Fermat's Library Cofounders João Batalha and Luís Batalha

Fermat's Library is a browser-based platform designed to annotate academic papers with support for LaTeX and Markdown, simulating the experience of an offline university journal club.

• The platform was founded by four co-founders with technical backgrounds in physics, computer science, economics, and data analysis (MIT alumni).
• It currently hosts over 1,000 annotated papers, with the original Bitcoin whitepaper remaining the most commented-on document.
• Notable early annotations include contributions from industry figures like Laurence Lessig and various members of the Bitcoin community.

Growth Strategy and Cold Start:

• The founders initiated the platform by annotating papers themselves to generate initial content.
• The initial "Bitcoin" annotation serves as a permanent reference point, often cited by news outlets and used by readers seeking clarity on complex concepts.
• The primary growth engine involves expanding from a simple journal club into a broader open science platform.

Core Mission and Open Science:

• The platform aims to advance "open science," encompassing open data sharing, open code availability, and open publishing (bypassing paywalls).
• It seeks to facilitate remote collaboration, citing the 2016 solution to the Erdős Discrepancy Problem as a prime example of online collaboration.
  • This problem was solved after a brief blog comment by a German programmer provided a key insight to Fields Medalist Terence Tao.
• The founders view the platform as a potential "Stack Overflow for science," enabling collaborative problem-solving on a global scale.

Technical Tools and Chrome Extension:

• The team released a Chrome extension for arXiv to address the lack of commenting on preprints.
  • arXiv hosts preprints (drafts) before journal publication, which is critical in fast-moving fields like machine learning.
  • The extension allows users to view annotations directly on the arXiv page, avoiding the need to navigate to a separate site.
• The extension includes features such as automatic BibTeX extraction, reference linking, and comment management.
• Future Features: The team is considering a rating mechanism (likes/dislikes or aspect-specific ratings) to help users distinguish quality work among unreviewed preprints, though this requires further user surveys.

Addressing Scientific Challenges:

• Negative Results: The founders acknowledge the systemic lack of incentives for publishing negative results or failed experiments, which hinders scientific progress (e.g., in biology and economics).
• P-Value Hacking: The team is investigating methods to highlight potential statistical manipulation (p-hacking) in economics and biology fields, potentially via API integrations or the Chrome extension.
• Annotation Quality: They advocate for annotations by non-authors, as authors often lack insight into where readers struggle due to their own internalized knowledge.

Content Discovery and Trends:

• Paper Length: The founders note a historical shift where papers have grown from 1-2 pages (e.g., Freeman Dyson's "Dyson Sphere" paper) to 15+ pages, attributing this to increased complexity and formal formatting requirements.
• Impact Metrics: The team critiques citation counts as an imperfect metric for impact, suggesting that review papers (e.g., Freeman Dyson's unification of Quantum Electrodynamics) often have more profound influence than original discovery papers.
• Book Annotation: They discuss the potential for annotating introductory textbooks (e.g., Goldstein's Classical Mechanics) to create open-source, evolving learning materials that overcome copyright and static content limitations.

Operational Philosophy:

• Side Project Status: Fermat's Library operates as a side project without current profitability goals, funded entirely by the founders to cover server costs.
• Resource Allocation: They prioritize features with the highest immediate impact (e.g., the arXiv extension) over long-term revenue models.
• Peer Review Evolution: The founders believe the current peer review system is too slow for modern research (especially in AI) and support the rise of "overlay journals" (e.g., Discrete Analysis) that host preprints without paywalls.

Community Engagement and Future Outlook:

• Twitter as a Learning Tool: The founders use Twitter to disseminate complex concepts in accessible "quantum of knowledge" tweets, observing high user engagement and addictive learning behaviors.
• Call to Action: They invite users to annotate papers, share annotations within university journal clubs, and upload personal papers for group collaboration.
• Monetization: The team is open to cryptocurrency donations but remains focused on non-profit sustainability, drawing parallels to the success of Wikipedia and Stack Overflow.