28: Published Does Not Mean Proven - The Shocking Truth About Scientific Corrections with Dr. Rob Karlinsey

What if some of the research shaping everyday products is quietly flawed, and the system designed to correct it often chooses not to?

In this episode of Dental Formulator’s Playbook, Dr. Rob and co-host Tami take a deep, unfiltered look at scientific ethics and research accountability. Prompted by recent survey data and real-world case studies, the conversation explores how flawed or falsified research can enter the scientific record and why it so often remains there uncorrected.

Rather than focusing on isolated mistakes, this episode examines systemic behaviors. These include informal admissions, delayed accountability, and the reluctance to issue corrections or retractions even when problems are known. From high-profile superconductivity claims to modern AI-driven research, Dr. Rob explains why transparency, replication, and proper testing models are essential, especially when scientific papers are used to justify claims in oral care products and other consumer-facing technologies.

This discussion challenges listeners to think more critically about what “published” really means and how to evaluate research beyond headlines and abstracts.

Key Topics Covered

Findings from a published survey examining how often chemists knowingly falsify information in scientific papers
What falsification can look like beyond simple errors, including selective data presentation and procedural shortcuts
Why many researchers avoid formal corrections and how those decisions are commonly rationalized
Informal and non-public ways errors are addressed instead of correcting the scientific record
How scrutiny can occur outside traditional journals through independent review and public forums
A detailed case study involving high-profile superconductivity claims and failed replication attempts
Why highly publicized fields like AI attract attention, while similar ethics risks exist across all areas of science
How Dr. Rob evaluates whether research is trustworthy, relevant, and clinically meaningful
The role of appropriate models, controls, and independent replication in supporting oral care product claims

Highlights and Takeaways

Published does not mean proven. Peer-reviewed publication alone is not a guarantee of accuracy or reliability.
The scientific record often remains uncorrected. Many known issues are never formally addressed through corrections or retractions.
Replication is essential. When independent groups cannot reproduce results, confidence in those findings should decrease.
Models and controls matter. Meaningful positive and negative controls are critical for interpreting results responsibly.
Bias influences what gets promoted. Financial incentives, career pressure, and marketing interests can shape scientific narratives.

Reference Links

Karlinsey & Karlinsey – Recently Published Research

Methods and Protocols paper examining formulation science and measurement models:

https://doi.org/10.3390/mps9010010

Scientific Accountability and Publication Ethics

Survey findings discussed by Accountability in Research and ACS Chemical & Engineering News on deliberate inclusion of misleading information in published papers:

https://doi.org/10.1080/08989621.2025.2564106

https://cen.acs.org/policy/publishing/One-five-chemists-deliberately-added/103/web/2025/10

Retracted Superconductivity Papers (Science and Nature)

Commentary and official retraction notices for eight papers in Science and one paper in Nature:

https://www.science.org/doi/10.1126/science.298.5595.961b

https://www.nature.com/articles/nature01467

University of Rochester Superconductivity Retractions (Nature 2020 & 2023)

Editorial notices addressing retracted claims of room-temperature superconductivity:

https://www.nature.com/articles/s41586-022-05294-9

https://www.nature.com/articles/s41586-023-05742-0.epdf

Fraudulent AI Research Activity at MIT

Wall Street Journal reporting and official MIT statement addressing inaccurate research records and ethics concerns: