Office Hours 201 — Meeting Notes

AI: Summary

This inaugural office hours session brought together participants from poetry, knowledge tools, and AI-for-the-commons research for an exploratory conversation about sense-making, creative synthesis, and the future of text. The discussion centred on a tool called Hyghlighter and its capacity to support what was termed “frame building” — the cognitive act of assembling fragments from multiple documents into a coherent narrative structure. From there, the conversation moved through constraint-based collaborative poetry (the Japanese renga form), the concept of Visual-Meta as a way to embed machine-readable metadata in human-readable documents, glossary design as lightweight knowledge infrastructure, and the problem of disciplinary tunnel vision reframed as “sanctioned cognitive biases.” Throughout, participants found unexpected convergences between poetic practice and serious knowledge work.

The experience of working through large bodies of research material was described as proceeding not incrementally but through a distinct phase change — days of uncertain engagement followed by a sudden morning where connections crystallise and everything clicks into place. This was likened to Pasteur’s notion that chance favours the prepared mind, but with an important specificity: the transition feels less like gradual accretion and more like a state change, from liquid uncertainty to crystalline understanding. The observation carries practical weight for tool design, because it implies that knowledge tools should be built to sustain a person through long periods of ambiguity rather than promising immediate payoff.

The tool Hyghlighter was reframed during the session not as a reading tool, annotation tool, or even an external memory device, but as a “frame building tool” — a system that supports the construction of cognitive schemas by allowing fragments from many documents to be spatially rearranged into narrative coherence. This reframing itself arrived as one of the phase-change insights: after a decade of development, the conceptual identity of the tool surfaced only through prolonged engagement with data frame theory literature. The implication is that the purpose of a tool can hide from its own creators until the right theoretical frame is encountered.

A tension emerged around the relationship between creative synthesis and originality, particularly in poetry. The observation was made that poets remain deeply resistant to any process that involves harvesting or recombining existing text — seeing it as cheating rather than as legitimate creative method. And yet, the oldest poetic forms, including the renga and the sonnet, are themselves highly constraint-based structures that regulate topic, syllable count, participants, and sequence. The provocative suggestion was offered that the future of poetry may be work written by machines for machines, and that human audiences and creators are the least interesting participants — a reframing that invites the question of what writing is for when authorship itself becomes distributed.

The idea of a cross-disciplinary concordance was raised as a way to address the deep problem of transdisciplinary collaboration. Rather than offering definitions, such a concordance would surface exemplars — showing how practitioners in different fields actually use the same term as a tool of their trade. This connects to the newly coined concept of “sanctioned cognitive biases”: the observation that disciplinary training requires adopting a particular perceptual lens, and that this lens functions as a form of institutionalised tunnel vision. When teams of specialists are brought together for large-scale problems like sustainability or pandemic preparedness, their disciplinary frames actively interfere with coherent synthesis. The concordance idea attempts to make these invisible frames visible and navigable.

Visual-Meta was presented as a method for embedding rich, structured metadata directly within a PDF at the same level as the text itself — human-readable in plain form, machine-readable in structure. One participant described it as “a barcode where the barcode itself contains the information,” making it both self-documenting and computationally accessible. The significance for knowledge work is that any flat document can carry layered meaning — dates that resolve to specific calendar entries, terms that link to glossary definitions, headings that enable document folding — without requiring any external system or database. For poetry specifically, this opens the possibility of marking up creative work with structured relationships (such as seasonal or lunar references in renga) while preserving the integrity of the surface text.

A practical glossary workflow was demonstrated in which selecting any term and invoking a simple keyboard shortcut creates a glossary entry appended to the document as plain text. When the user later searches the document, glossary terms appear at the top of results, surfacing conceptual anchors without requiring the user to know they exist. The design philosophy prioritises extreme simplicity after years of attempting more complex solutions — an acknowledgment that the most durable knowledge infrastructure may be the kind that looks like nothing special at all.

The suggestion was made that design patterns — originally from architecture, widely adopted in software — function as a kind of poetic form for practical knowledge: short, structured descriptions that embody how-to understanding and can travel across disciplines. A proposed university course in Transdisciplinary Design would ask each student to become expert in one domain, then collectively develop shared communicative capacity — an educational experiment in exactly the kind of distributed sense-making that tools like Hyghlighter attempt to scaffold.