Keynote: Towards Modular Open Science (Auditorium)

featuring Rowan Cockett (Curvenote, Continuous Science Fdn) and Matt Akamatsu (Asst. Prof, Discourse Graphs)

Keynote 9:15-10:00 AM

The traditional scientific record—anchored in static, monolithic PDFs and siloed journals—is increasingly ill-equipped to handle the speed and complexity of modern discovery. This keynote explores a transition toward Modular Open Science: a future where research is a continuous, federated, and computable graph of knowledge.

We outline a vision to move beyond isolated papers toward a world of interoperable research objects and introduce how the Open Exchange Architecture (OXA) and Discourse Graphs provide structural standards for this shift and can leverage the AT Protocol as a decentralized backbone. Additionally, we will discuss how tools like Curvenote and Discourse Graphs have enabled researchers to capture the computational origins of research and the granular logic of scientific inquiry in real-time, supporting networked research collaborations. Join us as we outline a roadmap for a modular ecosystem where data, code, and discourse are seamlessly integrated, verifiable, and interoperable.

Speaker Bios

Rowan Cockett

Rowan Cockett is the CEO and a founder of Curvenote (https://curvenote.com), where he leads the development of tools designed to free scientific communication from static PDF documents. Curvenote enables researchers, societies, and research institutes to publish interactive, reproducible, and richly linked scientific content, with a particular focus on computational research. By integrating narrative, data, code, and figures into cohesive, reusable research objects, Curvenote reflects how modern science is conducted.

Rowan is also a founder of the Continuous Science Foundation (https://continuousfoundation.org), a nonprofit organization for the next generation of scientific communication. Through this work, he helps lead the development of shared, open standards, including the Open Exchange Architecture (OXA, https://oxa.dev), which support modular, interoperable, and machine-actionable scientific content. These standards aim to enable experimentation, reuse, and long-term preservation across tools, publishers, and research communities, supporting a more continuous and composable model of science.

In addition, Rowan serves on the steering council for JupyterBook and MyST Markdown (https://mystmd.org), part of the Project Jupyter ecosystem. These widely adopted open-source tools support authoring and publishing of computational and scientific content, and Rowan has played a key role in shaping their technical direction and integration with emerging content standards.

Rowan holds a Ph.D. in computational geophysics from the University of British Columbia (UBC). While at UBC, he helped found SimPEG (https://simpeg.xyz), an open-source simulation and parameter-estimation framework for geophysical applications including electromagnetics, fluid flow, and gravity. SimPEG is now used across academia, national laboratories, and industry worldwide.

Rowan has received multiple awards for innovative research dissemination and open educational resources, including Visible Geology, an interactive geoscience modeling application used by over one million students globally. Prior to founding Curvenote, Rowan was Vice President of Cloud Architecture at Seequent, where he led large teams building computational platforms, visualization tools, and version-control systems for geoscientists.

Matt Akamatsu

Matt Akamatsu is an Assistant Professor of Biology at the University of Washington. He is also the cofounder of the Discourse Graphs Project, an organization supported by the Chan Zuckerberg Initiative and Navigation Fund creating systems and software tools for modular research collaboration in the age of AI.  He is a recipient of the 2020 Porter Prize for Research Excellence from the American Society for Cell Biology, K99 Pathway to Independence Award, and The Experiment Foundation's Beyond The Journal award.

Matt Akamatsu received his PhD in the department of Molecular Biophysics and Biochemistry at Yale University as the first student in Yale’s Integrated Graduate Program in Physics, Engineering and Biology. His postdoctoral work at UC Berkeley (...WIP!)

Building AT Science: Infrastructure (Auditorium)

featuring talks from Ronen Tamari, Maria Antoniak, and Aaron White

Presentations 10-10:45 AM

10:00 AM Presentation

10:15 AM Presentation

10:30 AM Presentation

AI/Data on ATProto (Auditorium)

featuring talks from Martin Karlsson, Dr. Maxine Levesque, and Dr. Sean Jungbluth

Presentations 11:00-11:45 AM

11:00 AM Presentation

11:15 AM Presentation

11:30 AM Lightning

Sensemaking Systems (Auditorium)

featuring lightning talks from Travis Simpson, Alex Garcia-Joyner, and Anish Lakhwara

Presentations 11:45-12:15 PM

11:45 Lightning

11:55 Lightning

12:05 Lightning

Can decentralists cooperate? Rethinking commons and collective action in the age of platforms and AI (Lounge)

a conversation with Laure Haak, Ellie DeSota, and Nick Vincent

Discussion 11-12:15 PM

With increasing commodification of research come challenges to connection, communication, and research integrity.  ATProto's open protocol, extensibility, and large uptake by researchers provides a unique moment of opportunity to reassert community control in the commons, and specifically in processes of science publishing and communication.  But is merely building new technology on open protocols enough? 

As argued in by Laurens Hof (

@laurenshof.online

) in a recent piece, open protocols need "not just the technical architecture that brings shared resources into being," but also new forms of governance: "the institutional arrangements that manage them, the collective-action mechanisms that sustain them, ... and the design principles ... that can help ensure they serve the communities that depend on them rather than the actors best positioned to capture them."

This panel will explore how challenges facing the broader atproto ecosystem are mirrored in its open science applications. Some questions we'll ask include:

• What is the research commons?

• What are the boundaries for inclusion in the commons?

• Who should be involved in decision-making on new protocol features? Who is responsible?  What does accountability look like?

• Who should pay for developing and running the infrastructure?

• How can the value created by individuals using ATProto balanced with the companies, bots, and agents mining user behaviors and activities?  How important is it to distinguish between humans and AI agents?

• How are new forms of data-based collective bargaining applicable to scientific publishing and communication? 

• Should researchers be thinking about joining unions to protect their creative work?

Further Reading

Paul Frazee, Practical Decentralization. (Feb 2026). 

The point of decentralization is to guarantee the rights of individuals and communities on the Internet.  Protocols are used to structure who can do what. He uses the term “Information Civics” for decentralization as a practice: a marriage of software engineering and civic design.  Protocols need to achieve ideological aims and they need to produce useable software. Good protocol design requires balancing the tensions of ideology and practicality. 

Laurens Hof, The Purpose of Protocols. March 2026 

If protocols are intended to guarantee rights, then they are governance structures whose design choices allocate power. Protocol design is a form of political design, and the appropriate way to evaluate protocols is not only by their technical properties but by the governance outcomes they produce.  However, protocols including ATProto are “silent” on governance – reflecting an engineering culture in which protocols were understood as infrastructure, and infrastructure was assumed to be politically neutral. The protocol's silence about governance, power, and purpose did not produce a neutral outcome but a highly specific one: the concentration of power in the hands of actors who could build on top of the protocol's open infrastructure without being constrained by its openness. 

Sabina Leonelli: Philosophy of Open Science, 2023

Openness is often viewed as a necessary complement to accountability and public scrutiny.  The Open Science movement maintains a strong continuity with values long viewed as definitive of scientific research – such as the critical questioning of dogmas, the search for reliable evidence, the privileging of rational reasoning and the emphasis on public scrutiny and debate. At the same time, OS has gathered momentum over the last three decades as a response to the broad transformations brought about by the digitalization, globalization and commodification of research. This evolution of OS, in its insistence on machine readability, interoperability, and resource sharing, is having the unwanted effect of constraining epistemic diversity and worsening epistemic injustice.  This results in unreliable and unethical scientific knowledge.  OS should be reframed to be an effort to establish connection – meaningful communication – among humans doing research and their systems of practice. Only by building in ethics will science achieve openness.

Building AT Science: Community (Auditorium)

featuring presentations from Dr. Emily Hunt, Dr. Teon Brooks, and Dr. Scott McGrath

Presentations 13:15-13:50 PM

13:15 PM Presentation

13:30 PM Lightning Talk

13:40 PM Lightning Talk

Future of Science Social Media (Auditorium)

panel featuring Dr. Scott McGrath, Maria Antoniak, Ariel Lighty, and Ronen Tamari

Panel 13:50-14:30 PM

In this panel we will explore why, despite early momentum, the migration of researchers to Bluesky has waned. We will discuss better ways to onboard, retain, and attract researchers by highlighting the flexibility and extensibility of the AT Protocol. Panelists will share what's worked, what hasn't, and what a coordinated push to build a science ecosystem on Bluesky might look like.

An atproto.science workshop

Sensemaking Systems + AI for Science (Lounge)

demos from Travis Simpson, Alex Garcia-Joyner, Anish Lakhwara, Semble, Martin Karlsson, and Sean Jungbluth

Workshop 13:15-14:30 PM

Sensemaking Systems

• Seams (Anish)

• ViewSift (Alex)

• Skysquare (Travis)

AI for Science

• Agentis (Sean)

• Coordination Network (Martin)

An atproto.science workshop

Social Media Research (Auditorium)

featuring presentations from Jay Patel, Sophie Greenwood, Francisco Carvalho, and Billy Pierce

Presentations 14:45-15:40 PM

14:45 PM Lightning Talk

14:55 PM Presentation

15:10 PM Presentation

15:25 PM Presentation

Building ATScience: Sustainability (Lounge)

A conversation with Mathew Lowry, Ronen (Semble), Brendan (Leaflet), Tyler (Sill), Travis (SkySquare), Paul Fuxjäger & others (tbc).

Workshop 14:45-15:50 PM

We'll meet in the lounge to unpack, discuss, tear apart and rebuild the following idea:

We already have four five speakers, but drop me a line if you'd like to contribute and I'll update the outline agenda (to be agreed when we get started):

• Short scene-setter for atproto-curious scientists: What is ATProto, how does it work, and why should I care? (Mathew)

• First steps up the value chain (Mathew):

  • Organise your Bluesky footprint with starter packs and custom feeds (source: How newsrooms, scientific institutions & governments can best use Bluesky, Mathew, Nov 2025)

  • What It Means for Universities to Run Their Own PDS (source: The Potential of DID as Academic Infrastructure, Nighthaven)

• What would a research team website look like? Part 1: Leaflet, Standard.site and Atmospheric publishing (Bernard Schlagel - see also Atmospheric Publishing Discussion, just after this session)

• Collective knowledge discovery & sensemaking

  • From MySill to OurSill (Tyler Fisher - see also Rethinking the Client: Why User Choice is the Key to Growth for ATProto, during AtmoshereConf)

  • Aggregating Semble and Margin (Ronen - see also Semble: A social knowledge network for research on ATProto, in the Building AT Science: Infrastructure session, 10-10:45)

  • The web as a socially annotated layer of context (Travis - see also Skysquare is context as a service, in the Sensemaking Systems session, 11:45-12:15)

• What would a research team website look like? Part 2 (Mathew): combining the above

• Discussion: what do the Universities in the room think?