Atmospheric Reviews

Modular Research as a Commons

Torsten Goerke
Technische Universität Dresden
Center for Open Digital Innovation and Participation (CODIP)

Guest Lecture · Simon Fraser University, School of Computing Science · 2026
@tgoerke.bsky.social
Act 0 — The Utopia

What If

What if your most-cited figure could be cited independently of the paper it's in?
  • A dataset you built in 2019 accrued credit every time someone reused it — regardless of platform
  • Your community defined what "peer review" means for your discipline — not a journal
  • Your researcher identity moved with you — not with your institution or publisher
  • You had the vocabulary to describe what you actually produce — including artistic practices, co-creation, and cultural work — not just what a platform's affordances allow you to call it
"We need a seamless formal research environment — a standard protocol for easily modular publishing — with outputs interoperable across platforms." — Reuse Incentive Working Group, March 2026
Act 0 — The Utopia

From Albums to Streams

"The question is not whether scientific communication will evolve, but how that evolution will be shaped, and by whom." — Courtney Babott & Rowan Cockett, From Albums to Streams (2026) · Keynote speakers, ATScience 2026, Vancouver

Music changed when infrastructure made it possible to break apart bundled products and move value through use rather than control. The PDF bundles research the same way.

ATScience 2026 ecosystem — already building:

ChiveDecentralized eprints on ATProto
Margin.at / Seams.soSocial annotation in the atmosphere
LanyardsResearcher profiles on ATProto
SembleSocial knowledge network for research
Paper Skygest FeedPersonalized research feeds on Bluesky
Groundmist.xyzObsidian integrated with ATProto
AstroskyIndependent home for astronomy
Hypgen InfinityEvaluating scientific AI agents on ATProto
Act 1 — The Problem

POSIWID

The purpose of a system is what it does. — Stafford Beer, via Larens Hof, "The Purpose of Protocols" (2026)

  • SMTP, HTTP, RSS — claimed to be neutral infrastructure
  • Their silence on governance was not apolitical — it was a choice
  • The outcome: concentration of authority, lock-in, passive user
PlatformClaimed purposeActual outcome
Journal systemsDisseminate knowledgeSubscription lock-in, copyright transfer
Preprint serversOpen accessSingle-platform indexing monopolies
Citation indexesTrack impactCommercialized metrics, opaque algorithms
Act 1 — The Problem

The Governance Vacuum in Research

  • User lock-in → data commercialization → enshittification: platform quality degrades as users lose alternatives
  • Research platforms follow identical topology: proprietary infrastructure, opaque algorithms, extraction of community labor
  • Affordances researchers get: submit · review · publish · cite — designed for an undifferentiated "everyone" that doesn't exist
"Protocol design can constrain how actors operate within a system, but it cannot ensure the conditions that keep the broader ecosystem functioning." — Hof, 2026

No current research platform implements Ostrom's commons principles:

Defined community boundaries Collective choice mechanisms Accessible conflict resolution Proportional cost/benefit distribution
Act 1 — The Problem

The User/Developer Dualism

  • Communities select from a platform-defined set: submit, cite, download, share
  • They cannot specify: what counts as a contribution, how credit flows, what review means for their discipline
  • The user/developer dualism is an industrial narrative — "platforms build, researchers use"

The researcher loses two things:

The Process

AI tools optimized for task completion deliver outputs, not understanding. The cognitive work of science is outsourced.

The Literacy

The capacity to carry out peer review, evaluate evidence, situate claims — eroded by fluency illusion and metacognition bypass.

"Hyper-individualism — including AI-assisted 'vibe coding' — emerges as its latest manifestation." — Goerke & Barthold, Information Civics as Counter-Imaginary (2026)
Act 2 — Protocols as the Hinge

Protocols: Hard to Study, Easy to Ignore

"Hard to study and easy to ignore."— Rao et al., The Unreasonable Sufficiency of Protocols (2023)
FrameworkWhat protocols areWhat they enable
Actor-Network Theory (Latour)Bridge between ontologiesAgency in heterogeneous networks
Assemblages (Deleuze/Guattari)Sites of emergent propertiesConnections through shared experience
Cybernetics (Wiener)Governance of communicationCoordination across realms

Engineering ontologies (objects, functions, interfaces) must translate into social ontologies (power, identity, norms, credit). Protocols are the translation layer — but they are not neutral.

Act 2 — Protocols as the Hinge

The Governance Ladder

Early protocols
SMTP, HTTP, RSS — silent on governance → enabled monopolies through mimetic convergence
ActivityPub
Explicit about decentralization → recreates platform dynamics at instance level; governance by administrator discretion
AT Protocol
Most normatively explicit — "speech and reach should be two separate layers" → but Bluesky still dominates through market dynamics
"Protocol architecture shapes which governance mechanisms are 'natural' — but cannot enforce them." — Hof, 2026
Act 2 — Protocols as the Hinge

Hatching: Initializing the Commons

"Engineering any complex system is a consideration of its initialization conditions — in a Commons, we call these initial conditions the 'Hatch'." — Commons Stack (via Goerke & Barthold, 2025)
  • Not just building — co-constructing topology AND governance simultaneously
  • Not layering democracy onto pre-built infrastructure — designing democratic conditions from the start
  • The governor-engineer role: translating social frameworks into actionable technical steps

The 5-step governance pipeline:

① Identify initial conditions ② Introduce governor-engineer role ③ Apply technical knowledge ④ Shift focus to users ⑤ Implement deliberative practice
Act 3 — AT Protocol as Substrate

Why ATScience

ATScience — empowering science communities with open, democratic, researcher-owned infrastructure. (atproto.science · co-founded by Ronen Tamari, Torsten Goerke, Barry Prendergast)

Why AT Protocol as research substrate:

  • Sovereign identity — researcher identity independent of publisher, platform, institution
  • Portable data — research objects travel with provenance intact
  • Speech/reach separation — publishing ≠ amplification
  • Credible exit — structural condition for community self-determination

ATScience 2026

March 27 · Vancouver
A full-day exploration of AT Protocol in science, education, and open knowledge.

10+ projects building. The hatch is in progress.

Act 3 — AT Protocol as Substrate

Atmospheric Reviews: The Hatch in Progress

Hatching stepResearch commons equivalent
① Enable DeliberationWhat affordances does this research community need?
② Algorithmic ChoiceCommunity-governed discovery and curation
③ Sovereign IdentityResearcher identity independent of publisher/platform
④ Implement RulesGovernance encoded at protocol level
  • Modular research objects as first-class protocol citizens
  • Review, curation, annotation as specifiable community affordances — not platform defaults
  • The hatch sets governance conditions before concentration can occur
Act 3 — AT Protocol as Substrate

From Select to Specify

"Communities should specify affordances for situated needs — not select from a platform-defined set." — Goerke & Barthold, Information Civics as Counter-Imaginary (2026)

Current platforms (select)

submitreviewpublishcitedownload

AT Protocol commons (specify)

  • What counts as a contribution?
  • How does credit flow — to figure, dataset, method?
  • What does review mean — and who does it?
  • What is reuse and when is attribution required?

Information Civics: Network architecture as civic design. Not what algorithms can do — but what kinds of collective futures communities want to build, and with what authority.

Act 4 — Community in Motion

Reuse Incentive Working Group

Creative Commons + Continuous Science Foundation · Feb–March 2026

  • ~30 members: researchers, librarians, publishers, funders, technologists, infrastructure builders — global
  • Goal: surface friction around reuse and licensing of modular research; produce framework and recommendations
  • 3 meetings across 6 weeks — Meeting #3 was recently (March 19, 2026)
"Reuse is still treated as optional or downstream. What would it look like to treat reuse — and the permission to reuse — as core infrastructure for how science is shared?" — WG framing question, Feb 2026
Act 4 — Community in Motion

What Does Modular Research Actually Need?

Publishing for reuse

  • Know if and how to reuse a component as primary source
  • Know if and how to license their own components
  • Know what proper licensing requires

Setting attribution expectations

  • Track attribution per component, Panel of Evidence, Paper Repo
  • Distinguish citations for primary vs. secondary sources

Reuser guidance

  • Know what the license requires
  • Easily comply with license terms
  • Track their own compliance

Outside the PDF

  • Prevent gaming of citation metrics
  • Standard protocol for modular publishing, interoperable across platforms
  • Attribution during informal sharing
Act 4 — Community in Motion

Citation as Currency

"Citation as currency — but citations have no sentiment: support, reject, oppose. The hazard is that this might be gamed." — Reuse Incentive Working Group, March 2026

What modular research requires instead:

  • Attribution that propagates downward — to sub-components, figures, datasets, methods
  • Credit that travels across platforms — not locked to a single index
  • Licensing embedded at the object level — not as a separate document

TASLAR metadata standard (per research object):

Title Author Source Link License Link Attribution Statement Rights Statement
"Machine automation: embedding attribution as a cultural norm, and not just as a legal requirement."
Act 5 — The Research Horizon WIP

AIDLE: A Research Agenda

The remaining gap: who has the capacity to co-design? The governor-engineer role requires both technical and governance literacy — and AI tools as currently designed are making that rarer.

Thesis
Vibe coding = task completion = the latest form of the user/developer dualism
Antithesis
Hatching a commons — communities specifying affordances, deliberative co-design
Synthesis
AIDLE — AI redesigned for human capability development, not task completion
  • AI that knows when not to answer — scaffolds co-design rather than completing tasks
  • Skill Trajectory Modeling: tracks capability gaps, not preferences
  • Atmospheric Reviews as testbed: agentic tools that scaffold protocol co-design for research communities
"Not what algorithms can do — but what kinds of collective futures communities want to build."
Act 6 — Close

Protocol-Mediated Agency

The PDF bundled research the way albums bundled music. AT Protocol gives us the infrastructure to unbundle — and re-bundle on community terms.

🔑
Specify
affordances for your community
🏗
Hatch
governance before concentration
🌐
Own
identity, data, relationships

Hatching is underway: Reuse Incentive WG · ATScience 2026 (March 27, Vancouver) · Atmospheric Reviews · AIDLE

What affordances does your research community need to specify — and who currently decides?

Torsten Goerke · @tgoerke.bsky.social · atproto.science