> For the complete documentation index, see [llms.txt](https://docs.mosaic-program.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.mosaic-program.org/3.-mosaic-approach.md).
# 3. MOSAIC approach
Surprisingly, science publishing and communication has never been treated as a key R\&D priority. We have seen the potential of large-scale networked research through "accidental" emergence in platforms that were never designed for the purpose (e.g., Science Twitter, see below). We also have a good theoretical understanding of the principles driving large-scale collaborative efforts, like open-source code and Wikipedia (e.g., Michael Nielsen's Reinventing Discovery). These have never been applied to science. MOSAIC's Big-if-True bet is that catalytic funding and intentional engineering can deliver for science what has only emerged by accident on platforms built for other purposes. And like Wikipedia or open source, an open-protocol substrate is generative: researchers, labs, and AI builders can extend it themselves, growing the system beyond any fixed feature set.
#### **Core hypothesis**
Three mutually reinforcing developments are converging to create a unique opportunity. (1) Modular research tools (Discourse Graphs, Nanopublications) are making it possible to represent the actual process of research (hypotheses, questions, critiques, evidence, etc) as reusable, machine-readable units rather than burying them inside monolithic papers. (2) Open social network protocols, particularly Bluesky/ATProtocol, now offer a credible alternative to closed platforms and are attracting large science communities (see Appendix A). These spaces are already where many scientific microcontributions happen (discussion, critique, data sharing) but the infrastructure to recognize them as actual research contributions doesn't yet exist. (3) AI capabilities have reached the point where it can meaningfully contribute across the full research lifecycle but lacks the structured scaffolding needed to deliver — exactly what modular research tools and open protocols provide (see Appendix C).
None of these alone is enough: modular units without network adoption, social networks without recognition, AI without scaffolding. Together, they form the substrate for a fundamentally different architecture of science.
#### **Anchor evidence**
Three lines of evidence anchor this thesis (details in Appendix A).
First, networked research has already produced transformative outcomes on platforms not designed for it, such as COVID Moonshot's preclinical antiviral candidate (a 150-researcher, “Twitter-fueled” drug-discovery collaboration), informal peer review that resulted in consequential retractions, and scores of researchers for whom Science Twitter profoundly shaped their intellectual and professional development.
Second, the ATProto Science ecosystem, which we founded in December 2025, has seen more than 20 independent projects emerge in \~6 months — most volunteer or bootstrapped — showing that researchers and developers will build research infrastructure on shared open protocols once they exist.
Third, Discourse Graphs are actively deployed across 10+ research labs (100+ weekly active users, 6,000+ structured contributions in a year), with adopters reporting substantial improvements to the quality and confidence of their scientific thinking. Modular research is emerging as the medium of choice to scaffold agentic AI for science contributions (see Appendix C).
#### **Prior initiatives**
Prior efforts’ impact has been limited by multiple factors (see Appendix B for full comparison): closed platforms (ResearchGate), premature financialization (blockchain DeSci), AI-for-science tools built on paper-based substrates, and modular research initiatives without reward alignment. MOSAIC's power lies in novel integration: every component is designed to deliver tangible benefit to the working researcher — better sensemaking in their own work, clearer credit for the full range of their contributions, and productive collaboration with AI and larger teams — while also generating the structured signals needed for funding allocation and credible research assessment. In particular, MOSAIC connects modular contribution data directly to a research funding platform (Catalyze), closing the loop between scientific contribution and capital allocation in a way no prior effort has achieved. Adoption follows when contributing this way is itself the reward.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.mosaic-program.org/3.-mosaic-approach.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.