Scenario Lab — Scenarios 7–9
Scenario 7 — The “Anonymous” Mobility Dataset
Difficulty: Difficult
Decisive clue: direct identifiers are gone, but combination reconstructs identity.
A city removes names and emails from detailed mobility traces and releases the dataset for research. A third party combines the traces with public home/workplace information and re-identifies people.
Model answer
Obfuscation/redaction reduced direct identification but did not guarantee anonymity. Downstream combination/re-identification risk still requires Governance and context-aware protection.
Review whether remaining attributes can identify people when combined, who can access/use the data, likely harms and whether aggregation/masking/classification/governed access fit the context.
Weaker: “Names were removed, so the ethical duty is complete.”
Changed fact: aggregation to a level that prevents individual tracing, plus governed access/use, could materially reduce the risk.
Operational mapping: Leading KA Data Security; supporting Data Governance and Metadata Management. Roles: Data Owner, security/privacy team, Governance and sharing/research stakeholders.
Source: §3.4.6, pp. 62–63.
Scenario 8 — The Ethics Statement Nobody Can Use
Difficulty: Difficult
Decisive clue: values statement exists; practices, controls, training, monitoring and safe escalation do not.
An organization publishes a polished Responsible Data statement. Internally, employees get no training, speed is rewarded over review, incidents are not audited, no safe escalation channel exists and employees fear retaliation.
Model answer
This is an ethical-culture failure. Start with current-state review and build principles/risks/practices/controls, strategy/roadmap, training/communication, monitoring, safe escalation and leadership/Governance support.
Weaker: “Buy automated ethics-monitoring software.” Automation cannot replace human judgment, culture or safe reporting.
Changed fact: if culture mechanisms were strong but the organization lacked a written policy/code, formalizing that narrower gap could become the next priority.
Operational mapping: Leading KA Data Governance; supporting all Data Management functions that must implement the practices/controls. Roles: executive sponsor, Governance, managers, practitioners and audit/compliance/HR escalation stakeholders.
Source: §3.5, pp. 63–67.
Scenario 9 — Who Owns the High-Impact Scoring Decision?
Difficulty: Expert-discrimination
Decisive clue: consequential personal-data scoring with no Governance review.
A business unit wants to deploy a personal-data score that determines favorable financing terms. Data scientists call model performance technical. Legal sees no explicit prohibition. Governance has not reviewed data sources, fairness risk, use limitation or downstream consequence.
Model answer
Data Governance should review the handling standards and decision with legal counsel, while the business/data-science team remains responsible for surfacing and explaining ethical risks across:
- population selection;
- data/behavior capture;
- analysis/modeling;
- results/access/use.
Legal interpretation is necessary but is not the complete ethical standard. Governance is necessary but does not eliminate practitioner responsibility.
Weaker: “Let Legal decide.”
Changed fact: a low-impact internal experiment using non-personal data with no consequential decision could justify less intensive personal-data review, though general Data Ethics still applies.
Operational mapping: Leading KA Data Governance; supporting Security, Metadata and Data Quality depending on the scoring data. Roles: Governance council, Legal, business owner, analysts/data scientists and Data Owner/Steward.
Source: §3.5.4–3.6, pp. 65–67.