The May Project will revolve around document formats & information interaction in XR. We are proposing using EPB as a wrapper for HTML benefits with PDF level compatibility. We call it ‘Origami Text‘ since it is folded into legacy form but can be unfurled into spatial and dimension text.
AI: Summary
This session brought together regular members and two newcomers — one joining from Mumbai, another from Moscow — for a wide-ranging exploration that moved from a live demonstration of Author on macOS and Apple Vision Pro, through a proposal for a simplified EPUB publishing format called Origami Text, and into a deep theoretical exchange on cognitive load, adaptive interfaces, and the relationship between knowledge, context, and spatial computing. The demonstration showed concept extraction, spatial node interaction, focus/unfocus gestures, and saved views in XR, while the format discussion examined how to make scholarly documents more accessible to both humans and AI by stripping EPUB back to semantic essentials and enriching it with structured metadata.
The session surfaced a fundamental tension between presenting information and supporting genuine cognitive development. One participant, a psychologist specialising in human–computer interaction, introduced Vygotsky’s zone of proximal development as a design principle for spatial interfaces: rather than merely adapting to a user’s current level, interfaces should be calibrated slightly beyond present comprehension — scaffolding the next cognitive step rather than flattening complexity. This reframes the entire project of spatial knowledge tools from information management toward developmental augmentation.
A provocative inversion emerged around AI-assisted reading: instead of the prevailing instinct to summarise and simplify, the call was made to reveal complexity. The request was not “give me the summary” but “show me the density.” This stands in direct opposition to the default summarisation paradigm of current AI tools, and suggests that the real value of augmented text may lie in making conceptual difficulty navigable rather than invisible.
A taxonomy of books by semantic density was proposed, distinguishing at least three categories: single-concept books that justify one central idea across hundreds of pages, hyper-dense texts where a single paragraph may contain dozens of interlocking ideas, and narrative or fictional works that demand a conversational mode of engagement. Each type calls for fundamentally different reading interfaces — extraction for the first, disassembly for the second, dialogue for the third. Current document formats and reading tools treat all three identically.
The distinction between knowledge and information was sharpened. Knowledge was characterised as inherently positional — dependent on viewpoint, role, activity, and ontological framework — whereas information is context-free. This means spatial arrangement of concepts in XR is not merely organisational convenience but a representation of epistemic stance. Different people looking at the same material would legitimately produce different spatial layouts, and sharing those layouts becomes a way of sharing perspective itself.
The idea of connecting virtual knowledge objects to physical artefacts in the room — recognising a specific book cover or notebook and anchoring metadata to it — was raised as a near-term possibility already achievable with current vision models. The privacy constraints of Apple‘s platform were acknowledged as the primary barrier, not technical capability.
The concept of “focus” as a gestural interaction — pinching to collapse everything unrelated to a selected node, then expanding outward again — was identified as a powerful mechanism for managing cognitive overload in spatial environments. The gesture was described as feeling like physically pulling and pushing knowledge, and was contrasted with toolbar-based interactions, which were compared unfavourably to Microsoft Word-style paradigms.
A distinction between outline and overview was articulated as significant for authorship workflows. An outline shows structural headings; an overview shows marked-up content alongside keyword columns. Collapsing these into a single view creates unnecessary cognitive load for writers who sometimes need only the skeleton and other times need the annotated whole.
Stand-off markup was proposed as an alternative to HTML-based annotation within EPUB, using nano-IDs attached to individual characters and sparse bit vectors for attribute layers. This would allow infinite overlapping annotations without the structural nightmare of nested HTML spans. The counter-argument was that Origami Text is a publish-only format — a replacement for PDF, not for Word — and must remain compatible with existing EPUB readers, which limits the degree of structural experimentation.
AI: Resources Mentioned
Tom Haymes, “The System Strikes Back” — https://augmentingintelligence.substack.com/p/the-system-strikes-back
Tom Haymes, “The Modern Punchcard” — https://augmentingintelligence.substack.com/p/the-modern-punchcard (described as parts 1 and 2 of an Engelbart trilogy, with part 3 still in progress)
Dave Millard session write-up — https://futuretextlab.info/2026/05/02/millard-1-may-2026/
Origami Text specification — https://visual-meta.info/origami-text/
Origami Text structure — https://visual-meta.info/origami-text-structure/
Douglas Hofstadter, Gödel, Escher, Bach — referenced as an example of extreme semantic density
Nassim Nicholas Taleb, Antifragile — discussed for its passage on apophatic knowledge: articulating what cannot be explicitly said
Yuval Noah Harari, Sapiens — cited as a single-concept book stretched across many pages
Harry Frankfurt, On Bullshit — mentioned as a short but significant work originally written as an essay
A revised edition of Bloom’s Taxonomy — referenced as a book containing one central idea with extensive justification
Margaret Livingstone — cited for her work on perception as information processing rather than image transmission
Santa Fe Institute — mentioned in connection with complex systems research
Bruce Horn — referenced as creator of the original Macintosh Finder and later Context
Dave Millard — referenced as PhD supervisor and collaborator on AI agent interface concepts
Vygotsky’s zone of proximal development — introduced as a framework for adaptive interface design