Knowledge Networks, LLM, BlockScience, Knowledge Management, Information Infrastructure

Knowledge Networks and the Politics of Protocols


Harmony without loss of Variety; 
Variety without loss of Harmony.

Building tools that allow users to provision, operate, maintain, and govern their own instances of an application is one of the most politically potent acts possible in open-source software development. In essence, doing so creates a new commons:  a shared forum within which communities of users can act, interact, and organize. Although the developers of such tools may intend for them to be broadly accessible, the tools themselves can never be truly politically neutral. Developers inevitably have to make subjective decisions that have downstream consequences, structuring and/or delimiting the action space available to users – hence cyborg anthropologist Amber Case’s claim that “design is governance.”1 The relationship between these design choices and their eventual outcomes, however, is often invisible to both software users (who have limited insight into the development process) and developers (who are often too far removed from specific instancing processes to recognize the ways that their decisions have shaped such processes in advance). 

In order to better understand the downstream consequences of decisions made in the course of developing open-source software, BlockScience is launching a new research initiative in collaboration with The Metagovernance Project – a non-profit known colloquially as “Metagov.” The goal of this initiative is to identify a “sweet spot” between designing for configurational freedom and designing for user-friendliness, so that the process of protocolization does not force users to choose between software that is useful and software that they can use. Furthermore, we hope to discover how to strike this balance in a way that preserves interoperability across instances, standardizing the dimensions that need to be standardized while allowing variability in those that do not.

These questions have become particularly urgent in light of the rapid emergence of Large Language Models (LLMs). Foundational models do not readily lend themselves to local instancing, and insofar as they can be fine-tuned locally, the existing processes for doing so are resource-intensive and lead to unstable instances. It is nevertheless common for foundational models to be used as components in applications, which results in those applications subtly recapitulating the politics hard-wired into the foundational models themselves: their fundamental assumptions, the ways they facilitate or discourage various forms of expression, how they configure and construe the space of possible action. These hidden politics are at work behind the scenes of any infrastructure that involves LLMs; there is thus an urgent need to better understand the political dimensions of the design trade-off between ease of use and degrees of user freedom in knowledge infrastructure, lest the advent of AI usher in a brave new world in which “user-friendliness” and “freedom" are mutually-exclusive properties. 

Our Research Question

Our research question is straightforward: 

How can developers ensure that user-instanced open-source software is customizable enough to meet the needs and reflect the values of heterogeneous communities, but also sufficiently standardized to allow interoperability across instances – all while remaining easy to maintain?  

This frame captures our investigation’s orientation – toward better understanding the politics latent in the design of user-instanced digital infrastructure – and its animating purpose – enabling developers to minimize the prescriptiveness inherent to protocols, and thus preserve the degrees of freedom required for users to customize instances that meet needs and represent values local to the context in which they are being deployed. 

Issues around entrenching politics in software are particularly salient when the digital infrastructure in question serves as the foundation for “knowledge commons,” or inter-organizational information-sharing networks – where the borders between representation, expression, and agency are blurry at best.  

By focusing on how organizations with different ontologies and values configure, govern, and experience peer-to-peer relationships with other instances of the same open-source application, we aim to revolutionize the collaboration between small specialized research entities, fostering collaborative ecosystems where specialized knowledge is mutually legible amongst diverse organizations. 

Methodology

Because of the typical separation between the developers of user-instanced open-source software and the user communities that provision, operate, maintain, and govern specific instances thereof, investigating our research question requires an unconventional and multifaceted approach. This approach is rooted in the realization that BlockScience’s ongoing work developing open-source digital infrastructure is itself an ideal “field site” for anthropological and ethnographic research concerning the development of open-source software.2 Our methodology, therefore, is fundamentally self-reflexive – researchers at BlockScience are engaged in a project of the type that we want to study, and we have decided to approach that work in ways that allow it to serve as a springboard to broader insight into topics of general interest to the community of open-source software developers writ large.

The project at the center of our inquiry is perfectly suited to this purpose: BlockScience has developed an open-source knowledge management system (KMS) and knowledge organization infrastructure (KOI) for our own internal use3, employing an LLM as the interface to a database of our organizational knowledge, to anchor interactions within the scope of the content of the knowledge database and thus foster a genuine user interaction. The inclusion of adjustable parameters and legible configuration interfaces is crucial for maintaining the balance between freedom of expression and user-friendliness, ensuring that organizations are not unduly constrained by the idiosyncrasies of their tools’ creators.

When members of The Metagovernance Project learned about BlockScience’s KMS, they wanted to know if we could protocolize it such that their organization could operate its own instance – a version of the same underlying system that is responsive to their organization’s specific needs, and reflective of their organization’s particular values. We accepted the challenge, and realized that, in doing so, we would have an opportunity to investigate some more general questions about the politics of protocols. Thus, the aim of our new collaborative research initiative involves protocolizing software that has already been developed – but also interrogating the processes of development and protocolization in order to shed new light on the ways that they invisibly parameterize the space of political possibility.  

Our methodology will thus include an extensive literature review, context mapping, ethnographic research, an exploration of technical implications, a research paper and public panel discussion presenting our findings – which, taken together, will provide a deeper understanding of the essential considerations involved in designing open-source software that users are meant to instance for themselves. The scope of our research will range from theoretical explorations and contextual understandings to experiential investigations and architectural refinements. We will focus on discerning and enhancing the intersection between organizational self-expression, interoperability, and user-centered design within a user-instanced open-source software environment. The insights we derive will contribute to the broader discourse on open digital infrastructure, network protocols, and collaborative information ecosystems. 

Why Collaborate?

This research grows out of an existing collaboration between BlockScience and The Metagovernance Project (Metagov), a vibrant, interdisciplinary, nonprofit collective of researchers investigating the governance of online communities. As a diverse community of over a thousand members – who come from disparate backgrounds, and are pursuing a wide array of research initiatives at any given time – Metagov is currently confronting challenges that are endemic in the broader research environment; in particular, they are grappling with the need to develop a unified language to articulate their organization’s diverse explorations, both within and beyond the borders of their community. 

Partnership with Metagov will be central to this project. BlockScience has a longstanding positive working relationship with Metagov; many of its team members are active contributing researchers within the Metagov community, and the company’s founder, Michael Zargham, is also a member of Metagov’s Board of Directors. Metagov is an ideal partner for this work because their community has an interest in the questions around which it centers (in addition to their concrete desire for the software that we are developing). Formulating rigorous questions about trade-offs in the design of open-source software requires high-resolution visibility into, understanding of, or involvement in the development process itself, and Metagov’s members will be able to engage directly in that process through our ethnographic research – while conducting their own inquiries into questions concerning the governance of their KMS instance, and investigating whether it can lead to novel research approaches.

Furthermore, the Metagov community’s interest in our research question means that there will immediately be an audience eager to explore our findings – and to interrogate our discoveries, providing a vital dialogue through which we can more precisely refine our conclusions. As mentioned previously, at the conclusion of our research we plan to hold a public panel discussion about our project facilitated by the Metagov community, which will be open to the general public (and which we hope to record and share online). Metagov’s membership includes community leaders from all over the world, thus ensuring our findings will be available to those best positioned to do something with that knowledge. 

Additionally, we will draw attention to our research through less formal public channels, such as blog posts and short videos on our YouTube channel, to ensure that those who might be interested in our work know what we have been studying – and where to find the fruits of our investigation. Metagov’s focus on digital governance allows us to collaboratively explore how to generalize the KMS architecture so that it can be safely shared with more vulnerable communities, who often cannot afford to experiment with infrastructure that may or may not put them at greater risk.

Finally, we hope to extend our research in this area after the conclusion of the proposed project. We will continue sharing our digital civic engineering praxis, and refining our open-source KMS software to engender a public network of knowledge commons. 

Thus, Metagov offers an ideal context within which we can both implement and observe the impacts of our research into the ways that automated knowledge management systems can facilitate alignments within varied research communities, bridging divides and enhancing coherent dialogues, while respecting the uniqueness of each participant’s position within those dialogues. This inquiry, in turn, opens onto the more general question of how digital infrastructure meant to synchronize the variances between disparate organizational landscapes can be designed in ways that make it possible to preserve the distinctness of the different organizations involved. 

Putting “Self-Infrastructuring” Into Practice

This research implements the concept of “self-infrastructuring” developed by BlockScience team member Kelsie Nabben in her doctoral dissertation4, which builds on the scholarship of Bowker and Star5

We define self-infrastructuring6 as the ability of an organization to collectively design, own, govern, and maintain its own infrastructure. This concept underscores the significance of entities provisioning and operating their digital infrastructure, which is especially pivotal for entities whose ideals and objectives markedly diverge from those of external infrastructure providers. Due to the economics of scale, private digital infrastructure providers disproportionately serve majoritarian interests. Self-infrastructuring is a crucial method in prefigurative politics, enabling unparalleled freedoms in expressing unique values and goals – and therefore operates as a mechanism for expanding the field of political possibility, as it renders a wider range of positions susceptible to representation. These benefits, however, bring a host of distinct operational and governance challenges along with them. 

Technological self-expression is not merely instrumental – it is essential to an organization’s ability to define and pursue its own interests. As a concept, “self-infrastructuring” stresses the imperative of fostering connections between disparate entities without imposing a unified value system or shared goals on those entities, thereby mitigating the potential coercion inherent in conventional infrastructural setups. Moreover, it unveils the inherent trade-offs between self-expression and operational ease, a dynamic that often propels entities to cede self-expression in favor of the convenience offered by Software as a Service (SaaS) tools.

Our research endeavors to operationalize these insights by exploring the aforementioned trade-off space within open-source software designed to facilitate the process of self-infrastructuring for communities of users. This exploration is poised to render self-managed infrastructural models viable for small research entities, thereby facilitating a more inclusive and diversified infrastructural landscape. By focusing on the delicate balance between user autonomy and system operability, our research aims to extend the benefits of infrastructural self-expression to a wider array of organizations, particularly emphasizing those with distinct and varying objectives and values.

In essence, the move to practice involves applying insights from the self-infrastructuring paradigm to the process of protocolizing open-source software, with the goal of enabling instances to interoperate while also adhering to (and representing) the divergent values and goals of different groups – and of different individuals within those groups. We believe that scrutinizing and navigating the trade-offs between self-expression and usability, with the aim of democratizing access to customized infrastructural solutions, will enable organizations to transcend the constraints of one-size-fits-none models prevalent in conventional SaaS offerings. This endeavor resonates with the aspirations of smaller entities seeking to maintain their unique identity and operational ethos, while navigating the complexities of the contemporary digital landscape.

Why This Work Matters

Broadly, our research casts its net over communities engaged in open-source science and project development. We aim to instill a refined understanding of the inherent trade-offs involved in standardization and parameterization decisions, underscoring the importance of recognizing the political and value-based nuances embedded in infrastructural developments. The objective is to enlighten developers about the value implications of their design choices, fostering a more reflective and intentional approach to the incorporation of values and the allocation of user freedoms within the code, enhancing the overall cognizance and deliberation in the creation of user-centric open-source projects. Special interest groups are especially affected by design decisions entrenched in knowledge infrastructure.

More narrowly, our research is meant to benefit small research organizations engaged in highly specialized scientific explorations. Our inquiry into open-source software protocolization is meant to facilitate scientific communication across specialized disciplinary dialects, offering a harmonious platform for interdisciplinary interoperability without the need to impose a restrictive universal language or framework. Our aim is to untangle the intricate web of specialized communications, paving the way for a more inclusive and interconnected scientific discourse – one that embraces the rich tapestry of specialized languages, while also fostering mutual understanding.

With developments in AI driving a rush to automate an ever-increasing share of human endeavor, it is absolutely critical that we, as a society, sharpen our understanding of how to navigate the tradeoffs between automation (and the standardization that it requires) and the capacity to express dissident positions – or how to understand one another, even if our priors are wildly divergent. It is not an exaggeration that this research – and research into self-infrastructuring, in general – implicates those very questions; indeed, insofar as we are investigating the technical infrastructure required for organizations to disagree agreeably, our ultimate aim is to contribute to the great collective project of engineering a solid foundation for a more functional civic discourse. 

Defining Success

Our initiative will be successful if, at its conclusion, we have reached a richer understanding of the trade-offs between customizability and standardization in the specific context of knowledge commons and open-source AI-enhanced knowledge management software, and succeed in disseminating those insights to community leaders in a position to put them into practice. We hope to offer other organizations the resources required to discuss and develop infrastructure that represents their values, politics, beliefs, and goals, not ours – and to do so in a way that allows various representations from various organizations to be put into productive conversation with one another, rather than seeking to drown one another out. Ultimately, this work is about clarifying how open source digital infrastructure allows different organizations to organize themselves and their activities around non-identical worldviews, while retaining the ability to make common cause, or otherwise coordinate and collaborate in the world we all share. 

We hope to discover new ways of acknowledging and respecting dissenting perspectives within (and differing approaches to) a shared field of action, such that organizations with different animating purposes and models of the world can work together without having to synthesize their positions, beliefs, and goals. We seek to better understand the politics of infrastructure by rigorously examining how the process of open-sourcing our KMS infrastructure requires us to tease out the political assumptions of our own that we have unwittingly embedded within it; we also aspire to develop the kind of infrastructure that could make it possible to build bridges across the widening divides that seem to characterize the politics of today, insofar as it is built around an acknowledgment that respect for different positionalities is the prerequisite for the possibility of joint endeavor. Ultimately, our aim is to equip organizations with the tools they need in order to represent themselves, both to themselves and to others, with enough clarity that they can work together without compromising their unique identities. 

This research initiative is animated by a belief that open digital infrastructure should be treated as a public good, and our proposed work is meant to democratize access to digital resources and to catalyze innovation through collective societal engagement. By investigating the nuanced trade-offs between standardization and customization, we intend for our project to contribute to the sustainability, longevity, and adaptability of open digital infrastructure. 


Acknowledgments

This piece was authored by Michael Zargham, Ilan Ben-Meir and Kelsie Nabben, with inputs from Orion Reed, David Sisson, and Amber Case. Special thanks to Ellie Rennie, Eugene Leventhal, and Luke Miller from the Metagovernance project.

Footnotes

1. “How Design is Governance”, Amber Case, 2023. https://uxdesign.cc/how-design-is-governance-7c8dd466d753

2. “A Language for Knowledge Networks”, Michael Zargham & Ilan Ben-Meir, 2023. https://blog.block.science/a-language-for-knowledge-networks/

3. “Constituting an AI: Accountability Lessons from an LLM Experiment”, Kelsie Nabben, 2023. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4561433

4. “Decentralised Technologies: ‘Self-Infrastructuring’ Resilience”, Kelsie Nabben, 2023. https://www.admscentre.org.au/kelsie-nabben/

5. “How to Infrastructure”, Susan Leigh Star & Geoffrey C. Bowker, 2010. https://doi.org/10.4135/9781446211304.n13.

  1. “Web3 as ‘self-infrastructuring’: The challenge is how.” Kelsie Nabben, 2023. https://doi.org/10.1177/20539517231159002

About BlockScience

BlockScience® is a complex systems engineering, R&D, and analytics firm. By integrating ethnography, applied mathematics, and computational science, we analyze and design safe and resilient socio-technical systems. With deep expertise in Market Design, Distributed Systems, and AI, we provide engineering, design, and analytics services to a wide range of clients including for-profit, non-profit, academic, and government organizations.

You've successfully subscribed to BlockScience Blog
You have successfully subscribed to the BlockScience Blog
Welcome back! You've successfully signed in.
Unable to sign you in. Please try again.
Success! Your account is fully activated, you now have access to all content.
Error! Stripe checkout failed.
Success! Your billing info is updated.
Error! Billing info update failed.