Skip to content

Runrig: A Plan for Socio-ecological Design


In its narrowest sense, Runrig is a technology framework for the communal management of a foodshed's digital assets, potentially spanning a network of cooperating foodsheds. To the degree that such information may control the flow of actual material resources, it may be viewed as a tool for the democratic management of food, labor, land and other resources that such information represents. In today's paradigm of "platform capitalism", technology companies derive their wealth and power by mediating interactions between users and capturing a portion of the value exchanged in the process. That portion only grows over time, even when the platform is no longer needed to mediate such an exchange.[1] By recognizing this, and reclaiming common ownership of the technologies that mediate relations between farmers, distributors, drivers, wholesalers, retailers, commercial buyers and end consumers, Runrig aspires to nothing less than the total redistribution of that value, giving control of the wealth and power it creates back to the people who created it.

Three Layers of Autonomy

To understand what Runrig does, we can start by examining what kind of digital assets it is intended to manage. Partly for explanatory purposes, but also for reasons that will become clear later on, let's separate assets into two very broad categories: the data itself, and the software programs that capture, process and generally make use of such data. Free software proponents and privacy advocates have spilt a good deal of ink advancing the argument that true autonomy in computing requires control of both the data and the software, not to mention the hardware the whole system runs on.[2] [3] [4] As the old saying goes: "There is no cloud, just someone else's computer."[5] Efforts to achieve computational autonomy, however, rarely exceed a vision for individual autonomy, and so are inevitably limited by individual capability. These solutions generally fall short in one of two ways: either they must curtail the range of functionality users have come to expect from so-called "Software-as-a-Service" (SaaS) platforms; or they foist the burdens of administering and maintaining those systems onto users who lack the capability to do so. Telling average users, who seek an alternative to predatory cloud platforms, that they can just spin up a service on their own server rack or VPS (then run backups and install updates all by themselves forevermore) is no better than playing the world's smallest violin for them.[6] If free software and open source advocates focus too narrowly on licensing and sharing source code, while overlooking the actual capabilities this affords their users, then at best they can only offer permissive freedom, not effective freedom, as Luis Villa observes.[7] In other words, they've granted permission to freely use the software, in a purely hypothetical sense, but not the real capability to use it in any practical or meaningful sense.

Runrig differs from other open source methodologies by coordinating the pooled capabilities of a community under democratic and cooperativist principles of governance. This way, ownership and control of both the data and the software can be shared collectively by all participants in a foodshed, while preserving the rights of individual users. To achieve this, Runrig comprises 2 - 3 functional layers:

  1. A single, collectively owned data storage provider, or "pod"[8] provider.
  2. Many federated cooperative service platforms.
  3. Local-first[9] and self-hosted applications (optional).

Stack diagram

With sufficient investment, the server racks, network infrastructure, brick-and-mortar facilities and other physical assets required to run these systems could be cooperatively owned as well. For now, however, Runrig restricts itself to the digital assets, since their storage and management can be cooperativized at a much lower upfront cost.

The Data Provider

The first layer is a global commons, providing networked data storage to all its members as a shared service, while diffusing the costs and responsibilities of managing the underlying databases and infrastructure such a system entails. This layer corresponds to the data category of digital assets, as outlined above, and should be relatively indifferent to the shape of that data or how it's intended to be used.

This data provider can be thought of as a kind of cloud-based file system, like Dropbox or Google Drive, accessible from anywhere with an Internet connection, but with two key differences. First, it is collectively owned by all its members, not some corporation and its shareholders. So while it may not be each user's private computer, it is still their computer in a very real sense, with full governance rights over the infrastructure hosting their data, in addition to individual protections under the terms of service. Secondly, while it could in fact be used as a file system, its true potential lies in being a source of structured data, like an SQL database or even more modern low-code databases like NocoDB, a libre alternative to Airtable. This gives it the power to integrate more readily with a myriad of existing services via its API.

Service Platforms

The second layer is a federation of platform cooperatives[10], which can run the gamut of web-based applications and other software services, like the ones we're all accustomed to today. This layer corresponds to the software category of digital assets, but it draws its core data from the first layer, the data provider, linking the two layers programmatically as well as contractually. Services might include consumer-facing ecommerce sites or wholesale buying clubs, crop management software and record keeping, logistical and fleet management, bookkeeping, timesharing for certified kitchen space, etc. A platform is not limited to providing a single service, and it can have separate websites set up for separate services, or integrated as one. For example, there may be one site where retail customers place orders and another site where farmers track inventory through a different service. Critically, however, the two systems would still be integrated, sharing infrastructure, operating costs, and a unity of purpose. Users would also enjoy a seamless "single sign-on" between services, because the data provider can double as a secure identity provider as well.

Each of these service-oriented platform coops would function best within a geographically contiguous region, what we loosely call a foodshed, food zone[11] or bioregion. This way they can curate the services provided on their platform in order to best suit the needs of their members. They can even tailor the configuration and features of specific services as they choose. For example, if a platform wants to add a service for aggregating CSA orders, they can do so easily. If they need a specialized feature for a citrus CSA and one exists, they can enable it. If it doesn't already exist, they can sponsor its development and contribute the open source feature back to the commons. Meanwhile, other platforms could add features for a flower CSA instead, or omit the CSA service entirely. In this way, geographic regions offer a sensible criterion for bounding a platform's scope of functionality, since some services might make sense in one region or climate but not another. It also keeps governance more manageable, when not trying to accommodate the needs of a larger and more dispersed membership. In theory nothing prevents a platform from being as small or large as desired, but each platform will require some combination of time, expertise and money to maintain, so the distribution of those costs should be balanced with the desire for more localized control. Nevertheless, it shouldn't be necessary to impose strict boundaries on any one platform's range of coverage, and users would be free to join more than one platform if available and desirable. It may also prove useful to cover a more dispersed geographic area, for niche sectors of production or distribution, like pasture-raised meats or high-value crops. These are tradeoffs that should be left to the thoughtful deliberation of the constituents of each platform, so they can strike the right balance to meet their own goals given their unique circumstances.

This second layer, regardless of the particular platform or service, is also the layer that users will typically interact with. It will be the service platform's website where most users sign-up for an account, assuming they've been approved for membership in the cooperative that operates and governs that platform. In the same stroke, they will gain membership to the global data provider as well. They will be given a data pod that belongs exclusively to them, designated by a unique address, so that the first layer will also serve as an identity provider, much like an email address. The technical process can remain invisible to the user, but they can feel assured that their data is theirs to control. This is backed up by the fact that they hold a share in the governance of how the data provider stores their data, as well as a share in how any service platform they join is allowed to use it. If at any time they wish to join a different service or leave entirely, they can more easily bring their data along with them, because portability is a fundamental aspect of the design.

Ultimately, the service platforms only mediate the exchange between the data provider and the user, both in terms of the actual data exchange, as well as the service level agreements between all three parties. That is not to say that power users are prevented from interacting directly with the data provider, especially if they have programming knowledge; they just don't have to do so directly. They don't have to be aware how those data transactions work, or that the data provider even exists, although hopefully they will appreciate that regardless of what service platform they use, they retain certain rights over both their data and how the storage of it is administered.

Local-first & Self-hosted Applications

The third layer is comprised of local-first applications, which individual users can install on their personal desktop and mobile devices, as well as self-hosted services, where the more technically savvy users are so inclined. These are not strictly necessary, however, and will only appeal to certain users at first. The other two layers are perfectly capable of providing fully featured web-based applications on their own, without the need to install and maintain purely local instances. Yet the possibility emerges quite spontaneously from the other two layers due to their intrinsically distributed nature. This affords the potential for additional peer-to-peer features and decentralized functionality, which could provide greater resilience to the overall network while also granting more freedom to individual users outside of those networks.

The details of this are still less critical to Runrig's initial development, but will be further explicated in the "Architecture" section of this proposal, though it is still in draft.


Part of a Larger Ecosystem

Brought to its full potential, Runrig goes far beyond what a single software project or platform can achieve. Its main value proposition is that it opens up new ways for complex, distributed networks of groups and individuals to collaborate freely and productively. It would be arrogant to presume that we could bootstrap an inherently collaborative, widespread network from the ground up, in a few short development cycles with only a handful of engineers, designers and agriculturalists. Fortunately, that is not our task; we draw upon a rich substratum of prior work that has already established the necessary standards, technical infrastructure, methodologies and other essential building blocks that make Runrig possible.

Food standards and ontologies have been pioneered by such groups as the Data Food Consortium[12], stewarded as an international collaboration by the Open Food Network. There also exists a full suite of agricultural tools from OpenTEAM[13], including the farmOS Data Model[14], which enjoy the continued support of the USDA, the Foundation for Food and Agricultural Research and other public and private partners.

These foundational projects are themselves part of a broader groundswell of technological advancements, particularly within the field of federated and peer-to-peer systems. Over the last two decades or more, Tim Berners-Lee and the W3C have spearheaded the drive for a Semantic Web,[15] which has made tremendous strides in just the last few years. Practical implementations of the Solid Protocol and WebID have now reached production, and community-supported servers are free to join or can be deployed independently as a testbed for Runrig's data provider.[16] [8:1] [17] Parallel developments like ActivityPub, while not strictly applicable to food and agriculture, provide a template for how standards and shared toolchains can evolve by methods of open innovation and collaboration.[18] [19]

By taking advantage of these open standards and tools, there is a lot that can be done by essentially snapping together off-the-shelf components. Combining that with the collaborative methodologies that have become all but customary in these communities, Runrig can enjoy a degree of near-term stability and long-term sustainability that are rarely found among early-stage startups.

Unique Advantages

These open source methodologies, free software libraries and proven models for standards development will all be instrumental in building our own network of service platforms for food production and distribution; however, Runrig's greatest advantage is how it can build directly on top of existing networks to deploy practical solutions today, while the larger technical infrastructure can grow incrementally, along with the networks of trust and social infrastructure undergirding the technical aspects.

This is where the separation between the data and software layers, as outlined above, proves so decisive. Because the service platforms are decoupled from the data source, they can be adapted to send and receive data to and from preexisting platforms based on compatible libre software systems, with very little additional overhead. This is not for the purpose of supplanting those platforms; rather, Runrig can complement their range of functionality and open up new avenues of potential development within the entire ecosystem.

The full value of this separation of concerns is best demonstrated by the way Runrig can prototype novel solutions that satisfy edge cases and fill critical gaps in the feature offerings of other libre software platforms. There will always be features that either fall outside the scope of more mature platforms, prove too costly to develop due to particular implementation details or accrued technical debt, or that merely get deprioritized in the context of more pressing concerns. These costs and other hindrances shrink to almost nothing, however, when an independent service can take full advantage of the existing platform's network and data storage, while still enjoying effective greenfield project status, unconstrained by previous design choices.

For example, one prototype we've discussed in the Skywoman MAIA Project is a post-planting survey for a producer coop to audit its member-farms and asses whether they are on target to match anticipated sales volumes. The coop manager only needs to send the survey out once or twice per season to about a dozen farms, but it could save her a lot of precious time otherwise spent on site visits.[20] Runrig could host the data, structured as farmOS plans, assets and logs,[14:1] while also being compatible with the Solid protocol. This could be done with a self-hosted Solid Server[17:1] or even just a simple JSON document which could be transformed to a spreadsheet once the responses were collected; for this proof-of-concept, at least, the data requirements are miniscule. The survey itself could be created using SurveyStack, a low-code solution provided by Our Sci and OpenTEAM, with integrations already available with the farmOS Data Model.[21]

While solutions like this have a payoff about on par with a Google Form plus a spreadsheet, the costs of developing and maintaining them with Runrig are still very low. Spreadsheets may be easier to create, but often prove much harder to maintain in the long-run. And what costs exist are not sunk costs; if the results aren't totally satisfactory, they can stick to the spreadsheets from then on, without losing that data or needing to export it from a third-party system. Maybe they like the concept, but want something more "off the shelf" like farmOS or Inrupt, in which case the data is easily ported there. And of course if they're happy enough with the system as it is, or want to continue building its potential to meet their unique circumstances, development can continue apace with Runrig. Best of all, none of these options are mutually exclusive, and in fact, the coop could keep on using the latter two options in parallel, getting the best of both worlds, without any serious costs associated with keeping the two redundant data sources synchronized.

That's where Runrig can most benefit other platforms in the ecosystem, by catching the spillover of feature creep that inevitably besets older software projects, while accommodating users for whom the lack of those features would otherwise be a deal breaker. A synergy arises between Runrig and other free software platforms, expanding the user base for each, which wouldn't be possible if Runrig's data and service layers were more tightly coupled.

Ecology Over Architecture

Runrig will center ecological design over any specific technological architecture. Ecology, in this sense, encompasses the technological, social and biotic aspects of design, together and all at once. Accordingly, we strive to make those designs adaptive to the whole, rather than prescriptive in the particular.

In a practical sense, this grants us greater freedom to experiment with these technologies. The example of the auditing survey mentioned above is a prime example. The solution is small, almost trifling, and it risks very little; however, it can have very real benefits, meeting the most immediate, material concerns of its users. While it can be tossed aside if it proves unhelpful, it could just as well form the kernel of a much larger system. In that sense, it's hardly even fair to call it a prototype. This is yet another direct consequence of separating the data provider from the services, an intrinsic element of the overall design that can be achieved from the start, without temporary expedients. In this way, the eventual contours of the superstructure are imprinted at the very laying of its foundations.

This design would give Runrig some rather attractive scaling properties, but "scale" alone is grossly inadequate for describing the full benefits. The continued ability for experimentation and free play, jeu libre, throughout the lifespan of the project, and the potential it yields cannot be understated. The design choice to enable the collective stewardship of data, together with the free and open sharing of the software, is as much a principle of social design as it is a technological one. It imparts a vitality and iterative quality to the project, akin to the Agile methodology, but extending well beyond mere product development. The evolution of the software system is only one factor in the evolution of the broader social and natural systems in which it is embedded. Experiments, whether they succeed or fail; ad hoc solutions, whether they can be adapted to general uses or not; shared techniques and information, whether that sharing is reciprocated or not — all of these outcomes will be metabolized just the same within an open system of communal ownership, and so their lessons and shared progress always nurture the whole.

It should be pointed out that the separation between the data and service layers is by no means absolute; again, such prescriptiveness is to be avoided. The services will always have some data which they persist themselves, and there's obviously no way the data provider could provide data if it were not itself acting as a service. The key abstraction is more social than technical in nature, intended to deconstruct the prevailing notion that both data and services must be centralized with strict top-down control, as well as the countervailing notion that our best alternative is strict decentralization, where data and services are still co-located in discreet instances. The aim is to empower new, imaginative configurations that instead emphasize the freedom of users, both as individuals and collectively, to determine the location of their data, how it is used and in what directions it flows. Tight coupling of the two layers restricts that movement, stifling the health of the overall ecosystem, to say nothing of what that means for the effective freedom of the people within it.

Technology can be far more than an extension of the greater capitalist apparatus, alienating us more and more each day from the living world and from each other; it can be reclaimed as a vital part of human life. To invite ecological characteristics back into our technologies and design is to reintegrate them with the natural and social worlds.

Next Steps

In the Runrig Roadmap for 2024 we propose a 3-phase business model for a workers cooperative, which can then steward the development of the platform cooperatives and eventually "One Big Data Co-op," à la Bill Haywood and the Wobblies. Right now as a volunteer-driven group, we're in Phase Zero. Before taking any contracts or pursuing grants for the following phases, we're building out one or two reference implementations to test the viability of the model at each phase of development. We're holding open workshops every other week, along with community chat rooms and discussion boards, so anyone can get involved. As we progress, we're actively seeking partnerships and collaborations with similarly aligned organizations, and eventually hope to recruit farmers, food & tech workers and activists as coop members, so sign up for our newsletter to receive occasional updates.

The Runrig Plan is a living document that describes the main design principles of the social and ecological technologies we are building now and in the near future. In addition to this overview, two appendices are also presented now, though they are still in a very rough stage of draft. The "Architecture" page will go into more technical detail of the underlying software and networking systems involved in the implementation of Runrig, while the "Ecology" page will cover more of the social, environmental and economic aspects of the plan, as well as governance structures and legal considerations.


  1. Muldoon, James. Platform Socialism: How to Reclaim our Digital Future from Big Tech. p. 14-15. ↩︎

  2. Stallman, Richard. "Who does that server really serve?" ↩︎

  3. Tisne, Martin. "It’s time for a Bill of Data Rights" ↩︎

  4. Cyphers, Bennett and Cory Doctorow. "Privacy Without Monopoly: Data Protection and Interoperability". Feb 12, 2021. ↩︎

  5. Watterson, Chris. "The Success Of My 'There Is No Cloud' Sticker". ↩︎

  6. "Fault-tolerance". ↩︎

  7. Villa, Luis. "Free as in...?" ↩︎

  8. Solid Project. "Get a Pod". ↩︎ ↩︎

  9. Kleppmann, Martin et al. "Local-first software: You own your data, in spite of the cloud". Ink & Switch. ↩︎

  10. Scholz, Trebor. Platform Cooperativism: Challenging the Corporate Sharing Economy. ↩︎

  11. "The Food Zones Explained" ↩︎

  12. Data Food Consortium. "Our Standard". ↩︎

  13. OpenTEAM. "Access Tools and Support". ↩︎

  14. farmOS. farmOS Data Model. ↩︎ ↩︎

  15. The World Wide Web Consortium. "Semantic Web". ↩︎

  16. Solid Project. "Solid Technical Reports". ↩︎

  17. Solid Project. "Running your own Solid server". ↩︎ ↩︎

  18. Lemmer Webber, Christine. "ActivityPub is a W3C Recommendation". ↩︎

  19. Lemmer Webber, Christine. "Victory for libre networks: ActivityPub is now a W3C recommended standard". ↩︎

  20. Skywoman MAIA Project. "OSU Microfarm Project & Richland Gro-Op" ↩︎

  21. SurveyStack. "Survey software designed to empower shared community knowledge." ↩︎