Abstract
This article presents a methodological case study of Chinook, a tool employed by the
Canadian government’s department known as Immigration, Refugees and Citizenship Canada
. Chinook assists officers with temporary resident visa applications. To examine
this tool, we use design fiction, a research-through-design method that builds speculative
prototypes in order to explore the social, ethical, and political implications of
technological systems and infrastructures. Given the system's ambiguity, fragmented
documentation, and limited public accountability, we adopt speculative reconstruction
as a critical and interpretive approach. This article relied on court affidavits,
Access to Information and Privacy releases, and legal commentary, to develop a design
fiction prototype and accompanying explainer video to visualize Chinook’s “Decision
Maker” module. Through iterative sketching, prototyping, and annotation, the project
translates bureaucratic and legal evidence into accessible interface logics. By anchoring
speculation in the documentary record rather than in the imagination, and by directing
it at an operational bureaucratic system rather than an emerging technology, the article
reframes design fiction as a method of evidentiary reconstruction, revealing how digital
infrastructures encode discretion and standardization within government decision-making.
Introduction
How can we study tools used to manage people that are not made public? In this article
we will discuss the design fiction approach we took to understand Chinook, an Immigration,
Refugees and Citizenship Canada (IRCC) tool used to deal with temporary resident visa
(TRC) applications.
IRCC is the Canadian federal government department responsible for immigration, refugee
resettlement, and citizenship, to manage high volumes of TRV applications. A TRV is
the document required of foreign nationals from visa-required countries who seek to
enter Canada for a temporary stay, whether as visitors, students, or workers, and
IRCC processes these applications at visa offices worldwide. IRCC describes Chinook
as an interface to a suite of modules that streamline decision-making using spreadsheet-based
tools, and claims it does not use artificial intelligence, algorithms, advanced analytics,
or built-in decision-making algorithms (Immigration, Refugees and Citizenship Canada,
2024, 2021a, 2021b; Tao, 2022).
What is publicly known about Chinook stems from a patchwork of heavily redacted Access
to Information and Privacy (ATIP) requests, court affidavits, court cross-examinations,
and policy commentary. These include the Ocran v. Canada litigation materials, as well as the ATIP-obtained Chinook training manuals (Daponte,
2021a, 2021b; Immigration, Refugees and Citizenship Canada, 2021a). Ocran v. Canada was a case for judicial review of a visa denial decision in which the role of Chinook
was at issue. Such cases are heard by the Federal Court of Canada, which reviews decisions
made by federal departments including IRCC. In response, IRCC filed an affidavit in
July 2021 by Andie Daponte, an official at the department, setting out how the tool
had been developed and how its modules operate. Daponte was subsequently cross-examined
on that affidavit. Together these two documents remain the most detailed account of
Chinook that the department has placed on the public record, and they anchor much
of the reconstruction that follows.
Legal experts and advocates, such as Tao (2022), have raised concerns about several
of Chinook's features: batch refusal capabilities, which allow an officer to issue
refusal decisions for many applications in a single action; standardized refusal templates,
which supply pre-written reasons that officers select rather than compose; and the
deletion of working notes that once formed part of the procedural record. We will
return to these.
This lack of available documentation presents both a methodological and ethical challenge.
How does one critically examine a government decision-making tool that is shielded
from public inquiry in a way that is both accurate and can engage the public? Our
research responds to this challenge by turning to design fiction. Design fiction is
a research-through-design method that creates speculative prototypes to explore the
social, ethical, and political implications of technological systems. Developed within
critical design and futures studies, it situates artifacts in plausible near-future
or alternative present contexts, allowing researchers and audiences to ask “what if”
questions where empirical observation is limited or impossible (Dunne and Raby, 2013;
Coulton et al., 2017; Blythe, 2014; Lukens and DiSalvo, 2011). We argue that design
fiction functions as a methodological bridge between limited public documentation
and speculative critique, enabling critical insight into black-boxed bureaucratic
systems and potentiating a new public conversation.
Our contribution to design fiction is twofold. First, design fiction has conventionally
proceeded from imagination, extrapolating from trends or provocations to build artifacts
that could plausibly exist. We instead anchor every speculative move in public documentation:
court affidavits and cross-examination transcripts, ATIP-released training materials,
and Federal Court rulings. Speculation begins only where the documentary record stops,
and we mark that boundary explicitly rather than allowing evidence and invention to
blur, using color-coded annotation in the prototype to distinguish elements grounded
in disclosed material from those inferred from textual description. Second, design
fiction has largely been directed at emerging technologies, asking what living with
them might entail. We direct it instead at an existing bureaucratic system that is
already operational and already producing consequences for applicants, but whose workings
are withheld from the people it governs. The object of speculation here is not a possible
future but a concealed present. Taken together, these two moves reposition design
fiction as a method of evidentiary reconstruction, suited to institutions that are
documented but not disclosed.
These two contributions have a consequence worth flagging early. Because we were building
something rather than writing an argument, we kept running into decisions that a written
critique could have left open. Every text field had to go somewhere and every button
had to do something, so each time the documents were vague, silent, or contradictory,
we had to take a position anyway. We treat these moments of friction as findings,
and we report them as we go, because they show where the bureaucracy's own logic becomes
visible.
It is also crucial to note that design fiction is not about predicting the future
or claiming factual accuracy. Rather, it constructs plausible representations to expose
assumptions, surface risks, and invite public reflection (Lindley and Coulton, 2016;
Galloway and Caudwell, 2018; Barendregt and Vaage, 2021). These are our goals in this
article and the associated animation we provide of Chinook based on design fiction.
Our reconstruction pursues them in the manner design fiction has always pursued them,
by building an artifact that can be examined and argued with rather than by making
an argument alone. Where design fiction more often asks what an emerging technology
might make possible, we ask instead what an existing institution has already made
routine. The literature we turn to next establishes the conceptual grounds for this
work: plausibility rather than accuracy as the measure of a speculative artifact,
world-building as a means of synthesizing heterogeneous evidence, and reflexivity
about interpretive limits. Our case extends that literature by testing whether those
commitments hold when the artifact is reconstructed from a documentary record rather
than imagined.
This article proceeds in five parts. First, we outline the conceptual foundations
of design fiction. Second, we describe our method of reconstructing Chinook through
speculative prototyping. Third, we analyze the evidentiary and narrative logic behind
each module’s reconstruction. Fourth, we reflect on the epistemological implications
of this method. Finally, we consider the broader potential of design fiction for studying
algorithmic governance, the use of computational systems to administer and shape public
decision-making (Galey, Ruecker, and Ruecker, 2010; Hinrichs et al., 2018), and deepening
collective understanding.
The Conceptual Foundations of Design Fiction: Considering the Literature
Design fiction has emerged as a critical method within human-computer interaction,
speculative design, and digital humanities, enabling researchers to interrogate sociotechnical
systems through imagined futures and material speculation. Building on the foundational
work of Dunne and Raby (2013), design fiction is specifically defined by its capacity
to challenge normative assumptions and stimulate civic imagination through plausible
but unreal artifacts and systems. Their notion of “social dreaming” recasts design
as a critical and speculative act (also see Farias et al., 2022), rather than a utilitarian
one, shifting attention from solutions to consequences.
This framework has since expanded to include a broader array of epistemological and
methodological concerns. Auger (2014) argues that speculative design, when grounded
in everyday material practices, allows users and designers to imagine what living
with emerging technologies might actually entail. This aligns with Gaver’s (2012)
provocation that research through design should not seek generalizable knowledge in
the traditional scientific sense, but instead generate rich interpretive insights
through engagement with concrete artifacts.
Design fiction is particularly effective for investigating opaque or black-boxed systems,
such as government decision-making software, where technical schematics or source
code are unavailable. Barendregt and Vaage (2021) helpfully frame speculative design
as a form of “thought experiment,” positioning it within a tradition of philosophical
inquiry that emphasizes inference and critique over verification. In this model, plausibility
becomes the primary metric, grounded not in internal coherence alone but in the alignment
between speculative reconstruction and publicly accessible fragments of documentation.
The “world-building” capacity of design fiction has also become central to its recent
development (Coulton et al., 2017; also see Crabtree et al., 2025, p. 2). Coulton
et al. (2017) describe this as a way to synthesize disparate evidentiary forms, such
as legal testimony, user manuals, and government briefings, into coherent interface
logics. The methodological stakes of this approach are significant: rather than claiming
definitive knowledge about a system’s functionality, design fiction acknowledges the
partial and interpretive nature of such reconstructions, while making those interpretive
boundaries explicit through annotation, commentary, and visual distinction.
Still other scholarship addresses the narrative and discursive dimensions of design
fiction. Blythe (2014, 2017) emphasizes the role of storytelling and character development
in transforming abstract system behavior into lived experience. His work explores
how fictional case studies, when well-grounded in context, can illuminate ethical
concerns and systemic contradictions that would otherwise remain buried in institutional
jargon. The use of characters with names familiar to many Canadians like “John” and
“Alex” in the accompanying Chinook explainer video produced through design fiction
(2025) reflects this tradition, offering viewers an accessible way to encounter the
inner workings of a decision-making system otherwise hidden from public view. The
use of fictional scenarios and characters in speculative production reflects this
tradition.
Design fiction also opens up possibilities for participatory critique. Farias, Bendor,
and Van Eekelen (2022) discuss how participatory speculative design encourages collective
sense-making around future sociotechnical imaginaries. This collaborative orientation
expands the utility of design fiction beyond individual authorship or institutional
critique, embedding it within broader conversations about public agency, civic accountability,
and alternative policy futures. Tsekleves et al. (2022) likewise demonstrate how speculative
design has been used in Malaysia as a tool for inclusive policy development, suggesting
its relevance beyond Western contexts and across diverse political regimes.
Synthesizing different bodies of literature, we found that design fiction scholarship
has focused on emerging technologies and participatory imagination; however, it has
rarely been used for evidentiary reconstruction of existing, hidden state systems.
We should also clarify that the transformative question of “what if” is not exclusively
a design fiction one; it is also a question that is being frequently asked in interdisciplinary
fields such as design justice and critical infrastructure studies. Design justice,
for instance, answers the question of “what if” to provide essential ethical frameworks
to include marginalized communities and challenge the multiple social classifications
(such as race, gender, class, and sexual orientation), which Patricia Hill Collins
describes as the “matrix of domination” (see Costanza-Chock, 2020). Similarly, critical
infrastructure studies expand this perspective and show how a specific infrastructure
can be beneficial and functional for some while structurally hostile and broken for
others (Liu, 2026). Costanza-Chock (2020) demonstrates that by using the methods of
speculative design, practitioners can break free from present-day resource constraints
to engage in “radical visioning.” This allows us to propose and prototype alternative
technologies that do not yet exist, and build a sustainable, pluriversal world where
many worlds fit (Costanza-Chock, 2020; Liu, 2026).
Finally, design fiction demands reflexivity. Galloway and Caudwell (2018) stress that
such work must be evaluated not by how closely it approximates reality, but by how
clearly it communicates its commitments, stakes, and interpretive limitations. By
foregrounding its methodological boundaries and limitations, a speculative reconstruction
like Chinook becomes an invitation to question how power, automation, and discretion
are encoded into systems, especially those that shape outcomes in areas that can have
negative consequences for disempowered groups such as immigration, policing, and public
services.
The Practice of Design Fiction
Having defined design fiction, we now turn to how it works in practice and why it
suits the study of an opaque system like Chinook. In this project, we adopt design
fiction as a critical method for examining Chinook, a system that, as Tao (2022) and
Daponte (2021a) point out, is intentionally shielded from public scrutiny.
Chinook’s technical infrastructure, decision-support features, and user interfaces
have never been fully disclosed. Key documents remain redacted, and officers' working
notes are regularly deleted, leaving minimal audit trails (Immigration, Refugees and
Citizenship Canada, 2021a; Daponte, 2021b). Standard forms and refusal templates reduce
variation in decision-making but obscure how discretion is exercised in practice (Tao,
2022). While some public-facing commentary and policy disclosures exist, they are
often fragmented, delayed, or contradictory (Colosi, 2023; Zahid, 2022). This evidentiary
gap between what is observable and what is withheld calls for a method that can responsibly
speculate from the margins.
Design fiction allows us to work within these margins. Rather than reverse-engineering
Chinook from leaked source code or unauthorized access, we construct speculative prototypes
grounded in publicly available but partial evidence: affidavits, court testimony,
refusal letter samples, and Access to Information and Privacy (ATIP)-disclosed training
materials (Daponte, 2021a, 2021b; Immigration, Refugees and Citizenship Canada, 2021a,
2021b; Tao, 2022). In Canada ATIP requests concerning government information subject
to the Access to Information Act may be made by citizens, permanent residents, any person residing in Canada, corporations,
and of course may be made by legal counsel. These sources combined form the evidentiary
spine of the reconstruction, allowing us to anchor fiction in fact while marking all
interpretive leaps. The result is not a simulation, but a provocation, a way to represent
how the system might function in ways that demand scrutiny (Gaver, 2012; Galey, Ruecker,
and Ruecker, 2010).
To provide a concrete overview of how we implemented this method, we first organized
Chinook's known components into a modular map, a diagram setting out the system's
constituent modules and the relationships between them, based on disclosed documentation.
We then prototyped selected modules in Figma, a collaborative digital design platform
suited for interface modeling and interaction mapping (Figma, 2019). Design conventions
were drawn from both public sector user interfaces and commercial decision-support
tools. Where documentation was missing, we generated visual structures and interface
behaviors, meaning the ways in which screen elements respond to user action, such
as what happens when a toggle is switched or how a refusal field is populated, that
matched the logic described in testimonies or legal analyses. For example, the batch
refusal templates or indicator management flags (Tao, 2022; Daponte, 2021a). These
speculative components were explicitly marked and explained in our accompanying documentation.
In the prototype, color-coded annotations distinguished between interface elements
grounded in evidentiary screenshots (in blue) and those inferred from textual descriptions
(in red), with each component cross-referenced to its evidentiary source. At no point
was factual ambiguity concealed; rather, we treated it as a design constraint that
demanded transparency and annotation (Blythe, 2017; Coulton et al., 2017).
Design fiction, in this context, operates as both a critique and a heuristic. It critiques
the opacity of digital bureaucracy by revealing how even stripped-down systems can
shape outcomes through interface defaults and workflow design (Galloway and Caudwell,
2018; Lindley and Coulton, 2016). It also offers a heuristic approach to studying
inaccessible systems, an alternative to traditional audit methods when researchers
lack insider access or traceable logs (Zimmerman et al., 2007; Farias et al., 2022).
To operationalize this approach, we draw on the available evidence base for Chinook,
mapping sources that could inform its speculative reconstruction.
Evidence Base and Source Mapping in Chinook
The design fiction prototype of Chinook was constructed using publicly available but
fragmented documentation. Given the absence of source code, system architecture diagrams,
or direct IRCC disclosures, our methodology relied on a distributed archive of legal,
journalistic, research, and institutional evidence. These sources provided varying
degrees of insight into how Chinook may function, which we used to inform the structural
logic, interface design, and speculative elements of our prototype (Daponte, 2021a;
Immigration, Refugees and Citizenship Canada, 2021a; Tao, 2022).
Much of what is known about Chinook comes from litigation-related materials. The Daponte
Affidavit (Ocran v. Canada) and its accompanying exhibits outline the basic modules
of Chinook, including File Management, Pre-Assessment, Decision-Maker, Post-Decision,
and Indicator Management (Daponte, 2021a). The cross-examination of Andie Daponte
added critical clarification regarding discretionary use, refusal note templates,
and the use (and deletion) of working notes (Daponte, 2021b). These materials provided
our clearest view of the system’s logic, workflows, and institutional intent.
This legal foundation was complemented by multiple Federal Court rulings, which fall
into two lines. In Safarian v. Canada and Khosravi v. Canada, decided by Justice Grammond
within a week of each other, the Court set the refusals aside. It held that boilerplate
reasoning is not objectionable in itself, but that a reviewing court must be satisfied
the officer turned their mind to the facts of the case, and observed in Khosravi that
if an assisted decision-making tool gives an officer a truncated view of an application,
the resulting decision may well be unreasonable (Khosravi, 2023; Safarian, 2023).
A second line of cases has been less receptive. In Jamali, Shirkavand, and Zargar,
the Court held that the mere use of Chinook, without evidence that it led to an unfair
decision, does not establish a breach of procedural fairness, while declining to rule
out that such issues could arise on a fuller evidentiary record (Jamali, 2023; Shirkavand,
2023; Zargar, 2023).
The reasoning in that second line begins earlier, with Haghshenas v. Canada and Raja
v. Canada, both challenges to C11 work permit refusals covering entrepreneurs and
the self-employed. In Haghshenas the Court accepted that the decision had input assembled
by artificial intelligence, but held that judicial review is concerned with the reasonableness
of the decision an officer made, whether or not such tools were used. In Raja the
Court found no evidence connecting the use of Chinook to a breach of fairness in the
officer's assessment, while indicating that such an argument remained open on a proper
record (Haghshenas, 2023; Raja, 2023).
Ocran is significant for a different reason. In dismissing the application on other
grounds, the Court left open the question of whether a certified tribunal record is
deficient when Chinook spreadsheets are neither retained nor included in it, an issue
that goes directly to what the system records and what it discards (Ocran, 2022).
Taken together, these decisions are valuable less for their outcomes than for the
descriptions of Chinook they place on the record, including how it interfaces with
the Global Case Management System used by IRCC and how it presents applications to
officers in aggregate. These judicial references helped us infer the probable function
of interface components, such as toggle-driven refusal sections and editable note
fields, which we then reconstructed in our speculative prototype.
In parallel, Access to Information and Privacy (ATIP)-released documents such as the
Chinook 1.5 user manual and internal training decks allowed us to visualize partial
screenshots, field labels, and interface behaviors (Immigration, Refugees and Citizenship
Canada, 2021a, 2021b). While many portions of these documents were redacted, enough
structural detail remained to suggest how officers might navigate through different
modules, and how indicators or risk flags were managed across cases. These training
materials frequently referenced “like-case” logic (batching/grouping applications
with similar attributes for streamlined processing), dropdown categorization, and
pre-assessment shortcuts, all of which influenced how our interface was built in Figma.
Beyond court and training documents, we consulted official records from the House
of Commons Standing Committee on Citizenship and Immigration (CIMM), especially the
February 2022 sessions that addressed the development and implementation of Chinook
(Zahid, 2022). Government officials repeatedly emphasized that Chinook is “not artificial
intelligence,” yet described operational behaviors, such as risk-based sorting and
refusal generation, that functionally resemble automated decision support. This rhetorical
distancing from AI terminology informed our decision not to visually depict Chinook
as algorithmic, but rather as an acceleration layer over human judgment constrained
by interface defaults. However, we also recognize that the insistence of federal officials
that Chinook is not AI may work to shield Chinook from the accountability mechanisms
in place for systems that are labeled AI. Côté-Boucher and Paquet (2026) show that
this framing proved effective: as the dispute moved from immigration lawyers into
media and parliamentary debate, the question of whether Chinook constituted artificial
intelligence displaced the original concerns about procedural fairness, leaving the
department's account of the tool largely intact.
In addition to official records and court disclosures, legal commentary and public-facing
blogs provided important interpretive context, especially where formal documentation
was silent or ambiguous. Much of this perspective was drawn from the writings of immigration
lawyer Will Tao, whose detailed commentaries helped interpret the policy implications
of Chinook’s deployment (Tao, 2022). His reflections on transparency, fairness, and
due process framed the ethical stakes of our design fiction, grounding it in lived
practitioner concerns. Tao's blog and Meurrens Law further contributed insight by
tracking public statements from IRCC, refusal rate trends, and evolving government
rhetoric about Chinook's role in application triaging (Tao, 2022; Meurrens Law, 2024).
Other critical reflections, such as those from Meurrens Law, outlined the rise of
advanced analytics and automation within Canadian immigration, reinforcing our decision
to position Chinook not as an anomaly, but as part of a broader shift toward computational
governance (Meurrens Law, 2024). These legal and media sources did not supply interface
details, but were instrumental in tracing patterns of concern that informed the speculative
direction and tone of our prototype.
Drawing Boards for the Design Fiction of Chinook
Before any digital mockups or video explainers were produced, the foundation of the
Chinook design fiction was built on drawing boards, literal and conceptual. These
spaces, whether whiteboards in shared rooms or digital sketchpads during calls, became
our initial laboratory for making sense of the evidence. They also allowed us to think
aloud, sketch freely, and test the spatial logic of the interface as described in
the aforementioned fragmented documents.
The process was grounded in collaboration. Team members reviewed legal affidavits,
court decisions, redacted training manuals, and policy briefings together, line by
line, while actively translating excerpts into sketched components. For instance,
when the Daponte affidavit described the “Decision Maker Module” we paused to ask:
what might that actually look like? Where would it appear on screen? What kind of
input or toggle would the officer use? Each question became a sketch, and each sketch
was tested against other parts of the evidence to check for consistency or contradiction
(Daponte, 2021a; Daponte, 2021b).
This phase was deeply iterative. We routinely circled back from the drawing board
to the documents and back again, using the visual space as a kind of evidence map.
Sometimes we revised interface sketches based on a new line discovered in a court
decision (Khosravi, 2023; Safarian, 2023). At other times, a note scribbled on the
board raised questions that sent us back to the documents for clarification. As Galey,
Ruecker, and Ruecker (2010) put it, evidence informed the sketches, and sketches in
turn reshaped how we interpreted the evidence.
The drawing board process was not just preparatory. It was a crucial interpretive
step that gave form to otherwise abstract or bureaucratic language. It let us work
through ambiguity with others, flag contradictions, and surface new lines of inquiry
before anything was formalized. More importantly, it made visible our commitments:
to transparency, to careful reading, and to treating speculation as disciplined public
inquiry, not as fantasy (Blythe, 2014; Dunne and Raby, 2013). At the same time, we
fully recognize that our reconstruction may not completely capture Chinook as it operates
in practice. By explicitly marking uncertainties and sharing provisional interpretations,
we aim to invite feedback and correction, positioning the prototype as an open contribution
to ongoing dialogue rather than a final or authoritative account.
This early stage also clarified the limits of the available information. By sketching
directly from the language of the documents, we confronted the gaps, omissions, and
redactions embedded in official records (Immigration, Refugees and Citizenship Canada,
2021a; Daponte, 2021a). Some modules could be diagrammed with relative confidence,
while others remained vague, requiring conditional interpretations. Rather than resolving
these uncertainties prematurely, we used the boards to hold multiple possibilities
in parallel, treating ambiguity as a methodological resource rather than a flaw (Galey,
Ruecker, and Ruecker, 2010; Gaver, 2012).
The redactions became a design problem in their own right, and working through them
taught us more than reading around them had. A blacked-out passage in a training manual
is not just missing information. On a drawing board it takes up space, and something
has to fill it. We could leave the area blank, which would misrepresent a system whose
officers clearly see something at that point in the workflow, or we could sketch a
plausible component, which would credit the department with a design it has never
confirmed. Neither choice is neutral, and a sketch cannot postpone the decision the
way a paragraph can, since an interface has no blank spots when someone is using it.
Running into this again and again made something clear that reading had not. The redactions
are not gaps in an otherwise complete account. They shape the account we have, which
means we were never reconstructing Chinook. We were reconstructing Chinook as the
state has allowed it to be seen.
In this sense, the drawing boards operated not only as a site of visualization but
also as a space for methodological reflection. They helped distinguish what was explicitly
stated from what was inferred, and what remained speculative. This allowed us to trace
the threshold between evidence and interpretation and to begin building a design fiction
that remained accountable to its source material (Blythe, 2017; Coulton et al., 2017;
Galloway and Caudwell, 2018).
Figma Digital Prototype and the Design Fiction of Chinook
Following the analog sketching phase, the project advanced to digital prototyping
using Figma, with a concentrated focus on reconstructing Module 3: Decision Maker.
This module received the most consistent and detailed references across internal documentation,
court affidavits, and public commentary. It is understood to be the core environment
where immigration officers review cases, apply risk indicators, and generate final
decisions using standardized refusal templates (Daponte, 2021a; Immigration, Refugees
and Citizenship Canada, 2021a).
Figma was selected for its flexibility in modular interface construction and its ability
to support collaborative design and revision. Module 3 was developed using the structure
and terminology described in key source materials. Particular attention was given
to the arrangement of decision-making fields, drop-downs for boilerplate refusals,
embedded notes, and batch processing tools. These elements were anchored in direct
references wherever available and were annotated accordingly within the prototype.
Visual fidelity was prioritized for interface elements that had clear correlates in
the evidence. Terms like “Refusal Notes,” “Group Numbers,” and “Working Notes” were
used verbatim from policy and training documents. Interface groupings, such as applicant
summaries, risk indicators, and decision input areas, were laid out based on inferred
officer workflows, especially in contexts involving high-volume triage or streamlined
refusals.
Where source material lacked specificity, design choices were guided by institutional
logic and interface conventions typical of government case management systems (Galloway
and Caudwell, 2018). These speculative components were clearly distinguished using
visual labels and were cross-referenced with corresponding justification in the project
documentation. This method preserved interpretive transparency and emphasized the
boundaries of what could be responsibly reconstructed.
Two components were harder to build than we expected, and in both cases the problem
turned out to be in the system rather than in our drawing of it. The first was working
notes. We know officers write notes while assessing a file, and we know these notes
are routinely deleted, leaving little audit trail (Daponte, 2021b; Immigration, Refugees
and Citizenship Canada, 2021a). But an interface has to put the field somewhere, and
wherever we put it, it looks like a record. A bordered text box inside a case view
reads as documentation. It implies the text will be kept and can be found later. We
could not find a way to draw a field whose main institutional feature is that what
you write in it disappears. We included it and annotated its status, but the failure
is worth stating, because it points to something odd about the tool itself: Chinook
gives officers a place to write down reasons that the record does not keep. This is
not only our reading. In Mehrara, the Federal Court observed that IRCC's systemic
deletion of materials generated by processing technology may not reflect best practice,
though it declined to find the omission relevant on the facts before it (Mehrara,
2024; Otchengco, 2025). What we could not draw is a feature the Court has also noticed,
without either of us being able to say what the deleted notes contained.
The second was batch processing. The documents describe grouping applications by shared
attributes, including citizenship and school, and handling them together (Tao, 2022;
Daponte, 2021a). Drawing this meant designing a control, and a control takes the same
effort no matter what it does. Ticking a box for forty files is the same gesture as
ticking it for one. We could not draw the element in a way that made that difference
visible, which is the point: the interface does not make it visible either. The gap
between how much the outcome matters to an applicant and how little effort the action
takes is not something our reconstruction introduced. It is how the tool works.
A third problem ran through the whole phase. IRCC says Chinook is not artificial intelligence,
while the same documents describe risk-based sorting and refusal generation (Zahid,
2022; Côté-Boucher and Paquet, 2026). There is no standard way to draw that combination,
and an image cannot hold a contradiction the way a sentence can. Our solution, showing
an acceleration layer over human judgment constrained by interface defaults, is a
position we had to take, not one the department has stated. Having to draw something
forced us to settle a question the documents leave open.
The prototype evolved through multiple revision cycles. As additional evidence was
incorporated or inconsistencies were identified, the interface design was adjusted
for coherence and plausibility. Figma’s component system enabled efficient updates
across screens, ensuring uniformity while allowing for iterative refinement. Each
revision was tracked and aligned with updates to the video explainer and documentation,
maintaining internal consistency across project outputs.
By the end of this phase, the digital prototype offered a structured and interpretable
reconstruction of Module 3. It does not simulate decision-making outcomes but instead
visualizes how such decisions may be operationalized within a semi-automated workflow.
Making the Design Fiction of Chinook Publicly Legible: Video Production
To more widely share the speculative interface reconstruction, a 20-minute explainer
video
[1] was developed to communicate Chinook’s seeming operational logic to a broader public
audience, especially those affected by Chinook. This phase translated the design fiction
and supporting research into a narrative format that emphasized accessibility, clarity,
and evidence-based reconstruction.
The storyboard was constructed based on the modular logic of the system, with particular
focus on Module 3 (Decision Maker). Sequences were developed to depict key components
and workflows as described in affidavits, internal manuals, and court testimony (Daponte,
2021a; Immigration, Refugees and Citizenship Canada, 2021a). Each scene was anchored
in available evidence, visualized through the Figma prototype, and narrated using
a plain-language voiceover with commentary drawn from the project’s documentation
and legal commentary.
Fictional placeholders were introduced for illustrative purposes. An officer named
“John” and an applicant named “Alex” served as narrative anchors to visualize both
the interface and its implications for temporary resident visa applicants. These characters
framed the system’s operation from administrative and applicant-facing perspectives.
Scenes were structured to walk viewers through login procedures, dashboard configuration,
batch processing, attribute-based filtering (e.g., by citizenship or school), use
of risk indicators, and the generation of refusal notes using standardized templates.
Animations were derived directly from the digital prototype to ensure alignment between
the storyboard and the interface. Interface behaviors, such as toggling indicators,
adjusting columns, and exporting decisions, were illustrated through simulated interactions
based on institutional logic and described functionality.
Throughout production, annotations and disclaimers were integrated to distinguish
confirmed features from speculative ones. These distinctions were reinforced through
visual markers, voiceover language, and supporting documents. The video also included
cues for cross-referencing scenes with relevant documentation, ensuring coherence
across all project outputs.
Assessing the Strengths and Limitations of Design Fiction and the case of Chinook
The use of design fiction in this project revealed several methodological strengths,
particularly in addressing the challenges of studying an opaque digital system like
Chinook. One of the most significant strengths was its ability to render the operations
of a black-boxed system interpretable and accessible to researchers across disciplines
and the broader public. By visualizing possible interface elements and workflows,
the project translated abstract legal and bureaucratic concepts into concrete, navigable
designs. This made it possible to engage audiences, including those outside of legal
or policy circles, in meaningful critique and discussion.
Another key strength was the method’s capacity to synthesize fragmented and heterogeneous
forms of evidence. Legal affidavits, redacted training manuals, Access to Information
and Privacy (ATIP) disclosures, and media commentary were all integrated into a shared
design logic. The visual and iterative nature of the prototyping process helped the
team triangulate across sources, testing how interface components described in one
context aligned or contradicted with details from others.
At the same time, the approach made explicit its own assumptions and interpretive
leaps. Speculative elements were consistently marked, annotated, and documented, reinforcing
a norm of transparency rather than authority. In this way, the design fiction methodology
encouraged reflexivity. It foregrounded the interpretive labor of the researchers
and invited readers and viewers to question what is known, what is inferred, and what
remains concealed. The speculative prototype did not pretend to represent Chinook
in full fidelity. Rather, it was a vehicle to pose difficult questions about how power
is encoded in software systems.
Design fiction also helps to visualize and interrogate opaque systems in the absence
of source code, insider access, or detailed technical documentation. It allowed for
the synthesis of disparate forms of evidence, legal, technical, policy, and narrative,
into a coherent interface logic. This synthesis was not neutral. Every design choice
was embedded with assumptions, which is why iterative documentation, internal peer
review, and public annotation were integral throughout the process.
Design fiction also helped surface ethical questions not always visible in textual
analysis alone. By placing speculative users into reconstructed workflows, the design
raises questions about discretion, accountability, and automation that were otherwise
dispersed across court documents and policy briefings. The speculative lens, in this
way, made visible the affective and procedural stakes of decision-making systems,
reinforcing critiques raised by legal advocates and civil society organizations (Tao,
2022).
The most useful strength of all, however, was the friction the method produced. Building
an interface does not let one skip past things the way writing does. A sentence can
note that working notes are deleted and move on; a screen has to show where they get
written. A paragraph can say applications are processed in groups; a control has to
specify what the officer clicks. Each of these requirements brought us to a point
where the record was not enough, not because it was redacted, but because what it
describes does not add up. The notes field that documents nothing, the control that
refuses forty applicants as easily as one, and the system that sorts by risk while
officials deny it is artificial intelligence are not readings we brought to the prototype.
They are what the prototype would not let us leave unsaid. The reconstruction worked
less as an illustration of concerns raised elsewhere and more as a way of finding
them, and the oddities it turned up are the everyday operating conditions of a bureaucracy
handling a very large volume of applications.
Still, the method carries limitations. Design fiction does not offer empirical verification.
Its strength lies in plausibility and interpretive framing, not in technical accuracy.
The final prototype does not claim to depict Chinook as it exists when using an IRCC
workstation, but rather as it may function based on what is observable from the outside.
This distinction was maintained throughout the project by marking speculative components,
avoiding deterministic conclusions, and situating all claims within a bounded evidentiary
frame (Dunne and Raby, 2013; Barendregt and Vaage, 2021).
Design fiction also raises the risk of over-interpretation. Despite annotations and
disclaimers, there is always a possibility that viewers may interpret the design as
a literal interface rather than a research artifact that involves speculation. The
narrative and visual clarity of the prototype, while useful for accessibility, can
inadvertently imply authority. Moreover, this methodology may not generalize easily
across other black-boxed systems. Chinook was a viable subject for design fiction
because of the volume of related litigation, public records, and insider testimony.
In systems where such documentation is unavailable or heavily censored, the speculative
method would struggle to meet ethical or evidentiary thresholds.
A further limitation concerns the position from which this critique is made. Our reconstruction
addresses a system whose developers have had no part in it, and the account IRCC has
given of Chinook's origins deserves to be engaged rather than treated as mere justification.
The Daponte affidavit describes Chinook as an in-house response to operational conditions:
sharply rising application volumes, and a case management system that performed poorly
over the low-bandwidth connections common at visa offices abroad and obliged officers
to move between multiple windows to assemble a single file. Chinook was not procured
from an external vendor, nor designed as an instrument of automated adjudication.
It was built by officers, in Excel, to make an unwieldy system workable against performance
targets they did not set (Daponte, 2021a). Read this way, Chinook is less an imposition
of algorithmic governance from above than an accumulation of local workarounds that
hardened into a new layer of infrastructure, which is a different problem and arguably
a more common one.
Taking that account seriously does not blunt our critique so much as relocate it.
Whether or not batch processing and templated refusals were intended to constrain
discretion, they shape how discretion is exercised, and an efficiency measure that
becomes the routine environment of decision-making acquires governing force regardless
of the intentions behind it. The question our prototype poses to IRCC is therefore
not whether its officers acted in bad faith, but why a tool built to relieve administrative
pressure has been shielded from the scrutiny that its effects on applicants would
ordinarily invite?.
Two implications follow for design fiction as a form of critique. The first is an
asymmetry of response. We reconstructed a system whose developers cannot correct our
errors without disclosing the details the department has chosen to withhold, so the
same institutional opacity that makes speculation necessary also prevents the people
best placed to refute it from doing so. Marking speculative elements and inviting
correction are procedural safeguards, but they do not dissolve this asymmetry. The
second implication concerns what happens to a speculative artifact once it circulates.
A reconstruction that is visually coherent and easy to follow may come to stand in
for the system in public debate, displacing the messier object it represents. The
Chinook controversy has already shown how readily attention can migrate from the substantive
question of procedural fairness to a narrower dispute over whether the tool meets
the definition of artificial intelligence, a framing that has served the department
well (Côté-Boucher and Paquet, 2026). A design fiction can contribute to that drift
as easily as it can resist it, which is why we have tied every element of the prototype
to a cited source and treated the reconstruction as an argument to be contested rather
than a depiction to be believed.
Maintaining coherence between the written documentation and the animated explainer
was central to the methodology. Each segment of the video is linked to a corresponding
section of the documentation, enabling viewers to cross-check the visual narrative
against cited evidence. The animation serves as an accessible entry point, while the
documentation provides detailed context and references.
[2]
Together they operated as complementary outputs: the video communicates the system’s
speculative logic to a wider public, and the documentation reinforces transparency
by showing exactly how each design choice was grounded or marked as uncertain.
Finally, while ambiguity was treated as a resource during the sketching and prototyping
phases, it also presented challenges. Ambiguity enabled a more open-ended exploration
of possibilities, but it complicated efforts to communicate clear findings to policymakers
or the public. Without sufficient framing, speculative work risks being dismissed
as conjecture rather than grounded inquiry. In an era often described as “post-truth,”
where misinformation and skepticism toward expertise are widespread, this risk becomes
even more pronounced. For this reason, every stage of the project emphasized interpretive
discipline, cautious language, and consistent citation, even when working within the
speculative space.
Conclusion
This article has explored the use of design fiction as a methodological tool for the
critical modeling and examination of Chinook, an internal decision-support system
used by the IRCC. By grounding our speculative reconstruction in court affidavits,
redacted training manuals, and legal commentary, we were able to prototype Chinook’s
Decision Maker module (module 3) and visualize how its interface may influence visa
decision-making processes.
Each phase of the project (such as drawing board sketching, Figma prototyping, and
video production) worked to show the interpretation required when dealing with partial,
contradictory, or deliberately redacted information. The interpretive work was also
confrontational. At each stage, having to build something specific exposed features
of the system that its own documentation records without seeming to notice. This is
where the urgency of the case is clearest: not in the fact that Chinook is hidden,
but in what shows up as soon as anyone tries to draw it. The resulting design fiction
is not a claim to technical accuracy but a structured provocation, meant to invite
scrutiny and encourage deeper public engagement with the procedural realities of the
digital shift in immigration governance. This approach acknowledges the evidentiary
limits of working from the outside while also affirming the importance of critical
inquiry, especially when democratic oversight is also limited.
By treating design fiction as both critique and inquiry, the article contributes to
broader debates in digital humanities and beyond about methods, algorithmic governance,
digital bureaucracy, and public accountability. Methodologically, this article contributes
to how design fiction is usually practiced. Rather than speculating from imagination,
we speculated from the available documents such as affidavits, ATIP releases, and
court rulings. These documents served as the evidentiary spine of the reconstruction
and marked the point at which that record gave out. Second, rather than directing
speculation at an emerging technology, we directed it at a bureaucratic system already
in operation, where the object of inquiry is not a possible future but a concealed
present. Design fiction, practiced this way, becomes a method of evidentiary reconstruction
suited to institutions that are documented but not disclosed. In doing so, this article
complements and extends work in digital humanities, human-computer interaction, socio-legal
studies, and policy studies concerned with the ethics of automation and the infrastructures
of state power.
Future interdisciplinary research may build on this method to explore other black-boxed
systems across policy domains, including education, healthcare, or border security.
As decision-making processes become increasingly data-driven and interface-mediated,
speculative reconstruction can help scholars from various disciplines, advocates,
and affected communities interrogate the tools that shape access to rights and resources
and powerfully determine life chances. Even emerging fields such as quantum computing,
with their potential to reshape data processing at scale, may benefit from speculative
engagement to surface assumptions before these systems solidify into institutional
norms.