Metasphere

The concept of Metasphere emerged from a simple but powerful observation: the way we interact with digital information feels fundamentally disconnected from how our minds naturally work. I noticed this disconnect during my time as a lecturer in information design and emerging technologies in 2014/2015. While teaching students about maps, clustering algorithms, and pattern recognition, I became increasingly fascinated by how we process and navigate information.

Traditional interfaces, with their rigid hierarchies and office metaphors, fail to use our innate spatial cognition and pattern recognition. I kept circling related ideas while most of my work was still branding and design systems: experimenting with parametric systems, and noticing how procedural generation can produce more organic, legible visual structure.

Some early experiments with clustering and gravitational relationships to influence corporate design systems (ScaleUp, 2011) laid the groundwork for what would later become core principles in Metasphere. The teaching experience that followed helped crystallize and concretize these concepts, as I observed how students naturally gravitated toward spatial and visual metaphors when trying to understand complex information structures.

Historical Foundations

The vision of Metasphere builds upon a rich history of pioneering work in hypertext and information organization. Vannevar Bush’s 1945 essay “As We May Think” introduced the concept of the Memex, a hypothetical device that would allow users to create and follow associative trails through vast information spaces. Ted Nelson’s subsequent work on hypertext systems expanded this vision, introducing concepts like transclusion and bidirectional links that remain relevant to modern knowledge management.

These early visionaries recognized that traditional, hierarchical methods of organizing information failed to capture the associative nature of human thought. Their work laid the foundation for thinking about information spaces as networks rather than trees, an insight that deeply influences Metasphere‘s approach to knowledge organization.

Ubiquitous Structures

The core of the Metasphere concept stems from the observation that the most common structure in the universe as we know it is hierarchical networks embedded in clusters. Networked structures are the fundamental organizational form, repeating in fractal patterns from the macroscopic to the microscopic. What we perceive as hierarchies are often just useful abstractions within these more complex networked structures.

This pattern reveals itself at every scale of observation:

• At the cosmic scale, we see it in the cosmic microwave background radiation’s filamentary structure
• At galactic scales, the way galaxies form into clusters and superclusters along cosmic filaments
• At stellar scales, in the distribution of stars and their gravitational relationships
• On our planet, in the branching patterns of river systems and mountain formations
• In human settlements, where cities cluster along coastlines and transportation networks
• In our biology, from the branching structure of our lungs to the organization of neurons in the brain
• At the cellular level, in the networks of protein interactions that govern cell function

These self-similar structures can be observed in informational networks as well as biological ones. The equally recursive nature of language and our symbolic thinking leads us to believe that a spatial navigation paradigm is suitable for navigating complex sets of information.

Remarkably, our memories appear to be organized in similar fractal patterns. Research in neuroscience suggests that our memories aren’t stored as discrete files but exist as networks of associations that activate in patterns. We recall information through associative pathways, moving from broader concepts to specific details in ways that mirror how we navigate physical spaces. This is why spatial metaphors for memory (“it’s on the tip of my tongue,” “let me explore that idea,” “I lost my train of thought”) are so pervasive and intuitive.

Even though the network metaphor is far better suited for organizing knowledge, we lack the capability to easily comprehend and navigate network structures at a glance. Our brains aren’t wired to directly perceive complex multi-dimensional networks in their raw form. Luckily, there is a visual representation that encompasses the same characteristics. And we have been trained to read and understand it for centuries.

Throughout history, humans developed a powerful tool to navigate these complex structures: maps. Maps allow us to externalize spatial relationships, compress multidimensional information into comprehensible forms, and navigate efficiently between points of interest. The art of cartography evolved to help us understand and traverse physical terrain, and now we can apply similar principles to navigate information landscapes.

This natural affinity for spatial metaphors is evident in how we talk about information in everyday language. We constantly use spatial terms to discuss abstract concepts: we explore “domains” of knowledge, travel down “information highways,” navigate “cyberspace,” and discuss different “spheres” of understanding. We create “mind maps,” climb “mountains of data,” and sometimes feel like we’re “drowning in seas of information.” We follow “routes” through arguments and gather at “hubs” of knowledge. This pervasive spatial language isn’t coincidental — it reflects how deeply our understanding of information is grounded in our spatial cognition.

This recognition of universal patterns provides a powerful foundation for the Metasphere approach. By aligning our interfaces with these ubiquitous structures, we create systems that feel intuitively “right” to users because they mirror patterns they’ve encountered throughout their lives across different scales and contexts.

What’s particularly interesting about this evolution is that the initial intuitions about spatial interfaces and pattern recognition came well before I encountered the neuroscience research that would later validate these approaches. The work of Barbara Tversky and Jeff Hawkins on the spatio-temporal nature of cognition and the hierarchical structure of information processing would later confirm what I had intuitively sensed: that our minds are fundamentally wired to think in spatial and temporal patterns.

This convergence of practical experience and theoretical validation led to a crucial insight: if we could navigate information as intuitively as we navigate physical spaces, we might be able to create interfaces that feel like natural extensions of our cognitive processes rather than artificial constraints we must adapt to.

The fundamental premise of Metasphere is that knowledge acquisition isn’t a linear process of accumulating facts. When we encounter new information, we don’t just store it — we test it against what we already know, follow associations, revise frames, and gradually fit it into a wider network of concepts.

This understanding led to three core principles that guide the Metasphere vision:

1. Embodied Cognition: Our thoughts, feelings, and behaviors are deeply intertwined with our physical experiences. We use spatial metaphors to describe abstract concepts not just as linguistic convenience, but because our cognitive processes are fundamentally grounded in spatial understanding. This suggests interfaces should use our innate spatial cognition rather than fight against it.

2. Visual Variance and Pattern Recognition: Human pattern recognition abilities, particularly our tendency toward apophenia, can be used deliberately through visual differentiation. By creating information landscapes with meaningful visual variance — mimicking real topographical features like peaks, valleys, and shorelines — we can support more intuitive navigation and understanding of complex information spaces.

   Tapping into Apophenia

   Apophenia, the human tendency to perceive meaningful patterns in data (even when such patterns do not objectively exist), is a cognitive bias that has aided our survival. Our ancestors who could quickly identify threats in their environment were more likely to survive and pass on their pattern recognition abilities.

   This instinct is utilized in Metasphere‘s topographical information maps, where conceptual features and relationships are shaped into landscape-like visual clusters. These clusters mimic the way ideas self-organize according to semantic relationships, much like how villages form around areas of geographic importance. Visual anchors such as shorelines, peaks, valleys, and slopes allow users to segment concepts and navigate the idea space intuitively.

   Strategic Visual Variance

   The key to this approach is the intentional introduction of visual variance in the data representation. In information visualizations, a certain degree of intentional variance is necessary to stimulate the human visual system’s contrast-dependent perception.

   Intentionally increasing visual variance by mimicking real topological features allows for visual delineation between conceptual clusters. It leverages our innate capacity for visual apophenia, highlighting the patterns in the semantic data we seek to discern. Users can orient themselves more easily and intuitively decode conceptual relationships nested in the information landscape.

   This approach also reinforces pattern recognition abilities over time. Users gradually memorize “their” map, quickly finding clusters of information because they become familiar with what specific terrain looks like. The approach therefore not only enhances immediate orientation but also facilitates long-term information retrieval.

3. Progressive Disclosure through Semantic Zooming: Just as we navigate physical spaces at multiple scales — from city maps to street views — knowledge interfaces should support fluid movement between different levels of abstraction. This is achieved through semantic zooming, where the level of detail and type of information displayed adapts to match our current cognitive needs and context.

   Text Zooming Implementation

   The implementation of this principle in Metasphere takes the form of a spatial hypernym zooming system:

   • At the deepest zoom level, the full verbatim text is displayed.
   • As the user zooms out, the text is summarized into key details and supporting evidence using extractive summarization techniques.
   • At even higher zoom levels, the core ideas and main topics are exposed as the text is reduced to topic-level summaries.
   • Further zooming provides a high-level thematic overview, collapsing the text to focus only on the central themes and concepts.
   • At the highest zoom levels, the user is presented with a broad bird’s-eye perspective showing relationships between documents and the overarching topics in a corpus.

   This staggered reveal of textual detail lets users navigate hierarchies of understanding. It supports focused reading by exposing relevant passages, while still enabling high-level orientation and concept mapping. The dynamic zooming interaction paradigm is designed to surface nuance without forcing a linear read.

These principles come together in Metasphere‘s core vision: an interface that turns complex information into navigable, dynamic landscapes. Like a city that develops organically around areas of importance, these information landscapes evolve based on usage patterns and semantic relationships. Frequently accessed concepts form peaks and clusters, while less visited areas settle into valleys or plains. Users can actively shape this terrain, creating landmark formations that serve as visual anchors for future exploration.

I worked on the original concept of Metasphere during the cryptocurrency boom of 2017–2018, when blockchain technology promised new possibilities for decentralized systems. The original vision positioned Metasphere as a decentralized knowledge management network that would combine novel visualization approaches with economic incentives for knowledge creation and curation.

This early version proposed using blockchain technology to create a bi-directional micropayment system that would reward both content creators and active readers. The idea was to incentivize not just the creation of content, but also its thoughtful consumption, annotation, and integration into the broader knowledge network. While the cryptocurrency aspects of this vision may now seem tied to that particular moment in time, the core insight about needing better incentive structures for knowledge work remains relevant.

The theoretical principles of Metasphere found practical application during the early stages of the COVID-19 pandemic. At the beginning of 2020, I formed a research group to investigate the technical aspects of the concept. The goal was to apply it to a dataset of papers related to the beginning surge of research into COVID-19. In order to help researchers make sense of the vast amount of literature, we worked on the extraction and embedding of textual features.

Working with an international team, we developed tools to help researchers navigate the rapidly growing corpus of COVID-19 research papers.

Using the CORD-19 dataset of approximately 30,000 articles, we implemented clustering algorithms and procedural landform generation techniques to create navigable maps of the research landscape. The project demonstrated how Metasphere‘s principles could be applied to real-world information management challenges:

• Semantic Clustering: Papers were grouped based on thematic similarity, creating natural “neighborhoods” of related research
• Topographic Visualization: Clusters were represented as terrain features, making the research landscape intuitively navigable
• Multi-scale Exploration: Researchers could zoom from high-level overviews to detailed paper analysis

Implementation Challenges

At the time when the concept was conceived, pretrained transformer models weren’t commonly available. Embedding models didn’t perform as neatly as they do today, and AI as a service didn’t exist. Despite these limitations, we built a prototype that mapped a dataset of roughly 20,000 papers onto a vector tileset in Mapbox, letting researchers explore similarities between papers.

What initially felt like technical limitations ultimately provided valuable insights into the core principles of spatial knowledge interfaces. The constraints forced us to focus on fundamental user experience challenges rather than getting lost in technical possibilities, helping to clarify which aspects of the spatial paradigm were most essential.

This project validated key aspects of the Metasphere vision while highlighting the practical challenges of implementing spatial knowledge interfaces at scale.

The path from early intuition to prototypes has been one of testing and revision. What began as a simple observation — that digital interfaces often mismatch how cognition actually moves — kept reopening into a broader question: how do we build tools that align with perception, memory, and attention rather than constantly fighting them?

While the original vision of a blockchain-based knowledge network may not have materialized as initially conceived, the core principles of Metasphere have proven remarkably resilient and valuable. The success of projects like ECCHR Explore, Wonder, and Trails has shown that these ideas resonate beyond theory — they address real needs in how we work with and understand information.

The COVID-19 research mapping project demonstrated both the practical utility of these concepts and their potential for scaling to handle large, complex information spaces. It’s still early, but the principles underlying Metasphere — embodied cognition, visual variance, and progressive disclosure — offer a foundation for designing interfaces that feel less like tools we must learn to use and more like natural extensions of our cognitive abilities.

As we move forward, the challenge will be to maintain this human-centered approach while embracing new technological possibilities. The goal remains the same: to create interfaces that don’t just store or display information, but actively support and enhance our natural ways of thinking, learning, and understanding.

Our work on metasphere has profoundly influenced my approach to interface design and knowledge management, serving as a foundation for numerous subsequent projects. Its principles of spatial cognition, visual variance, and progressive disclosure have shaped how I conceptualize human-computer interaction. The concepts first articulated in Metasphere — from spatial navigation to semantic zooming — have informed projects like ECCHR Explore‘s knowledge graph visualization, Wonder‘s collaborative spatial environment, and most significantly, the auto-associative workspaces of Trails. This foundational work has ultimately led to the development of more intuitive, context-aware knowledge management systems that better align with our natural cognitive processes.

The Metasphere project was made possible by the collaboration of a multidisciplinary team: