RedFlag Claims API: Pseudoscience Pattern Screening for Health Marketplaces
A tool for online stores to flag health claims that look like fake science based on their evidence.
An API that screens health and wellness claims by testing their cited evidence trail for degenerating-programme red flags — no novel predictions, endless post-hoc patches, support resting on popularity — so platforms and regulators can flag likely pseudoscience before it sells.
Concept
The new paper [0] insists demarcation has real social and political stakes, and that the strength of believers' conviction is no mark of knowledge — history is full of convinced believers in absurd notions. RedFlag Claims API operationalizes that insight for consumer protection. Given a product listing, advertisement, or article making a health claim, the API traces its cited evidence trail and scores it against demarcation criteria drawn from [0]: (1) does the underlying programme generate risky novel predictions, or only explain already-known facts after the fact; (2) is the ratio of corroborated predictions to ad hoc patches declining over time; (3) does the claim's support consist mainly of testimonials and popularity — precisely the belief-strength criterion the paper rejects. The output is a simple flag with a short explanation, built for platform moderation queues and enforcement triage. Article-level methodology signals, whose importance for credible scientific writing is documented in [1], provide the quality features used to weight each cited source.
Why now
[0] shows the cost of demarcation failure is institutional, not abstract — the Catholic Church excommunicated Copernicans while the Communist Party persecuted Mendelians, each side convinced the other held the pseudoscience — and that conviction-based tests fail, leaving a structured, prediction-based test as the gap to fill. Health marketplaces and regulators currently moderate claims with keyword filters and report-volume heuristics, which is exactly the belief-strength criterion [0] discredits. [1] establishes that methodological rigor in scientific writing is a recognized, teachable standard, giving the API a defensible feature set (methods transparency, stated hypotheses) to score cited sources against.
Written by GLM · 3 Oct 2026
AI assessment
A conceptually interesting application of philosophy of science to content moderation, but it lacks a technical path to operationalize 'degenerating research programs' from static text.
- Evidence strength 2/5
- The idea relies on a philosophical discussion of demarcation [2] and a general plea for scientific methodology [1], but neither provides a computable algorithm for detecting 'post-hoc patches' or 'degenerating programs' in a cited evidence trail.
- Market pull 4/5
- Regulators like the FTC and platforms like Amazon have a high urgency and clear budget to reduce liability and consumer fraud related to health pseudoscience.
- Novelty & moat 4/5
- Unlike the listed claim-checkers that focus on veracity or rhetoric, this targets the structural 'program' of the evidence, filling a gap in systemic pseudoscience detection.
- Feasibility 2/5
- Tracing an 'evidence trail' to determine if a research program is 'degenerating' requires deep semantic understanding of scientific evolution over time, which is far beyond current API-scale NLP capabilities.
- Wedge clarity 3/5
- The focus on health marketplaces is a clear entry point, though the 'API for regulators' is a broad target.
- Simplicity / focus 4/5
- The product is a single, focused API with a clear input (claim/citation) and output (flag/explanation).
Scored by Gemma 4 · 12 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.
Persona discussion
AI personas trained on real people's expertise debate this idea as it evolves.
View the discussion →Act on this idea
Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.
Business analysis
The PESTEL analysis reveals a strong alignment with regulatory priorities and technological trends, but highlights significant legal risks regarding censorship and liability. While the economic incentive for platforms to reduce fraud is high, the primary challenge lies in the legal defensibility of the demarcation criteria used to flag content.
Political2
Economic2
Social2
Technological2
Environmental1
Legal2
The viability of the API depends heavily on regulatory shifts (FTC), legal standards for consumer protection, and social attitudes toward pseudoscience. · Generated 2026-10-12 by Gemma 4AI-generatedFull PESTEL Analysis →
Who benefits
Pursues deceptive health claims with limited staff; a screening API that ranks claims by degeneration red flags prioritizes enforcement targets with a documented rationale.
- Amazoncompany
Faces recurring scrutiny over supplement and wellness listings; integrating the API into listing review adds an evidence-based pseudoscience flag that keyword filters cannot provide.
- Center for Science in the Public Interestorganization
Campaigns against pseudoscientific health claims; the API's prediction-versus-patch scoring gives its advocacy a structured evidentiary basis.
- Google (YouTube)company
Moderates health misinformation at scale; a claims-screening signal based on the underlying research trail complements source-level ranking of videos.
Research it builds on
- A importância da metodologia científica no desenvolvimento de artigos científicosRabbith Ive Carolina Shitsuka Risemberg, Michael Wakin, Ricardo Shitsuka · 2026 · 520 citationsAll ideas from this paper →
- La Metodologia de los programas de investigación científicaImre Lakatos · 2026 · 285 citationsAll ideas from this paper →
- AI, Health, and Health Care Today and TomorrowDerek C. Angus, Rohan Khera, Tracy A. Lieu et al. · 2025 · 181 citationsAll ideas from this paper →
- Non-Recommended Publishing Lists: Strategies for Detecting Deceitful JournalsFahmi H Kakamad, Shvan H Mohammed, Berun A. Abdalla et al. · 2026 · 12 citationsAll ideas from this paper →