A conference on inference and the pre-print record
Ars Inquirendi 2026
Querying Pre-Print Cultures with LLMs
Pre-print archives are made of loss, dispersal, and uneven survival — exactly the shape of evidence that inference, not just retrieval, is built to face.
Call for papers → 31 August 2026 Early draft · permanent home arsinq.com coming soon
Why the pre-print world, why now
“The inferential prowess of LLMs is especially promising for pre-print cultures, since their archives are overwhelmingly one of loss, dispersal and uneven survival.”
That promise is not generic. It rests on what these models actually are:
“LLMs are language-shaped machines, trained on human expression and responsive to all the breadth and detail of natural language. Some of the most striking new techniques for eliciting powerful behaviour from models depend not on esoteric programming, but on skills such as rhetoric, philology, hermeneutics, poetics, psychology, and the art of asking.”
But the same evidentiary conditions that make pre-print cultures a promising target also make them a dangerous one to work carelessly:
“pre-print-era scholarship is also exceptionally vulnerable to the floods of synthetic text, shallow automation, and dangerously plausible AI nonsense … only a fraction of those materials fall within the data of current models, which train overwhelmingly on post-print production, above all twenty-first-century internet culture.”
Scope: cultures before movable-type dominance — Western Europe to the 16th century, Russia to the 18th, Central Asia to the late 19th.
Papers, ecosystem, and practice
From Dream to Actuality
Demonstrations of LLM projects making the previously unreachable real: reconstructions of lost texts, libraries, subcultures, and lost knowledge, and the relations between them; re-imaginings of history as a field of possibility; scholastic disputation engines; historical route, map, and landscape systems; prosopographical workbenches; catalogue-enrichment tools; and projects that turn scholarly argument itself into a working system.
The Pre-Print AI Ecosystem
What it means to build a genuinely pre-print AI ecosystem: sovereign LLMs that exclude later material versus boosting pre-print material within frontier training; copyright, public-domain distortions, transcription error, and low-resource languages; LLMs situated within the wider ecology — HTR/OCR, RAG, metadata enrichment, entity recognition, knowledge graphs, corpus analysis — and their strengths and weaknesses against other methods.
Workshops
Practical 1–2 hour sessions:
- Using LLMs well, in chat and in more elaborate interfaces
- LLMs in teaching
- Ethics and pitfalls
- Comparative use of models
- Coding with LLMs, natural-language-driven
- How LLMs work under the hood: making, adapting, fine-tuning, distilling
- Humanistic practice itself as a source of method and standards — wiki-like approaches to explanation and staged understanding, prompt design as rhetoric New
Submit by 31 August 2026
Papers, roundtables, and workshops are invited across all three strands. The conference is hybrid; all talks except keynotes are pre-recorded and released a week ahead of the conference, with live session time reserved for discussion — so every session is followable both in person and online.
Papers & roundtables
~250–300 words.
Workshops
~500–700 words. State intended audience, duration, and technical requirements; sessions must be followable both in person and online.
Short bio
≤100 words.
Format & mode
State your preferred format (paper, roundtable, or workshop) and whether you intend to present online, in person, or either.
On site
- Lunch provided 20 & 21 November
- Conference dinner, evening of 21 November
Send submissions to
arsinquirendi@gmail.comStrands I–II, already running
Two working exhibits from the Plerome suite instantiate Strands I–II directly — useful as concrete conversation-starters, not as anything the conference has vetted or endorsed.
One Thousand and One Conjectures →
An LLM posing 1,000+ falsifiable conjectures about the pre-print world, each carrying a dated prior-art check. A live instance of “reconstructing lost knowledge” run at scale — and, by the same token, a measured, checkable answer to the survival-bias and “dangerously plausible” problem the rationale names.
plerom.org/conjectures →Inferpedia — the encyclopedia of the missing →
An evidence-led wiki for unattested but warranted entities in the pre-print record: lost books, unnamed people, vanished buildings, events known only from their echoes — every article an audited argument, never a settled fact.
plerom.org/inferpedia →Both are working previews built by one of the organisers, offered here as a concrete case for discussion — not as the conference's proceedings or a claim of endorsement.
Organising committee
- Mathew BarberAga Khan University
- Rubén González CuervaInstituto de Historia–CSIC, Madrid
- Yasmin FaghihiUniversity of Cambridge
- Estelle GuévilleYale University & Bibliothèque nationale de France
- Huw JonesUniversity of Cambridge
- Anthony John LappinUniversity of Stockholm
- Henrike LähnemannUniversity of Oxford
- Roger Martínez-DávilaUniversity of Colorado & Plus Ultra Collective
- Stephen PinkPlus Ultra Collective
What the first meeting established
Held online and in Oxford, 4–7 December 2025, the first Ars Inquirendi convened pre-modernists, archaeologists, digital humanists, and AI researchers “to test — rather than merely celebrate — what large multimodal models can do for pre-print-era evidence.”
Keynotes: Maurizio Forte (Duke) and Roger Martínez-Dávila (Colorado).
- Old English NLP benchmarks
- The Graphom project, modelling the pre-print “graphosphere”
- AI-assisted manuscript catalogues
- Visual language models for Church Slavonic, Glagolitic, and Ukrainian
- Birchbark-letter reconstruction
- Historical-map exploration with MapReader
- Workshops on prompt design, hallucination control, and tokenization
Its “central achievement was to shift discussion from general claims about AI to concrete scholarly evaluation.”
Conference recordings on the Oxford Medieval Studies webpage →
Synthetic proceedings at the Plus Ultra Collective website (link pending).
Dispatches
Title pending
Lede pending — to be filled from a @darkarchives article supplied by the operator.
Read → (pending)Title pending
Lede pending — to be filled from a @darkarchives article supplied by the operator.
Read → (pending)