No One Left Behind: Cross-Level Analysis for Sustainable Software Engineering

I. INTRODUCTION

• Software sustainability is multidimensional, extending beyond energy optimization to technical durability, social equity, and economic viability across the lifecycle.The cited account also notes that a multidimensional view is largely missing in empirical software-engineering research.Source: §I · p. 1
• Research on energy efficiency, software quality, and developer wellbeing often develops in separate communities with distinct assumptions, metrics, and methods.This fragmentation provides limited support for understanding how sustainability problems emerge, interact, and reinforce one another over time.Source: §I · p. 1
• Organizational, process, and product decisions are interconnected, so a delivery priority may reduce testing and coincide with greater complexity, energy use, and developer frustration.The passage presents this sequence as an example of cross-level consequences.Source: §I · p. 1
• The paper introduces sustainability antipatterns to capture systemic interactions across socio-technical levels and sustainability dimensions.It also outlines research directions for cataloging, detecting, measuring impacts, and mitigating these antipatterns.Source: §I · p. 1

II. THE NEED FOR SUSTAINABILITY ANTI-PATTERNS

• Software-engineering sustainability research has contributed since 2010, yet remains fragmented and does not fully capture interconnected, recurring sustainability problems.The section identifies two major limitations as motivation for its vision.Source: §II · pp. 1–2
• Figure 1 presents an overview of a socio-technical sustainability lens spanning organizational, process, and product levels and sustainability dimensions.The caption describes sustainability antipatterns as capturing interactions across these levels and dimensions.Source: Figure 1 · p. 2
• The identified limitations motivate the paper's vision for sustainability antipatterns.Source: §II · p. 2

A. Patterns and Anti-patterns

• Patterns are reusable solutions to recurring problems, whereas antipatterns are repeated solutions that appear beneficial but ultimately produce negative consequences.The passage attributes the pattern concept to Gamma et al. and the anti-pattern concept to.Source: A. Patterns and Anti-patterns · p. 2
• Code smells, process antipatterns, and community smells have been studied at product, process, and organizational levels, typically in isolation and rarely through a sustainability perspective.The cited examples respectively concern design problems, project-management effectiveness, and dysfunctional social structures.Source: A. Patterns and Anti-patterns · p. 2
• Sustainability smells, energy smells, and sustainability design patterns link code and energy consumption but still frame the issue as technical optimization.The passage contrasts this focus with sustainability problems that can co-occur and amplify one another.Source: A. Patterns and Anti-patterns · p. 2
• Technical debt captures accumulated product-level implementation shortcuts but omits organizational drivers, process practices, and downstream sustainability consequences.The passage names developer burnout and wasted energy as examples of downstream consequences.Source: A. Patterns and Anti-patterns · p. 2
• Addressing technical debt alone may be incomplete when the cross-level socio-technical conditions that produced it remain unchanged.Source: A. Patterns and Anti-patterns · p. 2

B. Systemic Nature of Sustainability

• Systems thinking focuses on how system components interact and influence one another over time, matching the complex, multi-level nature of sustainability problems.Such problems may not be fully explained by one line of code, development practice, or organizational policy in isolation.Source: B. Systemic Nature of Sustainability · p. 2
• Other researchers have acknowledged sustainability's systemic nature.Source: B. Systemic Nature of Sustainability · p. 3
• Empirical software-engineering research largely optimizes isolated components, while sustainability is described as an emergent property of an entire socio-technical system.Prior proposals include leverage points, requirements engineering as a systemic bridge, and a sustainability capability maturity framework.Source: B. Systemic Nature of Sustainability · p. 3
• Existing systems-thinking efforts remain largely theoretical and treat sustainability as static levels rather than dynamic behavior.The section concludes that a practical concept is still lacking for identifying, cataloging, analyzing, and mitigating systemic problems in real-world projects.Source: B. Systemic Nature of Sustainability · p. 3

• A sustainability anti-pattern links interacting organizational priorities, development practices, and technical decisions that offer short-term local benefits but generate long-term unsustainable outcomes across dimensions.Source: §III · p. 3
• The concept extends traditional anti-patterns beyond isolated artifacts to multi-level patterns, cross-dimensional consequences, and long-term systemic failure from seemingly rational short-term decisions.Source: §III · p. 3
• Figure 2 illustrates sustainability problems propagating across socio-technical levels and amplifying.Source: Figure 2 · p. 3
• The vision shifts attention from isolated artifacts or practices toward system-level effects of local decisions and the evolution of interrelated patterns over time.Operationalizing the patterns requires reasoning about complex dependencies rather than simple observations.Source: §III · p. 3
• The delivery-focused example traces organizational priorities to process shortcuts, technical debt, harder maintenance, possible energy waste, developer burnout, and reinforcing short-term priorities.The authors explicitly present this as a conceptual instantiation, not an empirical finding; specific causal paths and contextual triggers may vary.Source: §III · pp. 3–4

Understanding Sustainability Anti-Patterns

• Researchers should synthesize siloed findings and investigate real-world contexts to identify how sustainability concerns connect, including dynamics from AI-assisted and agentic development.Suggested approaches include literature synthesis, systems-thinking methods, practitioner interviews, and repository mining of process artifacts and organizational signals.Source: §IV · p. 4

Formalization

• A shared template should capture each anti-pattern’s name, socio-technical context, observable symptoms, impacted sustainability dimensions, and potential mitigations.Source: §IV · p. 4
• Anti-patterns should recur across projects and contexts and affect more than one sustainability dimension; operationalizing these proposed thresholds remains a research challenge.Strictly one-dimensional issues may be better framed as technical debt or energy smells.Source: §IV · p. 4

Detection and Tooling

• Detection should combine repository signals with practitioner inquiries, while future tools support continuous monitoring and flag threshold violations.Some indicators, such as organizational pressure and implicit team norms, exist only in practitioners’ knowledge.Source: §IV · p. 4

Intervention

• Mitigation may require coordinated changes to incentives, development practices, and architecture, with strategies evaluated for cross-dimensional trade-offs.Longitudinal studies, quasi-experimental field studies, and simulations may help assess interventions that are too risky or slow to test in organizations.Source: §IV · p. 4

Sustainability Impact

• Impact measurement must account for different timescales, difficulty attributing effects to one anti-pattern, and unequal distribution of harm across stakeholders.The authors propose longitudinal comparisons of similar projects to estimate impacts and what might have happened without an anti-pattern.Source: §IV · pp. 4–5

• Sustainability anti-patterns are recurring socio-technical patterns that systematically produce long-term unsustainable outcomes across sustainability dimensions.Source: §V · p. 5
• The paper argues that understanding how sustainability problems interact and reinforce one another over time is important alongside optimizing isolated concerns such as energy and maintainability.Source: §V · p. 5
• The proposed roadmap covers identifying, formalizing, detecting, intervening on, and measuring the impact of sustainability anti-patterns.The conclusion highlights AI-assisted development and agentic systems as contexts where decision-making speed and automation may accelerate unsustainable outcomes and make them harder to trace.Source: §V · p. 5