# A Brief Intro to Knowledge Organisation Infrastructure
*This piece introduces and explores the concept of Knowledge Organisation Infrastructure (KOI). KOI is a research effort to enable the self-organisation of knowledge at scale by addressing technical and conceptual shortcomings of traditional approaches to knowledge management. This research draws from BlockScience's experience in complex systems, cybernetics, governance, and knowledge representation.*

#### Why Infrastructure?
By framing the problem as an infrastructural one we shift the focus from singular, self-contained systems (such as "knowledge management systems") to a more comprehensive foundation that supports the organization of knowledge across multiple interconnected systems, organisations, and contexts. This perspective considers that the production and organisation of knowledge has complex dynamics which cannot be controlled from any central point of authority, nor can the problem be solved by any one system or piece of software.

The success of traditional and contemporary approaches to knowledge management is contrained by the infrastructure on which it is built, making infrastructural improvements both a bottleneck and a point of high-leverage. In practice, this means that new approaches, mechanisms, and systems must address the needs of knowledge organisation *in general* and be applicable in a manner that matches the decentralised nature of real knowledge.

#### Research Goals and Progress
The goals of this research can be broadly divided into two areas. Some progress towards these goals has been included and will be expanded upon in more depth in the future.

**Theoretical:** Improve our ability to reason about knowledge organisation. This involves tackling basic questions of knowledge organisation such as:
- When we talk of knowledge organisation, what are the *things* which are actually being structured and organised?
>This question has resulted in a pragmatic and generalised model of knowledge objects called *"Objects as Reference"* — that these objects (and in fact all digital objects) exist through the ability to *refer* to them. The category of objects thus includes members as diverse as unicode characters, PDF files, UI buttons, calendar events, and Python functions. This definition means that objects are actually a *relation* between a *reference* and a *referent* — where a reference uses some means of reference to refer (such as an ID, filepath or CSS selector) and a referent (the thing referred to) can be anything from a single string of bytes to an abstract concept which exists only in the minds of people (the *socio* in a *socio*-technical system). This definition alone only goes so far but has lead to ways to characterise these objects and reason about the ways they can be organised.

- As knowledge is produced, organised, communicated and interacted with, what dynamics govern the observed evolution of this knowledge and its organisation? What changes can we make to these dynamics and how can these changes be observed and measured?

**Technical:** Informed by a conceptual understanding of KOI, improve our ability to structure, reason, and interact with knowledge. This involves, among other things:

- Developing generalised approaches to representing structure which allow knowledge to be organised in pluralistic manner (i.e. enabling local differences in organisation without conflict)
>The need for pluralism and local differences has lead to a clear infrastructural requirement: The separation of *reference* from *structure* at the infrastructure level. In short, this comes from an observation that tight coupling between knowledge objects and their structure (such as referring to files by their location in a tree) creates a hard limit on organising these objects in other ways.
- Creating new modes of interaction for organised knowledge which minimise required knowledge of how it is organised
>Recent experiments with Large Language Models within BlockScience have provided a tantalising glimpse into one possible approach to interacting with freely structured knowledge. However, text is not by itself a sufficient medium for knowledge as a whole, so other approaches (such as automated layout of graph-like structures) is also being explored.
- New ways for individuals and organisations to structure knowledge *collectively* over time - this requires new applications of digital governance as well as mechanisms which can adequately represent the expressed intent of the people involved
>This goal has lead to an approach to managing object *identity* (the way in which something is the same over time or space) which replaces implicit authority with social agreement. When a knowledge object is mutable, such as a document which evolves as people make edits, it is assumed that the identity of this document has some kind of continuity over time (e.g. that a document titled "Project X Roadmap" shares an identity with its past and future versions). However it is not the name (this could change to "Plan for Project X") nor the reference (such as a Google Doc URL, which could refer to something entirely different if someone were to replace its content) that defines its identity — instead it is the fact that the people involved *agree* that the identity of the document is consistent. This observation has motivated a governance-based approach to identity which will be explored over the summer and aims to address a basic infrastructural shortcoming as well as solving some pragmatic needs within the BlockScience organisation.

#### The BlockScience KMS
Over the last year BlockScience has been developing an internal system for knowledge management which aims to be a pragmatic solution to the needs of BlockScience researchers while also acting as a testbed for mechanisms and system designs which over time can be dissolved into infrastructural components.

This system focuses on *research artifacts* such as writeups, research papers, working documents and internal communication. Instead of creating a new centralised platform to store these artifacts, the system connects to the tools that BlockScience already uses such as HackMD, Google Docs, Slack, Github, etc. and tracks the content of these artifacts so that they can be searched, queried, and organised.


# WIP Sections/Paragraphs

#### on knowledge in orgs (part of why infra)
As organisations and institutions go about their work they produce digital objects like messages, plans, working documents, design files, code, descriptions of procedures and roles, and an endless number of other diverse objects. These objects embody the knowledge of the organisation and become an essential part of how the organisation perpetuates itself. These diverse objects become essential to the operation of an org and its continued existence and cohesion as membership evolves over time. This is doubly true for decentralised organisations, which must maintain cohesion without centralised authority.

These objects are at the center of institutional, organisational, and individual knowledge and yet they appear so disparate that attempts to improve their organisability almost always takes the form of a new software service or product, like a project management, notes, or task manager app. When these solutions fall short, we invariably see new software which tries to fill the gap or big monolithic tools which attempt to cover ever more use-cases.

These solutions result in walled-gardens, limited interoperability, and a continued effort to reimplement basic features found in similar tools. While each new problem is perhaps solvable with enough perseverance, there is an alternative perspective which can help resolve the contradictions of the organisation-as-software paradigm.

#### more detail on objects
Will pull from WIP "Objects as Reference" paper

#### What comes next?
... might not write this section in the end, leave it for a future piece

## NOTES (to be deleted)
- title TBD, recommendations welcome
- reference fields it touches: knowledge representation, knowledge management systems, CAG, cybernetics
- framings and concepts: cyberinfrastructure, provenance, self-organisation, cybernetics, inter-org knowledge sharing and synthesis

## call with Ilan
- more detail about the conceptual objects
- organisations and institutional artifacts, how do decentralised organisations retain cohesion without centralised authority
- speculate about further possibilities at the end, what are these first steps, how do they connect to what can come next? what is *exciting* about this work?
- organisation is getting harder
- orgs representing knowledge lets them reason collectively