Validating and reproducing science

1.1 The Rational Basis for Skepticism

It is essential to acknowledge that public skepticism toward scientific institutions is often well-founded, rooted in historical and ongoing failures.

The influence of corporate interests on research outcomes is a significant, salient concern. As claimed by critics like Goldacre (2012), the pharmaceutical industry often controls the design, execution, and publication of research on its own products. This leads to pervasive publication bias, where negative results are suppressed, distorting the evidence base.

At the moment, people often point to the opioid crisis. Pharmaceutical companies deceptively marketed opioids as non-addictive, citing flawed or misrepresented literature. This demonstrates how industry interests can corrupt the validation process with catastrophic societal consequences. As historians Oreskes and Conway argued in Oreskes and Conway (2022), the strategy of manufacturing uncertainty to delay regulation—first perfected by the tobacco industry—remains a standard playbook for industries facing inconvenient scientific truths.

Historical abuses have also left scars, particularly among marginalized communities. For example, the infamous “Tuskegee Study of Untreated Syphilis” (1932–1972), where U.S. Public Health Service researchers withheld known treatments from Black men, remains a symbol of institutional betrayal. Research suggests the disclosure of this study contributed to lasting medical mistrust and measurable negative health outcomes (Alsan and Wanamaker 2018). (cf Nazi experiments, Henrietta Lacks, etc.)

These examples highlight that skepticism is often directed not at the scientific method itself, but at the institutions charged with implementing it.

1.2 Failures of Science Communication

The credibility problem is exacerbated by the way science is communicated. The incentives driving academic careers (novelty, impact, funding) often bleed into communication strategies, prioritizing hype over substance.

This is reinforced by the internal dynamics of science itself. A recent study found that papers that failed to replicate are cited significantly more often than those that succeeded (Serra-Garcia and Gneezy 2021). This suggests the system favours “interesting” or surprising results, regardless of their validity.

This tendency toward novelty creates a “hype pipeline” (Caulfield and Condit 2012). Academic pressures lead to exaggerated claims in papers, university press offices amplify those claims to attract media attention, and journalists prioritize sensational findings. Studies have shown that press releases frequently exaggerate the importance of findings while minimizing limitations (Sumner et al. 2014). This constant stream of “breakthroughs” sets the public up for disillusionment.

Furthermore, institutions sometimes prioritize messaging that encourages desired public behaviour over complete transparency. As sociologist Zeynep Tufekci argued, when public health authorities appear opaque or misleading (as seen in debates during the COVID-19 pandemic), it damages institutional credibility. Paternalistic messaging often backfires, as the public recognize inconsistency and interprets it as incompetence or deception.

1.3 Funding pressure

TBD — I really need to write this one up. It’s of general interest; one day I’ll be permitted to mention my personal experiences with industry pressure.

• Industry Sponsorship bias | Catalog of Bias
• The Cost of Manipulation: The Irresponsible Abuse of Technological Opacity in the Pharmaceutical Industry

1.4 Do Your Own Research

In this environment of justified skepticism and communication failures, the rise of the “Do Your Own Research” (DYOR) phenomenon is understandable. It reflects a collapse of faith in established mediating institutions.

While the impulse behind DYOR frequently stems from legitimate critiques—the recognition that published research can be biased and that experts can be conflicted—the method doesn’t fix the institutional failures of mainstream science.

DYOR mistakes access to information (e.g., reading preprints or isolated studies) for the rigorous, collective process required to evaluate and synthesize it. An individual cannot solve the problems of P-hacking or publication bias alone; these require institutional reform.

Indeed, someone close to me died needlessly and slowly because they trusted their own research over medical advice. This shit is serious. At the same time, other people close to me have been ignored or shunned for questioning established narratives when they have direct evidence of adverse effects from medical treatments.

As researcher danah boyd argues, we are living through a crisis of epistemology—a disagreement about how we know whether something is true. In the digital age, DYOR often involves navigating an information landscape strategically polluted by media manipulation. Boyd warns that simplistic approaches to critical thinking—teaching people to doubt without providing robust frameworks for understanding—can backfire, making individuals more vulnerable to conspiratorial thinking.

The DYOR movement seems great at surfacing flaws in existing practice. It’s less good at producing calibrated criticism and synthesis, and it doesn’t offer a scalable mechanism for reliable scientific self-correction. It often leads individuals into alternative epistemic communities that lack rigorous validation processes.

The challenge for the scientific community is not to demand blind trust, but to earn it by reforming the internal mechanisms of validation and communication, making the enterprise demonstrably rigorous, transparent, and accountable.

2.1 The Evolution and Limits of Peer Review

It is often assumed that peer review is a timeless feature of the scientific process, but its current dominance is a relatively recent invention.

Baldwin (2018) notes:

Throughout the nineteenth century and into much of the twentieth, external referee reports were considered an optional part of journal editing or grant making. The idea that refereeing is a requirement for scientific legitimacy seems to have arisen first in the Cold War United States. In the 1970s, in the wake of a series of attacks on scientific funding, American scientists faced a dilemma: there was increasing pressure for science to be accountable to those who funded it, but scientists wanted to ensure their continuing influence over funding decisions. Scientists and their supporters cast expert refereeing—or “peer review,” as it was increasingly called—as the crucial process that ensured the credibility of science as a whole.

This history suggests peer review evolved as much to maintain professional autonomy as to ensure empirical rigour. Today its effectiveness is hotly debated: critics say it’s slow, biased, and often fails to detect errors. Adam Mastroianni argues in The rise and fall of peer review that the system often fails to deliver on its promises.

For a deeper dive, see peer review.

• Related question: How do we discover research to send for peer review?

2.2 Gatekeeping and Institutional Culture

The validation process inherently involves gatekeeping. Institutions decide who can participate and which standards must be met. That helps explain why we’re perennially dissatisfied with academic culture.

Thomas Basbøll observes the tension between innovation and tradition:

It is commonplace today to talk about “knowledge production” and the university as a site of innovation. But the institution was never designed to “produce” something nor even to be especially innovative. Its function was to conserve what we know. It just happens to be in the nature of knowledge that it cannot be conserved if it does not grow.

This conservatism can make the validation process hostile to radically new ideas. (See Andrew Marzoni, Academia is a cult).

Friction in the traditional system leads some researchers to bypass that system entirely. Stephen Wolfram’s recent approach provides a telling example, as detailed by Adam Becker (Wolfram’s latest positioning):

So why did Wolfram announce his ideas this way? Why not go the traditional route? “I don’t really believe in anonymous peer review,” he says. “I think it’s corrupt. It’s all a giant story of somewhat corrupt gaming, I would say. I think it’s sort of inevitable that happens with these very large systems. It’s a pity”.

2.3 Grants (TODO)

How does the funding mechanism shape what gets validated? Does the grant review process prioritize novelty over rigour?

3.1 Publication Bias and the File Drawer

Journals prioritize novel, statistically significant findings. This creates a “publication sieve”.

Null results, or studies that fail to find an effect, are often relegated to the “file drawer”—they’re never published.

This leads to a distorted view of reality in the published literature. If 20 researchers each test a hypothesis, and only the one who (by chance) achieves \(P\leq 0.05\) publishes their result, the literature will suggest a strong effect where none exists. This is essentially multiple testing across an entire scientific field.

3.2 P-Hacking and Researcher Degrees of Freedom

For individual researchers, the pressure to publish significant results incentivizes “researcher degrees of freedom”—the many choices researchers make during data analysis that can be exploited (consciously or unconsciously) to push a result below the significance threshold. This is often called P-hacking.

We can detect the signature of systematic p-hacking in the literature. See Uri Simonsohn’s work on the p-curve.

5.1 Transparency and Pre-registration

A major push is toward greater transparency to reduce researchers’ degrees of freedom. This includes open notebook science.

A key intervention is pre-registration, in which researchers specify their hypotheses, methods, and analysis plan before collecting data.

Tom Stafford’s 2-minute guide to experiment pre-registration explains the options:

• you can use the Open Science Framework to timestamp and archive a pre-registration, so you can prove you made a prediction ahead of time.
• you can visit AsPredicted.org which provides a form to complete…
• “Registered Reports”: more and more journals are committing to published pre-registered studies. They review the method and analysis plan before data collection and agree to publish once the results are in (however they turn out).

Registered reports are particularly useful for decoupling the publication decision from the results, potentially solving the incentive problem. However, they require more upfront work and may slow down the research process, which can clash with existing career incentives.

5.2 Review