The Conscilium Manifesto

Conscilium is a digital research environment for historians and humanities scholars. It is not designed around a particular artificial intelligence model, computational technique, or technological trend.

It begins instead with the practices through which researchers actually produce knowledge. Conscilium is not interested in proving what a technology can achieve when applied according to a predetermined formula; it is interested in shaping itself around the rarely linear sequence that scholarly work actually follows, with its own actions, methods, revisions, and decisions.

Technology should adapt itself to scholarly practice rather than forcing scholarship to conform to the assumptions of software. Its role is to complement the researcher’s work, simplify unnecessary passages, and enrich the possibilities available throughout the research process.

Conscilium therefore seeks to organise computational instruments around research operations rather than around isolated technological capabilities. The question should not be: What can this technology achieve on its own based on our request? It should be: What does the researcher need to examine, understand, test, or demonstrate, and how can we provide them with the elements necessary to make that process easier and richer?

Historical knowledge must remain connected to historical evidence. Every transcription, annotation, classification, observation, relationship, interpretation, and conclusion produced inside Conscilium should be capable of remaining linked to the material that supports it: a manuscript, image, folio, region, passage, document, object, record, dataset, or bibliographic source.

A scholarly environment should make it possible to move in both directions, from the source towards an interpretation and from the interpretation back towards the source. The connection between them must never disappear behind an interface.

Conscilium therefore rejects systems in which computational outputs become detached from the evidence from which they originated. A historical claim without recoverable evidence is not strengthened because a machine produced it quickly.

Traceability has to be one of the foundations of the research environment.

Artificial intelligence can become an extraordinary scholarly instrument. It can propose and suggest an enormous range of potentially valuable elements: transcriptions, patterns, related material, similarities, organisational structures, extracted entities, comparisons, contradictions, and ways of accelerating repetitive operations or navigating increasingly large collections.

These possibilities are the product of an exceptional capacity for data analysis, but suggestion is not demonstration, probability is not evidence, and data must still be carefully questioned and curated through the historian’s own capacities. Conscilium therefore treats computational systems as participants in a research workflow, not as autonomous scholarly authorities.

The researcher must remain able to distinguish between different epistemic states, and the interface should never silently collapse observation into interpretation, prediction into fact, similarity into identity, model output into scholarly conclusion, or uncertainty into certainty. Conscilium should make these distinctions visible.

A reading proposed by a model is a model suggestion; a reading accepted or corrected by a researcher is a scholarly intervention and ultimately subject to scholarly criteria. A hypothesis remains a hypothesis until the researcher decides that the available evidence justifies something stronger.

AI may assist the process of the researcher’s reasoning, but it must not conceal that process.

Humanistic evidence is often incomplete. Important information such as dates, authorship, and places may remain uncertain, and because of the nature of historical evidence several readings of the same abbreviation may remain plausible, sources may contradict one another, and interpretations may coexist without any immediate possibility of resolving them.

A research environment designed around historical practice must therefore permit uncertainty to exist. Conscilium supports unknown values, ranges, probabilities where appropriate, competing readings, alternative identifications, contradictory evidence, provisional classifications, and unresolved hypotheses.

It should never force researchers to invent precision merely because a database field requires a single answer. Ambiguity is not corrupted data.

Sometimes ambiguity is precisely what the evidence tells us, and preserving it is part of preserving the historical record.

The digital humanities does not suffer from a lack of tools. It suffers from their isolation.

Take the example of a scholar working on a manuscript: they might use one platform to view IIIF images, another for transcription, another for palaeographical comparison, another for bibliographic references, another for prosopographical data, another for maps, another for networks, another for notes, and still another for writing. Every transition risks breaking the chain connecting one scholarly operation to another and complicates the research process even further.

Conscilium seeks to reduce this fragmentation. It should eventually contain specialised but interoperable research applications capable of addressing different forms of historical work while sharing a common research infrastructure.

Projects, sources, bibliographic materials, entities, annotations, evidence, provenance, and scholarly outputs should be capable of moving across the environment when intellectually appropriate. The purpose of this integration is not simply to make several tools accessible from the same interface, but to preserve the integrity and continuity of the research process itself.

When the different stages of scholarly work are separated across unrelated systems, the connections through which an interpretation was constructed can become difficult to maintain and reconstruct. An integrated environment should allow the researcher to move through different operations without repeatedly breaking and rebuilding those relationships, so that the research process remains coherent as it develops.

Integration should follow the logic of scholarship.

No research environment should attempt to become the entirety of digital scholarship, and Conscilium is not intended as a universal solution that replaces every existing digital humanities tool. Its objective is instead to provide a coherent research environment that can also communicate with the wider technological and scholarly ecosystem around it.

Where another instrument, infrastructure, or specialised service is appropriate to a researcher’s work, Conscilium should be willing and able to interoperate with it rather than requiring the researcher to abandon the continuity of their virtual research process in order to use it. Integration should therefore extend beyond the internal boundaries of Conscilium itself.

Conscilium should participate in a wider ecosystem of archives, libraries, repositories, standards, software, research infrastructures, and scholarly communities. Interoperability is therefore not simply an engineering convenience; it is a scholarly commitment.

Where appropriate, Conscilium should work with open and established standards such as IIIF, W3C Web Annotation, PAGE XML, ALTO, TEI, BibTeX, CSL JSON, linked data, and structured archival formats. Researchers must be able to import their work, but they must also be able to leave.

A project should remain intelligible outside the software in which it was created, and no scholar should lose years of annotations, transcriptions, classifications, or research relationships because a platform disappears, changes its business model, or becomes technically obsolete. Tools are temporary.

Scholarship must outlive them.

There is no universal workflow for historical research. A palaeographer, an economic historian, an archaeologist, a historian of institutions, and a scholar constructing a prosopographical database do not work in the same way, and even two researchers studying the same source may organise their investigations differently.

Conscilium should therefore resist the temptation to impose a single model of scholarship. Its architecture should be modular, allowing researchers to select the instruments appropriate to their questions, construct different workflows, and introduce new methods without rebuilding the entire environment.

Integration should not become homogenisation. A common infrastructure should allow different scholarly traditions to communicate while preserving their methodological diversity.

Digital research infrastructure should serve researchers rather than extract value from them. Private manuscripts, unpublished transcriptions, corrections, annotations, hypotheses, teaching materials, and research data should not automatically become training material for computational systems.

Researchers must retain meaningful control over where their projects are stored and how their data are used. Conscilium should therefore allow research to take place locally, offline where possible, through institutional infrastructure, or through shared services chosen by the researcher.

Participation in collaborative datasets or training corpora should be explicit rather than presumed. When scholars choose to contribute material, the contribution should preserve provenance, attribution, licensing, and rights information.

Openness must never become another word for appropriation. Sharing is valuable precisely because it is meaningful when researchers are able to choose it.

Digital scholarship cannot become genuinely transformative if sophisticated methods remain effectively restricted to researchers with advanced technical skills, substantial institutional support, or the financial capacity to sustain a fragmented ecosystem of proprietary services. Accessibility therefore cannot be reduced to interface design.

It is also an economic, technical, and infrastructural question. Democratisation does not mean replacing expertise with automation, nor does it mean reducing the intellectual demands of historical research.

It means widening access to the instruments through which expertise can be exercised. The growing fragmentation of digital research environments has raised the threshold for participation.

Researchers are increasingly expected to assemble their own workflows across multiple platforms, maintain access to separate services, navigate incompatible systems, and absorb recurring computational and subscription costs. Even when each individual component appears accessible, their cumulative cost and complexity can make advanced digital research impractical for students, independent scholars, small research groups, and less well-funded institutions.

Conscilium seeks to lower this threshold by bringing complementary research functions into a coherent environment and by developing that environment as open-source infrastructure. Its methods, architecture, and code should remain open to inspection, adaptation, extension, and independent deployment.

Researchers and institutions should not have to depend entirely on the pricing, continued availability, or strategic priorities of a single commercial provider in order to preserve or continue their work. Open source is therefore not only a licensing choice.

It is part of Conscilium’s understanding of scholarly access. It allows research communities to appropriate the infrastructure for their own needs, contribute to its development, deploy it under different institutional conditions, and retain greater control over the tools on which their scholarship depends.

Technical accessibility is equally important. Researchers should not need to become programmers, systems administrators, or machine-learning specialists before they can make meaningful use of computational methods.

The interface should absorb unnecessary technical friction while remaining transparent enough for scholars to understand how computational results were produced and how they should be evaluated. A student should be able to experiment with advanced digital methods without first constructing an entire technical stack.

A small research group should be able to undertake sophisticated computational work without reproducing the infrastructure of a major laboratory. Independent scholars, libraries, universities, and institutions operating under very different material conditions should all be able to participate meaningfully in digital scholarship.

This does not mean making scholarship effortless. Historical research is difficult because its evidence is fragmentary, its interpretations contestable, and its questions complex.

That difficulty should remain. What should not remain are avoidable barriers created by incompatible software, unnecessary technical expertise, fragmented infrastructure, and repeated financial costs.

The aim is to ensure that intellectual difficulty remains the meaningful difficulty. Advanced digital methods should become part of the common scholarly infrastructure of the humanities rather than privileges attached to technical specialisation, institutional wealth, or access to proprietary systems.

Digital scholarship should not be evaluated by how magical its technology appears. A system that produces an impressive or already closed answer without allowing the researcher to understand how that answer was reached, or without preserving its evidential foundations, may be less useful for scholarship than a more modest instrument whose operations are transparent and reproducible.

Conscilium therefore rejects technological spectacle as a measure of scholarly value. The objective is not to hide complexity behind the illusion of an omniscient machine, but to make sophisticated computational processes usable while preserving the possibility of scrutiny.

Researchers should be able to understand where information came from, how it was transformed, what assumptions entered the process, and where uncertainty remains. This is also why Conscilium keeps the research process user-centred: the technology offered by the software does not speak for the researcher and should never be allowed to do so.

The best research technology should not make the researcher disappear. It should make the research process clearer.

Digital scholarship has repeatedly produced valuable projects that become inaccessible when funding ends, software dependencies change, or institutional support disappears. Sustainability must therefore be considered from the beginning rather than treated as a problem for the future.

Open formats, modular components, documented interfaces, reproducible exports, standards-based interoperability, and the possibility of local deployment are key strategies for protecting scholarly work. Conscilium itself began as a project built by an undergraduate student with limited resources and limited possibilities for immediate scaling.

That origin matters because there are many other researchers, students, and small scholarly communities working under similar constraints. The infrastructure should therefore never be designed on the assumption that continued access to large-scale institutional or commercial resources can be guaranteed, nor should it require researchers to trust that a single platform will exist forever.

Conscilium should instead be designed around the assumption that technologies change while research must remain recoverable.

Conscilium is not an attempt to automate the historian. Historical interpretation cannot be reduced to a sequence of predictions.

It depends upon context, comparison, judgement, criticism, imagination, disagreement, revision, and argument. Computational systems can surely extend these practices, but their value depends upon how they are incorporated into research.

The central problem of the next generation of digital humanities is therefore not simply how powerful our models become. It is how responsibly, transparently, and coherently we integrate computational power into the production of knowledge.

Conscilium exists to explore that possibility: a research environment in which sources remain visible, evidence remains traceable, and instruments communicate instead of fragmenting the research process. Technology should not replace the scholarly process.

It should give that process more room to think.