Software Quality & Ethical Dev: Standards, Ethics & Best Practices

TAKEAWAYS

• Software quality is not just bug-free but aligns with user needs, ethical standards, and business goals (ISO/IEC 25010 defines 8 quality characteristics).
• Ethical software development avoids bias, respects privacy (GDPR, NDPR), and prioritizes transparency (e.g., Daraz’s order-tracking system must be fair and accountable).
• Standards like CMMI help organizations systematically improve quality, reducing costs by 30–50% in maintenance (SEI report).
• Open-source ethics (e.g., Linux kernel) show how collaborative development can balance innovation with licensing fairness.
• Ethical dilemmas in AI (e.g., Pathao’s surge pricing during traffic jams) require stakeholder analysis to avoid exploitation.
• Corporate social responsibility (CSR) in IT (e.g., Ncell’s digital literacy programs) turns ethical compliance into brand trust and regulatory advantage.

1. Defining Software Quality

Software quality is the degree to which a software product meets specified requirements and user expectations while adhering to ethical and legal standards. It encompasses functional correctness, reliability, usability, and maintainability.

Key Quality Characteristics (ISO/IEC 25010)

Why it matters:

• Poor quality software costs businesses $2.8 trillion annually in lost productivity (Standish Group).
• Ethical failures (e.g., Facebook-Cambridge Analytica) erode user trust and invite legal penalties (GDPR fines up to 4% of global revenue).

Worked Example: Daraz’s Order Fulfillment System

Scenario: Daraz’s delivery app must ensure:

1. Functional Suitability: Orders are processed correctly (no misdeliveries).
2. Performance Efficiency: Real-time tracking with <2s latency.
3. Usability: Clear error messages (e.g., "Delivery delayed due to traffic").
4. Security: Encrypted payment data (PCI-DSS compliance).
5. Ethical Consideration: Fair pricing during peak demand (avoiding dynamic pricing exploitation).

Trace:

2. Software Quality Models and Standards

A. ISO/IEC 25010: Quality in Use

• Focus: How software serves users in real-world contexts.
• Metrics:
  • Effectiveness: Accuracy, completeness, adaptability.
  • Productivity: Time saved, error rates.
  • Safety: No harm to users (e.g., autonomous vehicles).
  • Satisaction: User happiness surveys.

B. Capability Maturity Model Integration (CMMI)

• Levels: From Initial (chaotic) to Optimizing (proactive).
• Process Areas:
  • Requirements Management: Clear stakeholder agreements.
  • Technical Solution: Ethical design (e.g., bias-free algorithms).
  • Process Management: Continuous improvement.

flowchart TD
    A["Initial"] -->|"Disorganized"| B["Managed"]
    B -->|"Processes documented"| C["Defined"]
    C -->|"Standardized"| D["Quantitatively Managed"]
    D -->|"Optimizing"| E["Optimizing"]
    A--->|"Ethical focus"| F["Ethical Design"]
    F -->|"Bias-free algorithms"| C

Advantages:

• Reduces defects by 50% (SEI data).
• Ethical benefit: Structured reviews catch unintended biases (e.g., loan approval algorithms favoring certain demographics).

Disadvantages:

• High setup cost (training, tooling).
• Bureaucracy risk: Overemphasis on compliance over innovation.

C. Software Quality Assurance (SQA) vs. Quality Control (QC)

3. Ethical Software Development

Ethical software development goes beyond legal compliance to moral responsibility toward users, society, and the environment.

Key Ethical Principles

1. Transparency: Users must understand how data is used (e.g., eSewa’s privacy policy).
2. Fairness: Algorithms must not discriminate (e.g., loan approvals).
3. Accountability: Developers must take responsibility for harm (e.g., WhatsApp’s end-to-end encryption).
4. Privacy: Data must be minimized and secured (GDPR, NDPR).
5. Sustainability: Reduce e-waste and energy consumption (green computing).

Worked Example: Ncell’s Ethical AI in Customer Support

Scenario: Ncell’s chatbot must:

• Avoid bias: Not suggest higher-tier plans to low-income users.
• Be transparent: Disclose when a call is recorded.
• Respect privacy: Anonymize customer data in analytics.

Trace:

sequenceDiagram
    participant User
    participant Chatbot
    participant Backend
    User->>Chatbot: "Why is my bill higher?"
    Chatbot->>Backend: Query bill data (anonymized)
    Backend-->>Chatbot: "Usage details: [redacted]"
    Chatbot->>User: "Your bill increased due to [X]. Here’s how to reduce it."
    Note right of Chatbot: **Ethical check**: No upsell pressure

Ethical Dilemma:

• Should Ncell use customer call data to train AI for ads?
  • Unethical: Violates privacy.
  • Ethical: Use aggregated, anonymized data with user consent.

4. Open-Source Software (OSS) Ethics

Open-source projects (e.g., Linux, WordPress) rely on collaborative ethics:

• Licensing: Must allow free redistribution (GPL) or commercial use (MIT).
• Contribution: Code reviews must welcome diversity (avoid "brogrammer" culture).
• Security: Bug fixes must be transparent (e.g., Heartbleed in OpenSSL).

Comparison Table: OSS Licenses

Real-World Impact:

• Linux: Powers 90% of Ncell’s servers—its ethical governance ensures no vendor lock-in.
• WordPress: 43% of websites use it; its GPL license prevents monopolies.

5. Corporate Social Responsibility (CSR) in IT

CSR turns ethical compliance into business strategy:

• Ncell’s Digital Literacy Program: Trains rural users on safe online practices.
• Daraz’s Ethical Sourcing: Ensures suppliers follow fair labor laws.
• NEPSE’s Blockchain for Transparency: Reduces insider trading risks.

Why CSR Works:

• Regulatory advantage: NDPR fines for poor data handling are NPR 50M+.
• Brand trust: 90% of Nepali consumers prefer ethical brands (Nepal Brand Equity Survey).
• Cost savings: 30% lower turnover in companies with strong CSR (Harvard Business Review).

6. Software Quality Metrics and Tools

A. Quantitative Metrics

B. Ethical Metrics

• Bias Detection: Test algorithms for demographic disparities (e.g., loan approvals).
• Privacy Impact Assessment (PIA): Audit data collection practices (e.g., eSewa’s transaction logs).
• Green Computing Score: Measure energy efficiency (e.g., Ncell’s server consolidation).

7. Case Study: Ethical Dilemma in Pathao’s Surge Pricing

Scenario: During Kathmandu’s monsoon traffic, Pathao’s algorithm dynamically raises fares by 300%. Ethical Questions:

1. Is this exploitative? → Yes, if users have no alternative.
2. Is it transparent? → No, unless clearly disclosed.
3. Is it fair? → Debatable: Helps drivers earn more but punishes stranded users.

Solution:

• Transparency: Show real-time traffic data to justify pricing.
• Fairness: Cap surges for essential services (e.g., hospitals).
• Accountability: Return excess profits to users during disasters.

Lessons:

• Algorithmic fairness must be audited independently.
• User trust is fragile—one unethical move can lead to mass defection (e.g., Uber’s surge pricing backlash).

In the Real World

1. eSewa’s Ethical Data Handling

   • Idea Used: Privacy by Design (GDPR-compliant data minimization).
   • How: eSewa never stores full transaction details—only hashed data for fraud detection.
   • Real Impact: Avoids data breaches and builds user trust for 10M+ transactions/month.

2. Khalti’s Open-Source Payments

   • Idea Used: Transparency in Licensing (MIT license).
   • How: Khalti’s payment SDK is open-source, allowing third-party audits for security.
   • Real Impact: Reduces fraud risks and attracts merchants who prioritize security.

3. NTC’s Ethical AI in Network Management

   • Idea Used: Bias-Free Traffic Routing.
   • How: NTC’s AI avoids overloading rural networks by distributing load fairly.
   • Real Impact: Prevents digital divide and ensures equal access across Nepal.

Exam Tip

This unit is highly exam-friendly because it combines concepts (ISO/IEC 25010, CMMI) with real-world applications (eSewa, Pathao). Expect:

1. Short-answer questions on:
   • Definitions of software quality characteristics.
   • Ethical dilemmas in AI/algorithms (e.g., loan approvals).
   • CSR examples in Nepali IT companies.
2. Case study analysis (e.g., "How would you improve Daraz’s order system ethically?").
3. Comparison tables (e.g., CMMI levels vs. ISO standards).

Top Scoring Tips:

• Link theory to real examples (e.g., "Like Ncell’s chatbot, ethical AI must avoid bias").
• Use the ISO/IEC 25010 model to structure answers (e.g., "For Pathao’s surge pricing, evaluate fairness and transparency").
• Discuss trade-offs (e.g., "CMMI improves quality but increases costs—how can Ncell balance this?").
• Mention regulations (GDPR, NDPR) to show legal awareness.

Common Pitfalls:

• ❌ Ignoring ethics: Just describing CMMI levels without discussing moral implications loses marks.
• ❌ Overgeneralizing: Saying "software should be ethical" without specific examples (e.g., Pathao’s surge pricing).
• ❌ Forgetting metrics: Exams love quantitative examples (e.g., "Defect density reduced by 40% after CMMI Level 3").

Final Note: This unit is not just about coding—it’s about building systems that society trusts. Always ask: "Would this pass the ethical test in a court of public opinion?"