Institute for Social Vision Design
ISVD-LAB-002Foundations
1.4.1

Research Framework of the Agnotology Lab — An Inductive Coding Framework

From the perspective of agnotology, this note presents an inductive coding framework for multidimensional analysis of the 'production of ignorance.' Moving beyond conventional domain-based classification, a seven-axis tagging system structures research notes and allows cross-disciplinary patterns to emerge from the data.

This note provides an overview of the research framework of the Agnotology Lab. For details on each topic, see the individual notes.

Why Study "Ignorance"?

We are surrounded daily by vast quantities of information, yet remain ignorant of matters of decisive importance. This is no accident. The discipline that Robert Proctor (2008) named "agnotology" (無知学) investigates how ignorance is intentionally produced and maintained. The tobacco industry's decades-long, systematic concealment of evidence regarding health harms was the starting point of this field.

Tsuruta Souto and Tsukahara Togo, eds. (2025), in Invitation to Agnotology (『無知学への招待』), classified ignorance into three layers:

  1. Native Ignorance — What is not yet known. The frontier of science.
  2. Lost Knowledge — What was once known but has been forgotten.
  3. Strategically Produced Ignorance — The intentional maintenance of a state of "not knowing."

This lab focuses primarily on the third layer — strategically produced ignorance. Advertising, propaganda, media, educational systems, legal frameworks, and technology are often discussed in isolation. Yet from the vantage point of agnotology, they share the same structural patterns as "apparatuses for the production of ignorance."

Grounded Theory Coding Framework

1
MechanismsHow ignorance is produced (free tags)
2
Actor LevelIndividual / Organization / Institution / Social-field / Technical-system
3
Actor DetailsSpecific actors (free tags)
4
TargetsSuppressed knowledge domains (free tags)
5
IntentionalityStrategic / Structural / Emergent
6
Power DirectionTop-down / Lateral / Bottom-up / Algorithmic
7
Geo-CulturalRegional/cultural context (free tags)
Fig: 7-Axis Coding Framework — Multi-dimensional tagging for each research note

Research Approach: Inductive Dataset Construction

Why Single-Axis Classification Systems Fail

The phenomena addressed by agnotology are inherently multidimensional. A single case may simultaneously constitute advertising and propaganda, be jointly executed by state and corporate actors, and be reproduced through the education system — such layered complexity is the essence of ignorance production.

This lab initially designed a classification system based on six research domains (advertising, propaganda, media, education, religion/authority, national character/group psychology). However, this approach suffered from structural problems.

  • Inconsistent axes: D1–D5 were organized along an actor axis (who produces ignorance), while D6 (national character/group psychology) constituted a cultural-contextual axis — the classification criteria were not uniform.
  • Overlap: Advertising (D1) and propaganda (D2) merge in reality as government communications and political advertising, making mutually exclusive categorization difficult.
  • Gaps: Important mechanisms of ignorance production — restriction of knowledge through legal systems (secrecy laws, patent regimes), architectural control through technology (algorithmic filtering), and knowledge disparities arising from economic structures — fell outside the system.

Single-axis MECE classification is, in principle, unsuited to multidimensional phenomena such as those studied by agnotology.

Grounded Theory-Style Structured Coding

As a new approach, this lab adopts inductive dataset construction following grounded theory methodology. Rather than sorting cases into predefined domains, we assign multidimensional tags to individual research notes and allow classification taxonomies to emerge from accumulated data.

Each research note's frontmatter is tagged along the following seven coding axes:

#AxisDefinitionType
1mechanismsHow ignorance is produced (mechanism)Free tag
2actorsOrganizational level: individual / organization / institution / social-field / technical-systemConstrained tag
3actorDetailsSpecific actor namesFree tag
4targetsWhat kind of knowledge is suppressedFree tag
5intentionalitystrategic (intentional) / structural / emergentConstrained tag
6powerDirectiontop-down / lateral / bottom-up / algorithmicConstrained tag
7geoCulturalGeographic and cultural contextFree tag

"Constrained tags" are selected from predefined options, while "free tags" may be created anew as cases demand. Free tags grow in vocabulary as research notes accumulate, eventually forming bottom-up classification systems through clustering and co-occurrence analysis.

This approach offers three advantages:

  1. Preservation of multidimensionality: Multiple mechanisms, actors, and targets can be assigned simultaneously to a single case. For the tobacco industry case, one can write mechanisms: [doubt-manufacturing, regulatory-capture], actors: [organization, institution], actorDetails: [tobacco-industry, regulatory-agency], thereby describing the complexity of the phenomenon as it is.
  2. Structural prevention of gaps: When a new case cannot be described by existing axes, this signals the need to expand an axis or add a new one. In fixed domain classifications, cases that do not fit the framework are simply ignored.
  3. Facilitation of discovery: Co-occurrence patterns among tags reveal structural similarities that researchers did not anticipate in advance. The "useful ignorance" identified by McGoey (2012; 2019) — the structure in which not knowing serves a function for those in power — is a paradigmatic example of patterns that emerge from precisely this kind of inductive analysis.

Cross-Cutting Themes

Tags along the mechanisms axis form cross-cutting themes as structural patterns that transcend individual cases.

Cross-Cutting ThemeRelated MechanismsSummary
Cost Asymmetrydoubt-manufacturing, attention-controlThe cost of producing falsehood < the cost of correcting it. The asymmetry described by Brandolini's Law is a structural feature common to advertising, propaganda, and media, and has been amplified by orders of magnitude through AI-generated content.
Epistemic Injusticeepistemic-exclusionThe power structure governing whose voice is recognized as "knowledge." The pattern in which NPOs' field knowledge is dismissed as "subjective" is a typical instance of this mechanism (→ Details: Epistemic Injustice in NPOs and Information Access Disparities).
Strategic Ignorancestrategic-ignoranceInstitutional mechanisms for intentionally keeping inconvenient knowledge at bay. The pattern whereby evidence is dismissed as "insufficient data" in EBPM is typical (→ Details: The Obstructive Effects of Strategic Ignorance in EBPM).
Filter Structuresattention-control, complexity-weaponizationThe invisibilization of the "unseen" through selective presentation of the "seen." The same structure repeats at different scales, from algorithmic filtering to the selective content of educational curricula.
Conformity and Silencesilence-structuringThe structuring of "knowing but not speaking" by raising the cost of dissent. The "rule by atmosphere" (空気) analyzed by Yamamoto Shichihei represents the Japanese manifestation of this mechanism.

Details: AI Amplification of the Brandolini Asymmetry

What We Should Do / What We Need Not Do

What We Should Do

  • Accumulate and structure research notes based on the seven-axis coding framework
  • Build a corpus of essays that reexamine current affairs through the structural lens of agnotology
  • Construct bottom-up taxonomies through co-occurrence analysis and clustering of tags
  • Develop an original system of agnotology rooted in Japan's sociocultural context
  • Design information literacy education as counter-design

What We Need Not Do

  • Develop fact-checking methods per se (existing organizations are already working on this)
  • Collect and verify individual fake news cases (this belongs to the domain of journalism)
  • Attack specific religions, political parties, or organizations (this is structural analysis, not denunciation)
  • Assert that "ignorance is bad" as a moral claim (the purpose is to illuminate structures)
  • Design a complete, deductive classification system in advance (we build inductively from data)

References

Agnotology: The Making and Unmaking of Ignorance

Proctor, R. N. & Schiebinger, L.. Stanford University Press

Read source

無知学への招待 — 未知・無知・不可知の人文学

鶴田想人・塚原東吾 編. 明石書店

Read source

Epistemic Injustice: Power and the Ethics of Knowing

Fricker, M.. Oxford University Press

Read source

The Russian 'Firehose of Falsehood' Propaganda Model

Paul, C. & Matthews, M.. RAND Corporation, Perspectives PE-198

Read source

Strategic unknowns: towards a sociology of ignorance

McGoey, L.. Economy and Society, 41(1), 1-16

Read source

The Unknowers: How Strategic Ignorance Rules the World

McGoey, L.. Zed Books

Read source

Calling Bullshit: The Art of Skepticism in a Data-Driven World

Bergstrom, C. T. & West, J. D.. Random House

Read source

「空気」の研究

山本七平. 文藝春秋

Post-Truth

McIntyre, L.. MIT Press

Read source

Participate in & Support Research

If you're interested in ISVD's research, we welcome your participation as a cooperating member or your support for our projects.