Technical Summary: Stakeholder Criteria in Technical Debt Decision-Making

Problem Statement

Technical Debt (TD) decisions are rarely driven solely by technical factors; they involve a complex interplay of heterogeneous criteria including deadlines, client expectations, architectural consequences, and organizational constraints. While prior research has identified various factors influencing TD decisions (e.g., time pressure, cost, risk, maintainability), these factors are reported under inconsistent labels and at varying levels of abstraction. This fragmentation hinders the comparison of empirical results across different TD decision contexts.

Furthermore, existing taxonomies primarily focus on classifying the types of debt (e.g., code, architecture, test debt) or mapping management practices. They do not systematically organize the criteria stakeholders use when reasoning about TD decisions, nor do they distinguish how these criteria function differently during the two asymmetric decision moments: acquisition (legitimizing a compromise) and repayment (justifying the allocation of resources to address existing debt). This gap obscures the socio-technical nature of TD decision-making and the specific roles criteria play in justifying, delaying, or authorizing action.

Methodology

The authors conducted a qualitative study to develop a practitioner-informed taxonomy.

• Participants: 11 software practitioners from the Brazilian software industry, holding diverse roles including Project Managers, Developers, Technical Leads, Requirements Analysts, and Business Analysts. Experience levels ranged from less than 2 years to over 10 years, spanning private, public, and mixed sectors.
• Data Collection: Semi-structured interviews were conducted, focusing on how practitioners reason about TD acquisition and repayment. The protocol covered roles, development contexts, concrete TD examples, criteria used in decisions, and stakeholder interactions.
• Analysis Process: The study employed a five-step taxonomy construction process:
  1. Criterion Inventory: Extracting named criteria from interview codes.
  2. Consolidation and Normalization: Grouping semantically similar criteria (e.g., merging "client satisfaction" and "user value").
  3. Family Construction: Organizing criteria into high-level families covering both acquisition and repayment contexts.
  4. Cross-Decision Stress Test: Verifying that each family applies meaningfully to both decision moments.
  5. Literature-Informed Cross-Check: Refining terminology and conceptual boundaries against prior TD research without imposing categories a priori.
• Rigor: Traceability was maintained from raw interview codes to final taxonomic families. Anonymized data, including transcripts and coding tables, are available in a public repository.

Key Contributions

The paper proposes a practitioner-informed taxonomy represented as a conceptual model of stakeholder criteria for TD decision-making. Its primary contributions are:

1. A Six-Family Taxonomy: A structured classification of stakeholder criteria that moves beyond flat lists of factors to organize criteria by their decision function.
2. A Conceptual Model: A representation of TD decision-making as a stakeholder-mediated process involving interpretation, translation, and legitimation.
3. Identification of Structural Asymmetry: An analytical artifact distinguishing the functional roles of criteria in acquisition versus repayment, specifically highlighting the difference between "permission" and "authorization" mechanisms.

Results

The Six Taxonomic Families

The study identified six families of criteria that stakeholders mobilize when making TD decisions:

1. Stakeholder-facing Value: Criteria evaluating effects on clients, users, and market expectations (e.g., client satisfaction, time-to-market, user value).
2. Delivery and Resource Pressure: Criteria related to time, cost, capacity, and planning constraints (e.g., deadline pressure, resource availability, sprint capacity).
3. Technical Integrity and Systemic Risk: Criteria evaluating internal system health (e.g., architecture, stability, maintainability, security, bug severity).
4. Decision Basis and Epistemic Style: Criteria regarding how stakeholders know or justify a decision (e.g., evidence, intuition, prior experience, recurring incidents, ROI).
5. Governance and Legitimation: Criteria determining organizational acceptability (e.g., business prioritization, consensus, negotiation, process maturity).
6. Human and Team Sustainability: Criteria related to morale, cognitive load, and team alignment (e.g., developer satisfaction, fear of breaking things, team friction).

The Conceptual Model

The model posits that TD decision-making is not a direct reaction to a debt item but a mediated process. Criteria become actionable through three mechanisms:

• Interpretation: Assigning meaning to a criterion in a specific context (e.g., interpreting "pressure" as an unavoidable deadline vs. a lack of capacity).
• Translation: Reformulating concerns to be understood across roles (e.g., translating "technical debt" into "business continuity risk").
• Legitimation: Making a criterion acceptable as a basis for action (e.g., using deadline pressure to justify a shortcut).

Structural Asymmetries: Permission vs. Authorization

A central finding is that while acquisition and repayment share the same taxonomic families, they perform different decision functions:

• Acquisition (Permission Mechanism): Criteria often act to permit shortcuts. They justify compromises under current constraints (e.g., "We must deliver now, so a shortcut is acceptable"). The focus is on preserving momentum and immediate value.
• Repayment (Authorization Mechanism): Criteria act to authorize the allocation of time and resources to existing debt. They must justify interrupting feature delivery (e.g., "The debt is causing recurring incidents, so we must fix it"). The focus is on making hidden costs visible and legitimate.

The paper notes that acquisition is often easier because it is justified by immediate delivery value, whereas repayment is harder because it requires demonstrating that delayed consequences warrant present resource allocation.

Significance and Claims

The authors position this work as an empirically grounded artifact rather than a universal predictive theory. The significance of the paper lies in:

• Shifting the Focus: Moving from classifying what debt is to understanding how stakeholders reason about it.
• Clarifying Heterogeneity: Providing a stable vocabulary to compare decision contexts by distinguishing the function of criteria (e.g., defining value vs. creating pressure) rather than just their topic.
• Explaining the Acquisition/Repayment Gap: Offering an analytical explanation for why TD is often easier to acquire than to repay, rooted in the functional asymmetry of criteria (permission vs. authorization).
• Practical Utility: Supporting more balanced TD discussions in practice by helping teams distinguish between accepting a compromise, postponing action, or legitimizing an intervention.

The paper concludes that TD management depends not only on the type of debt involved but on how stakeholders justify, delay, negotiate, or authorize action. The proposed taxonomy and model provide a framework to structure these discussions and compare decision contexts across different organizations.