Kyūkei Master Hub
SCS2110 Material | Phase 8: Ethics, Integrity & Computing Dilemmas
👁️ -- opens | 0%

Material Phase 8: Research Ethics, Misconduct & Computing Dilemmas

Complete study guide for Course Unit 7. Master core ethical principles, the 3 cardinal sins of research misconduct (FFP), data privacy regulations (GDPR/POPIA), algorithmic bias, and dual-use cybersecurity dilemmas.

1. Core Ethical Principles in Scientific Research

Ethics in research protects participants, maintains scientific integrity, and ensures public trust in computing innovations.

Informed Consent

Participants must be fully informed of the study's purpose, risks, and data handling procedures, and voluntarily agree in writing before participating.

Anonymity vs. Confidentiality

Anonymity: No identifying data (IP, name) is collected at all.
Confidentiality: Identifiers are collected but stored securely under pseudonymized keys and never published.

Beneficence & Non-Maleficence

Maxmize research benefits while ensuring zero physical, psychological, social, or digital harm to participants.

2. The 3 Cardinal Sins of Research Misconduct (FFP)

Research misconduct invalidates scientific findings and results in academic expulsion or paper retraction.

Misconduct Scientific Definition CS Benchmark / Research Violation Example
Fabrication Making up data, fake benchmark execution logs, or non-existent user responses and recording them as real results. Generating artificial latency numbers in a CSV spreadsheet without running the benchmark experiment.
Falsification Manipulating research processes, code, or selectively deleting benchmark outliers to falsify findings. Deleting high tail latency log files from a graph to falsely claim an algorithm outperforms Redis.
Plagiarism Appropriating another person's ideas, algorithm logic, source code, or text without appropriate citation. Copying complex AST parsing code or paper text without attribution, including self-plagiarism without disclosure.
3. Computing-Specific Ethical Dilemmas

1. Data Privacy & Web Scraping

Scraping user profiles or private repos without explicit consent violates GDPR/POPIA regulations and site terms of service, even if data is publicly accessible.

2. Algorithmic Bias & ML Fairness

Models trained on skewed historical datasets can perpetuate societal discrimination (e.g., facial recognition systems with high error rates for minority demographics).

3. Dual-Use & Coordinated Disclosure

Cybersecurity exploits or zero-day malware analyses must follow Coordinated Vulnerability Disclosure (CVD), notifying vendors privately before public release to prevent malicious exploitation.

💡 Distinction Rule: Ethics Committee Review

Any CS research project involving human subjects (usability testing, developer surveys, user telemetry monitoring) MUST receive clearance from an Institutional Ethics Review Committee prior to data collection.