Methodology

This project is grounded in the principle that data is never raw. Every dataset reflects the conditions of its production: who had access to report, which events were visible enough to document, and which categories were available to describe them.

Informed by scholarship on data, power, and documentation practices, Nabard commits to: examining power in the documentation process; identifying and addressing structural omission; preserving human experience alongside quantitative records; and documenting its own biases rather than concealing them.

Documentation is not a purely technical activity; it is embedded in political, social, and material structures.

Nabard documents events related to political unrest, civil resistance, and state response in Iran. Documented events include but are not limited to:

An event is classified as related to political activity when at least one of the following conditions is met: it occurred during or in proximity to a documented protest; official charges reference assembly, expression, or political activity; or at least two independent sources establish the connection.

These criteria are designed to prevent both the inclusion of unverified claims and the structural exclusion of less visible cases. Inclusion in the archive reflects relevance to scope, not evidentiary certainty. Relevance to scope is determined by publicly stated criteria rather than discretionary judgment. Records outside scope are excluded regardless of their moral or political significance. Scope expansion, when it occurs, follows documented revision of criteria rather than reactive inclusion. Verification status is applied after inclusion and may evolve over time.

Our operational definitions determine what becomes visible within the archive. Event-based documentation privileges harms that can be temporally and spatially identified. Structural, slow, and continuous forms of harm may remain underrepresented. Forms of harm that are cumulative, diffuse, or embedded in long-term policy may not register within event-based systems in the same way. This does not imply lesser significance, only different archival visibility.

Categories and definitions are subject to revision as new patterns emerge or blind spots are identified. Methodology changes are versioned, timestamped, and archived. Previous versions remain accessible for review.

Classification shapes how events become searchable, comparable, and interpretable. For this reason, Nabard treats its taxonomy as a versioned public object rather than a fixed framework.

Categories, definitions, and structural groupings are timestamped and archived. When categories are added, merged, or redefined, the rationale is documented. Previous versions remain accessible.

Records may carry multiple tags to reflect complexity. Reclassification history is preserved as part of each record’s archival trace.

Operational definitions privilege certain forms of harm over others. Event-based categories foreground temporally bounded violence. Structural, cumulative, or slow forms of harm may not register within the archive in the same way. These limitations are documented through periodic methodological review.

Data is drawn from domestic and international media reporting, human rights organizations, social media platforms, community reports, direct testimonies, and official records.

No source category is treated as inherently definitive. Official sources may reflect censorship or distortion. Community sources may be incomplete or unverified. Media coverage is uneven and influenced by access, language, and editorial priorities.

Each source is evaluated on its own terms, accounting for its specific constraints, incentives, and risks of misrepresentation. No source is privileged solely by institutional status, follower count, or proximity to power. Source consistency is evaluated over time, not assumed at first appearance.

Nabard operates automated pipelines that continuously monitor publicly available content from social platforms, including Telegram channels, Instagram accounts, posts on X, and YouTube channels. The platform currently ingests approximately 500 new resources per day from these channels. Automated collection increases coverage but does not substitute for human judgment.

This automated layer is complemented by direct intake from community reporting, human rights documentation, and media monitoring.

After ingestion, each record passes through a structured labeling workflow:

1. Designated labelers review and classify records for searchability, categorization, and identification.

2. Reviewers independently assess each labeled record for accuracy, completeness, and consistency.

3. When a labeler and reviewer disagree, the dispute is escalated to the core team and resolved through anonymous vote with equal weight per member.

This separation of roles is a deliberate control designed to reduce individual bias and maintain quality across the archive.

Automated ingestion increases coverage but does not neutralize bias. Platform policies, digital access inequality, urban connectivity, language distribution, and diaspora amplification patterns shape archival visibility.

To address this, Nabard publishes periodic Coverage and Bias summaries including:

Changes in platform access may shift archival visibility. These shifts are documented.

Bias cannot be eliminated. It can be measured, disclosed, and revised.

Verification is not a single step but an iterative process. Each record is assessed through cross-source comparison, chronological and geographic consistency checks, and contextual review of supporting media.

Records carry an explicit verification status:

New records may be published with incomplete information, but always with a transparent status label. Nabard is committed to updating records as new evidence emerges. Verification status may change in either direction. Confirmation does not imply completeness.

Verification status reflects evidentiary confidence, not moral or political weight. Verification categories are epistemic tools for communicating confidence, not instruments of validation or invalidation. A record categorized as “Needs Additional Information” represents a claim requiring further substantiation, not a lesser form of harm.

In conditions of repression, information is structurally incomplete. Nabard does not treat this as a deficiency to be hidden but as a condition to be documented.

Approximate dates are recorded with explicit ranges. Names may be published as pseudonyms when safety requires it. Incomplete records are marked rather than excluded. Contradictory accounts from different sources are documented and contextualized rather than silently resolved.

Uncertainty is not eliminated; it is made visible. In some cases, uncertainty may persist indefinitely. Archival responsibility does not require epistemic closure. Closure may be impossible in environments where evidence is systematically suppressed.

The safety of individuals, including those documented, those who report, and their families, is a primary constraint on all editorial decisions.

Personally identifying information is collected only when necessary and minimized in publication. Anonymous reporting is supported. IP addresses and identifying metadata are neither stored nor published.

Requests for removal or correction from families and affected individuals are reviewed and honored when justified. When a record raises safety concerns, any team member can flag it for review; publication is held until the core team reaches consensus through anonymous vote.

As an aggregator of publicly available content, the information we archive is already accessible elsewhere. Public availability does not automatically justify amplification. Our safety obligation is to ensure that our curation, classification, and presentation do not create new risk that did not exist at the source. No record is worth publishing if our handling of it endangers the people it concerns. In such cases, omission is an ethical decision, not a failure of documentation.

Publication decisions involve visible trade-offs. No procedural safeguard fully eliminates risk. The decision to publish always weighs public record value against potential harm. Each case requires contextual judgment.

Making information searchable increases analytical value. It may also increase extractability.

Nabard operates under the assumption that structured data can be scraped, analyzed, or repurposed. Legibility is therefore balanced against safety through data minimization and selective omission.

Operational protest logistics, private contact networks, and relational graphs that could expose vulnerable individuals are not documented. Certain fields may be generalized even when publicly available elsewhere. Publication may be delayed or withheld when amplification would introduce new risk.

Bulk data exports are provided through controlled channels. Automated access is monitored to reduce abusive extraction.

Legibility is treated as a design constraint rather than an absolute goal.

Violence occurs both as discrete events and as recurring practices.

Where sufficient data exists, Nabard links individual records to broader recurring methods and longitudinal patterns. Event records may therefore reference related practices or aggregated pattern summaries.

This layered structure allows the archive to capture both rupture and continuity while remaining grounded in documented evidence.

Nabard acknowledges specific structural limitations in its documentation:

These limitations are not incidental; they are produced by the same power structures the project seeks to document. They are stated publicly, factored into analysis, and addressed through ongoing efforts to diversify sources and expand reach. Absence of documentation does not imply absence of harm. Patterns within the archive should therefore be interpreted as reflections of documented visibility, not total reality. Quantitative summaries derived from the archive should be understood as descriptive of recorded material, not exhaustive population measures.

Digital trace availability, urban connectivity, language access, and diaspora reporting patterns structurally influence visibility. These are not merely data gaps but reflections of unequal access to communicative infrastructure. Metrics of visibility—volume, frequency, repetition—may reflect communicative power rather than proportional harm.

Nabard currently operates in Persian and English. We recognize that this excludes many linguistic communities within Iran. Translation decisions necessarily shape meaning and are treated as editorial acts rather than neutral conversions.

Nabard does not quantify moral severity, rank suffering, or assign comparative value to different forms of harm. The archive records occurrences; it does not construct hierarchies of grief, legitimacy, or worth. It does not adjudicate truth claims beyond the evidentiary thresholds described above.

Nabard does not position itself as a guarantor of authenticity. In an environment shaped by state propaganda, partisan media with undisclosed funding, and increasingly convincing synthetic content, including AI-generated video, fabricated audio, and manipulated footage, no single organization can independently verify everything it archives. We acknowledge this limitation.

There is a deliberate tradeoff between coverage and certainty, and we prioritize building a comprehensive archive over claiming authority over every record. What we guarantee instead is transparency.

Every resource on the platform carries a full information panel showing its original source link, publication date, edit history, and all community reports received, categorized by type. Every resource links directly to its original source, allowing users to independently trace provenance and verify claims against the primary material.

Since Nabard operates as an aggregator, nearly all resources originate from external sources and are linked back to them; direct submissions are clearly tagged as such. Our labelers verify source attribution during the classification process, and community reporting provides an additional correction layer for misattributed or misrepresented content.

Users can report any resource. When a resource accumulates an anomalous volume of independent reports, it is automatically removed from public view pending re-verification. This threshold is adaptive rather than fixed; the system accounts for reporting velocity, account behavior patterns, and other signals designed to distinguish genuine community concern from coordinated manipulation.

Nabard can only become a reliable source if the community actively participates in correcting, contesting, and refining the archive. Our responsibility is not to be infallible, but to make every decision, edit, and report visible. Transparency is treated as a corrective mechanism, not a claim of authority. Authority, where it exists, derives from traceability rather than institutional status. Institutional longevity does not substitute for evidentiary transparency.

Disagreement between labelers, reviewers, or community reports is treated as part of the archival trace. Reclassification history is preserved. The archive increases reliability over time through accumulated revision rather than through claims of initial certainty.

Nabard documents events; it does not substitute for political organization or collective action. Documentation preserves record; change requires broader social processes beyond the archive.

Once publicly released, data may be used in ways beyond our control. Documentation of state violence can be incorporated into geopolitical narratives, advocacy campaigns, or policy arguments.

Nabard does not align itself with any specific political agenda or geopolitical position. Our commitment is to evidentiary rigor and methodological transparency, not to particular policy outcomes. The archive may inform debate, but it does not prescribe outcomes. Inclusion of an event in the archive should not be interpreted as endorsement, condemnation, or strategic alignment with any political actor involved.

Nabard is built as a research infrastructure for journalists, researchers, human rights advocates, political analysts, and anyone seeking reliable, structured information on Iran.

Records are organized across categories including individuals, events, geographic locations, organizations, media evidence, and editorial content, each with structured fields for verification status, source attribution, and bilingual documentation in English and Persian.

Every resource on the platform can be individually downloaded and shared via its unique link, accessible through the resource menu. Researchers and organizations needing bulk data exports can request them by contacting us directly.

Content on Nabard is published under a Creative Commons Attribution 4.0 International (CC BY 4.0) license. You are free to share, redistribute, and adapt the material for any purpose, provided you give appropriate credit, indicate if changes were made, and do not impose additional legal restrictions.

Since Nabard operates as an aggregator of publicly available content, this license applies to our curation, classification, and editorial contributions; original source rights remain with their respective owners. Attribution should reference Nabard (nabard.xyz) as the source. Derivative analyses or visualizations must clearly distinguish between Nabard’s data and external interpretation. Nabard is not responsible for conclusions drawn from derivative works.

We actively encourage independent scrutiny of both our data and our methodology.