crisis itself.

The scientific community’s lack of urgency in addressing the replication crisis is even more troubling than the crisis itself. Inertia? The deeper issue is that for too many scientists, truth-seeking is no longer the main priority. This is vividly illustrated by their increasing submission to authoritarian ideologies.

Ideological capture in universities reveals a deprioritization of truth-seeking, which also hinders efforts to overcome the replication crisis. Conversely, ideological capture has taken hold on already weakened ground—as illustrated by the replication crisis itself.

IDEOLOGICAL CAPTURE

Major universities, especially in the U.S., have been captured by authoritarian ideologies. Willingly or not, researchers have often repeated claims they know are false. To expose this ideological grip, Peter Boghossian, James Lindsay, and Helen Pluckrose managed to publish deliberately absurd but politically correct papers (they present their work in a video). Boghossian had to resign from his university and co-founded the University of Austin, which positions itself as one of the few alternatives to universities captured by wokeism.

Another alternative is the Peterson Academy, created by Jordan Peterson. He famously refused to . . .

[SNIP]