The history of computer science is marked by the tension between formal model and technical implementation. Drawing on the case of search algorithms, I theorize this dialectic and examine the dynamics that both maintain the separation between these domains and organize their interdependence.

I will show that this tension characterizing the history of computer science can be understood through two distinct stances rooted in the very origins of the concept of form in ancient Greek philosophy.

On the one hand, there is the formalist or computational perspective, which recalls the Platonic conception of form. In this view, mathematical models exist autonomously from their technical implementation and enjoy hierarchical priority over it. Algorithmic formalization represents the true essence of computers, where the computer is understood as a calculating machine, as an implementation of an ideal Turing machine.

On the other hand, there is a position that we trace back to Plato’s most celebrated student, one that recognizes the hybrid and relational nature of computer science. In this perspective, the influence of form on matter can be explained without needing to postulate a proper theory of abstraction. In what we call the Aristotelian approach, form is always the form of something and ensures the fulfillment of that thing’s specific function, whether it be a material body, a set of bodies, or even a determinate concept.

Rather than adopting an unequivocal stance toward one of the two positions, we propose considering form as relationality and immanence to be historically prior, while acknowledging that the Platonic tradition possesses genuine epistemological and ontological value, though derivative.

The case of computational search algorithms enables us to articulate this argument. The advent of modern computers, thus a technical necessity, created the need for mathematizing information retrieval, which initially took the form of simple sequential search. Subsequently, beginning with the first formulation of binary search, the progressive complexification of search algorithms developed in a relationship of causal co-determination with the practices and tools for organizing and managing data in computational memory.

We will theorize this essential entanglement between formalization and materiality as a recursive process. In this framework, recursivity refers to both the interdependence between form and matter and the progressive stabilization of the boundaries between them. The increasing complexity of formal aspects, driven in part by practical problems such as expanding computing memory and data availability, led to the development of increasingly sophisticated search strategies. Their formulation required recourse to mathematical competences and, with it, to a culture, way of thinking, and epistemological approach abstracted from the material concerns of implementation.

Through this case study, we will show how the recursive relationality between computation and matter aligns with Aristotelian form, conceived as an immanent entity enabling both the functioning of things and the proper organization of their constituent parts. We will further show that the representations of the computational approach can be theorized as a progressive reification and abstraction of form understood in the Platonic sense, as the mathematical essence of things.

Recent public discourse around “AI for the humanities” has foregrounded large language models (LLMs) whose outputs can feel interpretively powerful yet methodologically opaque. This paper argues that Voyant Tools offers a contrasting “glass-box” paradigm for digital humanities inquiry: its core instruments (e.g., term frequencies, concordance/KWIC retrieval, topic modeling, and co-occurrence measures) expose intermediate representations—tokens, counts, parameters, and textual evidence—so that claims can be traced, challenged, and reproduced.

Building on a Spyral library that lets you work with LLMs and an experimental Voyant workflow [https://www.experimentalvoyanttools.ca] that adds explicit, inspectable LLM endpoints (e.g., a general chat endpoint, topic-labeling, and document comparison), we show how LLMs can be combined with traditional text analysis tools and used without turning interpretation into a black box. The key design principle is constraint and auditability: using traditional tools where relevant, developing prompts that require citing supplied context snippets (e.g., “[doc N]”), forbid inventing evidence, and enforcing structured outputs (JSON-only or line-based schemas) that can then be examined with traditional tools. These constraints keep the LLM’s role closer to assisted organization and labeling than free-form meaning production.

We present four practical “beyond close reading” use cases that remain evidence-grounded: (1) an LLM-assisted stoplist/cleanup interface that proposes removals from Voyant’s ranked term lists, followed by deterministic re-runs of Cirrus/Topics/Collocates to measure impact; (2) topic cluster labeling that treats LDA outputs as inputs to a tightly constrained labeling step; (3) interpretation of distinctive-term lists as explicitly provisional hypotheses tied back to the ranked evidence list and validated through concordance checks; and (4) concordance “sense-making,” where KWIC lines are clustered into usage types with representative textual exemplars and ambiguity flags.

Our experimental research suggests that Voyant’s strength is precisely its visibility into the translations that computation performs (text → tokens → models → visualizations). Rather than hiding these remediations, the workflow makes them inspectable—turning distant reading into a documented, reflexive method pipeline suitable for DH pedagogy and research transparency.

International students have become one of the most intensively governed migrant groups in Canada, increasingly processed not through traditional adjudication but through technoscientific systems of assessment. Since 2019, Canada has expanded its use of automated triage, predictive analytics, cross-border data systems, and biometric infrastructures in visa processing. This shift—I describe as the emergence of immigration science—has transformed mobility governance from a discretionary administrative process into a data-driven risk-management regime. Yet, despite rapid uptake of these technologies, little scholarly attention has been paid to how algorithmic infrastructures are reshaping the rights and lived experiences of international students.

Problem

Existing scholarship in migration studies and digital governance highlights algorithmic bias, bureaucratic opacity, and securitization, but largely overlooks how rights themselves are being recoded as risk categories. In Canada, technoscientific tools are not merely expediting application processing—they are transforming international students into statistical entities governed through compliance scores, predictive risk models, and automated decision systems. This raises foundational political questions: What becomes of the right to education when access is determined through probabilistic classification? And how does technoscience subtly erode rights while maintaining the appearance of neutrality and efficiency?

Research Question

How do Canada’s technoscientific migration systems—namely automated triage, predictive analytics, and biometric infrastructures—reshape the rights and mobility of international students?

Argument

I argue that Canada’s adoption of technoscientific governance tools has produced a regime of riskification in which international students’ rights are reframed as potential risks to be managed rather than entitlements to be respected. Under this regime, applicants are increasingly governed as probabilities (e.g., likelihood of overstay, “compliance risk,” or potential for misrepresentation) rather than as legal subjects. This epistemic transformation shifts decision-making power from legal norms toward computational logics, thereby hollowing out procedural fairness and undermining the substantive right to education. Importantly, these logics reproduce longstanding racialized and colonial hierarchies, even as they present themselves as objective, neutral, and efficient.

Methods

This study employs a qualitative content-analysis methodology grounded in Science and Technology Studies (STS) and critical migration scholarship. Data sources include:

Government documents and technical reports (e.g., Advanced Analytics program, Algorithmic Impact Assessments, CBSA compliance scoring tools).

Parliamentary reports and ministerial briefings on immigration technology.

Publicly available policy statements and internal documents obtained through FOIP.

Watchdog and civil-society reports documenting the impacts of algorithmic migration governance.

These materials are analyzed using an interpretive STS framework that examines how computational tools construct categories, risks, and subjects. Special attention is given to how algorithms encode assumptions about credibility, security, and desirability.

Contribution

This paper contributes to digital humanities and migration governance scholarship by theorizing “immigration science” as a technoscientific regime that reshapes rights from within administrative infrastructures. It foregrounds international students—a rapidly expanding but understudied migrant population—as a critical site for examining how risk, data, and algorithmic decision-making transform mobility governance in Canada. More broadly, the paper advances interdisciplinary conversations on the ethics and politics of automated governance, offering conceptual tools to evaluate how technoscience redefines citizenship, belonging, and rights.