Matt Akamatsu and Evan Miyazono, in conversation with Tom Kalil
About
A shared structure for research work in motion
Discourse Graphs help teams move beyond static documents by representing research as connected claims, evidence, questions, and projects.
Compose work into a graph that can be queried, remixed, and carried across tools.
Keep findings live as teams interpret evidence and update their models.
Make research ideas composable
Map claims, evidence, questions, and projects as modular pieces that can be reused across people, tools, and contexts.
Liberate findings from static files
Turn high-signal findings into shareable graph objects that support attribution, reuse, and decentralized collaboration.
Synthesize and update continuously
Exchange work in a form that is easier to find, update, and build on than documents locked into one hierarchy.
Communicate with better structure
Separate observations from interpretations so researchers can compare, remix, and update shared knowledge without flattening disagreement.
Tool choice
Client-agnostic and researcher-aligned
Discourse Graphs are a decentralized knowledge exchange protocol designed to be implemented and owned by researchers rather than publishers.
The model can be implemented in networked notebook software like Roam Research, Obsidian, and other tools researchers already use.
Discourse Graphs are like GitHub for scientific communication.
Plugins
Start from the tool you already use
The project maintains plugin workflows for Roam Research and Obsidian, with docs for installation, graph building, querying, and configuration.
Roam Research
Roam Research plugin
Available via Roam Depot. Use the docs to set up the plugin and learn the core discourse graph workflows.
Obsidian
Obsidian plugin
Available via BRAT. Use the docs to set up the plugin and learn the core discourse graph workflows.
Cloud laboratory
A natural operating layer for active research
The flexible framework has been adapted to coordinate and share active research, lowering the barrier for interdisciplinary collaboration.
- Identify knowledge gaps and starter projects for new researchers
- Onboard collaborators into active research faster
- Support grassroots knowledge generation between researchers
- Share modular research findings with clearer attribution
Lower-friction collaboration
Faster discovery cycles
Resources
Read the ideas behind the project
A short set of writing and project notes for understanding discourse graphs as research infrastructure.
Project notes
Discourse Graphs for research lab coordination
Open resourcePreprint
Discourse graph plugin design and use cases
Open resourceConceptual model and practical guide
Knowledge synthesis
Open resourceSustainable authorship models for scholarly communication
Open resourceEvents
Recent talks, panels, and workshops
Places where the team and collaborators have presented the Discourse Graphs model and project.
Frontiers in Research: Open Science Catalyze Panel
June 18, 2026 | Zoom
Toward Modular Open Science
March 27, 2026 | ATScience Conference, Vancouver
Seminar: McGill University Quantitative Life Sciences program
March 24, 2026 | Montreal
Metagov x Future of Science Seminar: Interoperable LLM- and human-centered research with Discourse Graphs
November 19, 2025 | Zoom
IOSP '25 Winter Workshop: Discourse Graphs
February 23-24, 2025 | Denver Museum of Nature and Science
Talks
Watch the model explained
Videos and recordings that introduce the motivation, use cases, and research workflows behind Discourse Graphs.
Keynote: Towards Modular Open Science
Rowan Cockett, Matt Akamatsu
HCIL Brown Bag Speaker Series: Matt Akamatsu
Discourse Graphs: A New Model for Scientific Communication
Matt Akamatsu, Topos Institute
Open Sourcing Scientific Research with Lab Discourse Graphs
Matt Akamatsu, Desci Denver 2024
Accelerating Scientific Discovery with Discourse Graphs
Joel Chan, Protocol Labs Research Seminar
Research roadmapping with Discourse Graphs
Karola Kirsanow, NYC Protocol Labs Research Seminar
Team
Built by researchers, designers, and engineers
The project brings together research infrastructure, knowledge synthesis, and tool-building experience.
Joel Chan
Research
Matthew Akamatsu
Research
Michael Gartner
Tech Lead
John Morabito
UX Lead
Siddharth Yadav
Developer
Trang Doan
Developer
Marc-Antoine Parent
Developer
Karola Kirsanow
Pilots Lead
David Vargas
Advisor
Supporters
Supported by open science funders and partners
Organizations helping make modular, reusable scientific communication possible.
Contact
Help build better infrastructure for collaborative research
We are building user-friendly Discourse Graph plugins in tools for thought and would like to hear from researchers, developers, labs, and funders.