In 1948, a conference in the United States brought together librarians, engineers, and a scattering of other specialists who all felt, more or less, that they were working on the same thing — the systematic handling of information — but could not agree on what to call it. Seven decades later, the question of what a field is, what it contains, and what we should name its parts has not gone away. It has, if anything, grown more urgent. A small but striking cluster of papers published in the first days of October 2026, spanning a historical study of information science's conceptual roots and four tightly interlinked conceptual working papers on human-AI interaction, collectively turns the lens back onto the act of naming itself. The result is less a set of findings than a set of questions, and those questions are, in a very real sense, the questions of our moment.
What makes this moment distinctive is not that anyone is proposing a grand new discipline. Quite the opposite. Every one of these papers is explicitly bounded, explicitly not claiming to establish a new field, a validated instrument, or a global novelty. And yet, taken together, they sketch the intellectual infrastructure that will either make sense of the coming decades of human-AI collaboration or quietly fail, leaving researchers to improvise vocabulary in the dark. The stakes of getting the names right — or, more precisely, of getting the questions about the names right — are harder to overstate.
The Long Shadow of a Name
Alvin M. Schrader's In Search of a Name: Information Science and its Conceptual Antecedents [1] is, in the most literal sense, an archaeology of vocabulary. Schrader undertakes a historical study of the concepts, and what those concepts actually covered, used throughout the development of what is now understood as "information science." The work is not a narrative of great men and breakthroughs. It is a genealogy of terms: when did a word first appear in a given context, what did it mean to the person who used it, how did its meaning drift or contract over time, and who was responsible for the shift?
Two appendices give the paper its unusual texture. The first is a synoptic table tracking the dates of appearance and evolutionary stages of the principal macroterms in the field, with the author associated with each entry. The second is a list of conferences in the domain from 1948 to 1978. Together, these two tables turn what could have been a dry historiographical exercise into something closer to a map of an intellectual ecosystem — showing not just what was said, but where and when it was said, and by whom. The paper, published in the ERA: Education and Research Archive at the University of Alberta and already drawing fifty-three citations, reads as a corrective to the tendency in fast-moving fields to treat their current vocabulary as if it had always been there, as if the words simply fell out of the sky fully formed.
Why does this matter beyond a niche historiographical interest? Because every field inherits its vocabulary from a particular historical accident, and those inherited terms carry assumptions that rarely get examined. The word "information" itself, as Schrader's work implicitly reminds us, was not a neutral container. It was a contested, evolving, context-dependent concept whose boundaries were drawn by specific people at specific conferences in specific decades. Recognizing that history is the first step toward asking whether the inherited vocabulary is still doing its job.
When Does a New Word Actually Help?
Andreas Ehstand's Systematic Neology: Evaluating When New Terms Help in Human-AI Work [4] asks that question head-on, and with a level of methodological self-awareness that is almost uncomfortable. The central comparison is deceptively simple: does a proposed name add value beyond what adequate existing language and the same explanation without the name can already do? That is, if you can describe the phenomenon clearly using words you already have, why invent a new one?
Ehstand structures the answer around eight revisable methodological commitments, six record types, seven relations, three invented illustrations, and four research questions. The emphasis on revisable is doing important work here. These are not axioms. They are working hypotheses about how to evaluate a candidate term, and they are explicitly open to revision. The paper also carefully distinguishes Systematic Neology from two neighboring projects — Semanturgy and Cognitive Continuity Methodology — and acknowledges established conceptual and terminological work rather than claiming to supplant it.
The practical implication is sharp. In a field as volatile as human-AI interaction, the temptation to coin a new term for every slightly different configuration of tools, agents, and users is enormous. Ehstand's framework offers a brake: before you reach for the thesaurus, demonstrate that the existing language genuinely fails. The three invented illustrations serve as test cases, showing what the evaluation process looks like in practice. And the paper is explicit that it reports no empirical findings, no established new discipline, no global novelty. It is a method for evaluating names, not a catalogue of them.
Mapping the Terrain Before You Name It
If Systematic Neology is the question of whether to name, Ehstand's NEOMANITAI: A Faceted Research Framework for Human-AI Interaction [2] is the question of how to describe the territory that names will eventually cover. The paper develops a research framework for describing human-AI interaction across three settings: digital, embodied, and collective. It revises an existing umbrella proposal and, crucially, distinguishes between programme names, classification facets, concepts, observations, and evidential claims. That last distinction — between the name of a programme and the claims it supports — is where the intellectual honesty lives.
The framework's architecture is concrete: eight ordering conventions, seven descriptive facets, a small typed relation model, three constructed civilian cases, and four comparative research questions. The literature base is broad, covering distributed cognition, socio-technical systems, hybrid intelligence, collective intelligence, agents, human-robot interaction, conceptual work, and taxonomy development. The stated contribution is "an inspectable connection between programme navigation, episode descriptions, and claim-level provenance." In other words, NEOMANITAI is less a theory than a bookkeeping system — a way to keep track of what was said, by whom, in what context, and what evidence supports what.
The paper is equally explicit about its limits. It reports no new participant data, no empirical validation, no established disciplinary status. The constructed cases are explanatory examples, not findings. And the drafting and editorial checks were AI-assisted under the author's instruction, with no claim of external peer review or personal sentence-by-sentence author review. This transparency, while perhaps unusual in a field that often buries its methodological caveats in footnotes, is itself a form of intellectual rigor. It says: here is the scaffolding; the building is not yet finished.
What Survives the Change of Context?
Ehstand's Cognitive Continuity Methodology [3] addresses a question that sits at the intersection of epistemology and practical knowledge management: what happens to decision-relevant knowledge when the representation changes, when the user changes, when time passes, or when the context shifts? The paper reconstructs CCM as a bounded programme concerned with explicitly identified distinctions across those four axes of change, and it keeps four outcomes separate: preservation, documented revision, loss, and unresolved comparison.
That last category — unresolved comparison — is the most interesting. It acknowledges that sometimes you simply cannot determine whether knowledge was preserved, revised, or lost. The comparison is open. The framework does not force a verdict. This is a small but significant methodological choice, because most taxonomies and classification systems implicitly assume that every item can be sorted into a bin. CCM says: sometimes the bin is empty, and that is a legitimate outcome, not a failure of the system.
The paper describes six entities, five relation families, four research questions, and a single constructed civilian reading-room example. It relates its object to epistemology, distributed cognition, conceptual work, and provenance. And, like its sibling papers, it is explicit that no empirical study, validated instrument, or established new disciplinary status is reported. The historical source files are consulted selectively and remain separate restricted records. The open edition contains the bounded conceptual reconstruction, not technical source procedures or internal provenance files.
Co-Creation, Contribution, and the Boundary of Responsibility
The fifth paper, Synmanitik: A Research Programme on Human-AI Co-Creation of Scientific Concepts and Hypotheses [5], pushes the inquiry into the most philosophically charged territory of the set. It is concerned with episodes in which people and AI systems develop scientific concepts, questions, and hypotheses together. And it draws three distinctions that cut to the heart of current debates about authorship, credit, and epistemic responsibility:
- Contributions versus support. Not every act of assistance is a contribution. A system that reformulates a question, checks a citation, or generates a candidate hypothesis is not necessarily co-creating in the same sense as a human who commits to a claim and stands behind it.
- Human responsibility versus infallibility. The framework insists that human responsibility does not require human infallibility. A researcher can be responsible for a conclusion without being the sole generator of every step that led to it.
- Internal collaboration versus independent evaluation. Working together on a concept is not the same as independently testing it. Synmanitik keeps these two activities distinct, resisting the conflation that often blurs the line between generating an idea and validating it.
The paper specifies three recurring activities, seven methodological commitments, eight conceptual record types, ten named relation types, two invented civilian cases, and four prospective research questions. A critical requirement is that the connected episode record must demonstrate additional value over existing research and simpler logs containing the same substantive information. In other words, the framework must earn its complexity. It must show that the structured record tells you something a flat log would not. This is a built-in Occam's razor, and it is a welcome guard against the proliferation of ever-more-elaborate annotation schemes that add formalism without insight.
As with the other Ehstand papers, the caveats are extensive. No empirical study, no general human-AI synergy claim, no verified global novelty, no legal authorship rule, no autonomous discipline is established. The drafting and checks were AI-assisted, with no claim of personal sentence-by-sentence author review or external peer review. Older restricted records remain unchanged.
The Bigger Picture: Naming as a Discipline's Immune System
Read together, these five papers form a kind of intellectual immune system for a field that is growing faster than its vocabulary can keep up. Schrader [1] shows us that the messiness of naming is not a bug but a historical feature — every field has been through it, and the vocabulary that emerges is a sediment of decisions made in specific conferences, by specific people, at specific moments. Ehstand's four papers [2][3][4][5] then take the lessons of that history and build a set of bounded, inspectable, explicitly revisable tools for navigating the next round of conceptual growth in human-AI work.
What is striking is the collective refusal to overclaim. None of these papers announces a new discipline. None reports a breakthrough. None asks you to adopt a new term. Instead, they ask you to think more carefully about the terms you already use, the frameworks you already deploy, and the boundaries you already draw. They are, in a sense, papers about the process of making papers — the metadata, the provenance, the relation types, the record structures that allow a field to audit its own intellectual hygiene.
And that is, perhaps, the most important point. In a field where the tools are changing as fast as the questions, the most valuable contribution may not be a new finding but a new way of keeping score. Schrader's synoptic table of macroterms and conference lists [1] is a scorecard for information science's first three decades. Ehstand's typed relation models, facet structures, and episode records [2][3][4][5] are scorecards for the next. Neither is a destination. Both are maps of the road, drawn with the explicit acknowledgment that the road is still being paved, and that the map will need redrawing.
The question these papers leave with the reader is not what should we call the field? It is more subtle, and more important: how do we decide when a name has earned its place, and how do we know when it has outlived its usefulness? In a moment when the answers to that question will shape how millions of researchers describe, credit, and understand their collaboration with machine systems, the stakes of getting the vocabulary right are not merely academic. They are, in the most literal sense, the architecture of how we will think together for the next decade and beyond.
References
- Alvin M. Schrader (2026). In Search of a Name: Information Science and its Conceptual Antecedents. ERA: Education and Research Archive (University of Alberta).
- Andreas Ehstand (2026). NEOMANITAI: A Faceted Research Framework for Human-AI Interaction. Zenodo (CERN European Organization for Nuclear Research).
- Andreas Ehstand (2026). Cognitive Continuity Methodology: A Bounded Framework for Describing and Testing Continuity of Decision-Relevant Knowledge. Zenodo (CERN European Organization for Nuclear Research).
- Andreas Ehstand (2026). Systematic Neology: Evaluating When New Terms Help in Human-AI Work. Zenodo (CERN European Organization for Nuclear Research).
- Andreas Ehstand (2026). Synmanitik: A Research Programme on Human-AI Co-Creation of Scientific Concepts and Hypotheses. Zenodo (CERN European Organization for Nuclear Research).