The reality is that in order to foster a richly connected, richly interactive —more liquid — knowledge environments where thinking is truly augmented and innovation to make increasingly power tools increases, we need document formats to support that. To see a brief scenario of what this would be practically, have a look at: Origami Text In Use.
The Need
We need a robust document format to present our thoughts with so that they may connect to the wider discourse and allow for interactive reading. Current PDF is robust and HTML is interactive. We need both attributes and we need to go beyond current limitations.
For Developers
The Origami rules are clearly, openly available without license, simple and available for anyone to implement, same as for readers who want to extend the capabilities of their EPUB readers. The process is to encode what is already metadata in the user’s original document into EPUB, such as headings, references, glossary and spatial information. The simple Structure is defined and notes on Implementing Origami Text export for developers is available, as well as for adding Origami Text for augmented Reading. The Origami Text Structure is clear and functionally minimal.
Freedom to Soar
An analogy is thinking about how an aircraft takes off, which is poetic when we dream of ‘flying’ in our information environments. Wings support the aircraft and gives it lift when of the speed of interaction between it and the air reaches a threshold. When the interaction is too slow the aircraft is not much different from a car.
We need a format shaped to interact — addressable parts, embedded structured data, flexible layout, openness to tools that haven’t been built yet, and this we can’t brute-force to generate lift, we must design for it. PDF and HTML by themselves cannot support such needed rich interactions with their environments in terms of user interaction, AI interaction and connectivity.
The Problem
Academic knowledge is trapped in formats that don’t serve it well.
PDF preserves visual layout perfectly but hides structure. A heading looks like a heading to a human reader, but to a machine it’s just bold text in a larger font. Citations are formatted strings, not structured data. There’s no reliable way for software to distinguish a bibliography entry from a bold paragraph. And PDF has no concept of spatial relationships between ideas — because it was designed to reproduce printed pages, not to represent thinking.
HTML is structurally rich and machine-readable, but it has no boundaries. A webpage is not a document you can hand to someone. It depends on servers, stylesheets, scripts, and connectivity. It’s alive, which is its strength, but that also means it can break, change, or disappear.
Neither format carries the intellectual structure of a document — the concepts the author defined, the spatial arrangement of ideas they constructed, the connections they discovered — in a way that travels with the work.
The Approach
Rather than invent a new format, the approach is for a simplified EPUB — the open, W3C-maintained standard that already combines the strengths of both PDF and HTML. An EPUB file is a single portable file you can email, archive, and keep forever, like a PDF. Inside, it’s semantic HTML that any machine can parse, like a webpage. It’s already supported by every major reading platform. What authoring systems can add is a profile — a set of constraints and one addition. The constraints keep the HTML clean and semantic.
EPUB so far has not solved the issues because it has not addressed academic concerns, which will need to be done for Origami text to work.
The new approach has the addition of Visual-Meta: embedded in the HTML itself and as a JSON file inside the package, carrying the document’s intellectual structure — its defined concepts, citation data, and the spatial layout of knowledge nodes that the author built while thinking through their work. The result is a document that works at every level of capability.
- Open it in Apple Books or any other EPUB reader and you read a well-formatted academic paper. Compatibility comes from simplicity.
- Open the HTML in a text editor and every word is legible, every citation is a readable string.
- Feed it to an AI system and the semantic structure is immediately parseable — headings are headings, citations carry BibTeX and CSL-JSON data, defined concepts are explicitly marked.
- Open it in an XR-capable reader and the author’s spatial knowledge map reconstitutes around you. Nothing breaks at any level. Each layer adds richness without requiring the layers below it to change.
Openness
EPUB is an open standard. HTML is an open standard. CSS is an open standard. JSON is an open standard. There is nothing proprietary in this approach.
Any developer can write software that reads an Author EPUB Profile document. Any developer can write software that produces one. The specification fits on a few pages and every element in it is built from technologies that have been openly documented for years or decades. This matters because knowledge that can only be read by one application is not knowledge that’s free to move. A student who writes in Author today should be able to read their work in thirty years with whatever software exists then. An EPUB file from 2026 will still open. The HTML inside it will still render. The JSON will still parse. The Visual-Meta spatial layout data, even if no reader exists that can render it spatially, is plain text that any future system can interpret. Visual-Meta itself is designed around this principle. As Vint Cerf wrote, it adds “an exploitable self-contained self-awareness”* to documents. The document knows what it is. It doesn’t depend on an external database, a cloud service, or a particular application to explain its own structure.
Robustness
The format is robust because it’s layered and because every layer degrades gracefully. If the CSS is missing or ignored, the document is still readable. If the Visual-Meta JSON is ignored, the document is still a well-structured paper with proper citations. If the MathML appendix is ignored, the equations are still legible as Unicode text in the body. If the cross-document links can’t be resolved because the referenced documents aren’t present, the nodes still appear in the spatial layout — they just can’t be opened. Nothing depends on everything else working. Each piece adds value independently. This also means the format is resilient to the passage of time. Formats that require specific rendering engines, network access, or proprietary parsers become unreadable when those dependencies disappear. An EPUB Profile document is a zip file containing text files. The absolute minimum requirement for reading it is the ability to unzip an archive and open a text file. That requirement will be met for as long as computers exist.
Flexibility
The same document serves different readers in different ways without the author having to produce multiple versions.
- A journal reviewer opens it in their preferred EPUB reader and sees a paper. A collaborator opens the HTML and copies citation data directly from the structured attributes on each reference.
- An AI assistant reads the Visual-Meta JSON and understands the document’s conceptual topology before reading the body text.
- A student in a headset opens it in XR and walks through the author’s spatial arrangement of ideas, seeing how concepts connect and where citations cluster.
The author exports once. The document adapts to its reader. This flexibility extends to citation style. The inline citations can follow author-date or numbered conventions — the author chooses at export time — while the structured data underneath remains the same. A reader or tool that wants to reformat the citations for a different journal style has everything it needs in the BibTeX and CSL-JSON attributes. It also extends to the spatial dimension. The Z-depth data from the author’s visionOS workspace is carried in the Visual-Meta without affecting the two-dimensional reading experience. A flat reader ignores it. A spatial reader uses it. The same file serves both.
A Small Change, Not a Large One
This is not a new format, it’s EPUB with discipline. The HTML is standard HTML with a restricted element set — fewer elements than a typical webpage, not more. The CSS is a single optional stylesheet. The package structure follows the EPUB 3 specification exactly. The only addition is a JSON file containing optional metadata that the author’s tool already generates, data which is also added to the HTML but not rendered visually. For our proof of concept application Author specifically, every piece of this already exists in the .liquid working format. The defined concepts are there. The citation data is there. The spatial layouts are there. The Visual-Meta structure is there. Export is a matter of translating what Author already knows about the document into the EPUB container — not of building new capabilities. For the wider ecosystem, adoption requires nothing. Any EPUB reader already opens these files. Any tool that reads HTML already parses the content. Any reference manager that accepts BibTeX or CSL-JSON already imports the citations. Awareness of the Author EPUB Profile adds capability but is not required for basic use.
The distance between where we are and where this takes us is short. The difference in what becomes possible is large.
What This Makes Possible
A researcher writes a paper in a writing system like Author. While writing, they define concepts, build a spatial map of their ideas, and discover connections they hadn’t anticipated. When they export, all of that thinking — not just the final text, but the structure of the thinking itself — travels with the document. A colleague reads the paper and imports the citations into their own reference library with a click. An AI system reads the Visual-Meta and can answer questions about how the author’s concepts relate to each other. A student in fifty years opens the paper in whatever XR environment exists by then and sees the author’s original spatial arrangement of ideas unfold around them. The paper is still a paper. It still prints, still reads in a browser, still opens in Apple Books. But it carries more than text. It carries the author’s knowledge space — folded flat for transport, ready to be unfolded by anyone with the tools to do so. That’s what this format is for. Not to replace what exists, but to let documents carry more of what their authors know.
— Remember : Revolutions seem impossible before they happen & inevitable after they succeed —