AI: Summary
January’s sessions for the Future Text research community were anchored by live demonstrations of Author running natively on the Apple Vision Pro in visionOS, using a pitch letter embedded with Visual Meta as the primary test object.
The month moved between historical perspectives on the origins of writing, active design critique of spatial document interfaces, and deep conceptual debate about what — if anything — three-dimensional spatial arrangement actually adds to reading and knowledge work. Running beneath all of it was an unresolved tension between the excitement of a genuinely new medium and the honest acknowledgement that the group does not yet know what problem it is solving.
AI: Main Topic
The central object of inquiry across January was Frode Hegland’s demonstration of a spatial citation and knowledge graph derived from a single PDF letter, opened in Author on the Apple Vision Pro and displayed as a plane of floating nodes with embedded BibTeX references and Visual Meta metadata. The demo was shown live to the group and then critiqued — first for its design, then for its underlying assumptions — culminating in recurring challenges from Mark Anderson, Ken Perlin, and Tom Haymes about whether spatial arrangement generates genuine epistemic gain or merely aesthetic novelty.
AI: Highlights
Barbara Tversky’s email correspondence was flagged as a key reframe: in standard HCI practice you identify the problem first, but what this group is doing is different — arriving with a new set of paints and asking what they can do. Not a problem looking for a tool, but a tool looking for its picture. This reframe was treated as significant enough to open one of the January sessions.
Vincent’s “incunabula” framing was repeatedly returned to as a touchstone: the current moment resembles the Gutenbergpress arriving in the scriptorium, and you cannot tell a monk how a printing press works — you have to put him in front of it. The demonstration is the message.
Ken Perlin’s design critique was highlighted as a substantive intervention: a flat wall layout is the wrong form factor for what the demo is trying to demonstrate. The interface should be egocentric, curved, and proximity-based. When a pinch gesture fails, the user has stopped thinking about knowledge and started fighting the environment — that is the wrong cognitive place.
The open question Frode left standing at the close of the month was treated as the central unresolved problem: once you have walked into the knowledge, what did you gain that you could not have gotten by staying flat?
AI: Insights
The oldest known recorded writing are product tags — commercial inventory markers, not literature. Brandel Zachernuk’s observation, traced back by Frode to a connection at the British Museum, reframes the entire history of text: writing did not precede commerce, commerce produced writing. This recasts the group’s XR work not as invention but as a return — cycling a very ancient pattern through a new medium.
The document-environment interdependence is not merely a design claim but an ontological one. Frode’s observation — that the same letter opened in a dark Norwegian virtual room versus a Bora Bora environment carries different meaning — implies that the document is not separable from the space it inhabits. The workspace is the message. This has significant implications for how Visual Meta and similar metadata standards are conceived.
Tom Haymes proposed that the Z-axis in spatial document interfaces maps naturally to citation depth: walk backwards through your sources, then their sources, giving the third dimension genuine semantic logic rather than arbitrary placement. This is a concrete design hypothesis that gives spatial arrangement intellectual justification.
Mark Anderson immediately challenged this: citation chains are not linear — they loop, cluster, and compound into noise. The Z-axis-as-citation-depth model may produce what Tom himself called “a visualization that needs a visualization to explain it.” This tension between semantic depth and navigational chaos at scale is one of the central unresolved design problems of the month.
Brandel’s reading of Track Changes by Matthew Kirschenbaum surfaced a rarely-stated insight: the physical environment in which writing occurs — the cast of light, the specific machine, the muscle memory of a particular setup — is a cognitive scaffold, not just a backdrop. The writing environment is part of the tool. This strengthens the case for XR reading environments as cognitive infrastructure rather than mere display.
Brandel also disclosed that he has aphantasia — no visual mind’s eye, only spatial awareness without color. Writing’s power, he observed, is that you don’t need to visualize a tree: the word alone reinflates the concept. This introduces an unexpected angle: spatial-but-not-visual thinkers may relate to XR text interfaces in fundamentally different ways than the neurotypical assumption underlying most interface design.
Ken Perlin argued against collapsing all reading contexts into a single spatial interface. Reading wants white pages; conversation wants presence in a room. The assumption that one unified XR environment can serve all cognitive tasks may be wrong. Different affordances serve different epistemic states.
The annotation ownership question was raised and left entirely open: when a spatial annotation floats in an XR environment, does it belong to the document, the environment, or the person who placed it? This is territory that flat-document annotation law and convention has not yet entered, and the group recognized it as genuinely novel.
Frode’s reflection on using ChatGPT to review a written email produced a subtle but important concern: the model responded “This is intellectually rich — and very you.” The insight was not flattery, but danger — even a compliment is a current underneath, and if the model’s weights are set outside your moral frame, the nudging operates below the threshold of conscious awareness. This connects AI feedback tools to deeper questions about whose values shape the writing process.
The recurring phrase “solution looking for a problem” — said both by Frode directly and echoed in Barbara Tversky’s email — landed differently each time it was spoken. By the end of the month the group had collectively reframed it: at the incunabula moment, this is the correct and only honest methodology. You do not know what the printing press is for until you put the monk in front of it. Not knowing is not failure; it is the precondition for discovery.