Technical Debt Research Articles

Refining the Scrum Paradigm: A Comprehensive Study of Software Development Practices (2020-2023)

This article presents a complex vision on software production practices aimed at enhancing the Scrum methodology within software project management. The recommended best practices are closely aligned with contemporary trends in the IT sector, including the total digitalization and virtualization of production processes, the transition to fully remote software development models, the incorporation of artificial intelligence technologies, and the implementation of cost-effective models in team organization. The modifications to software production processes discussed herein are grounded in research conducted between 2020 and 2023, encompassing insights from leading Russian, European, and global IT companies. This research underscores a significant transformation in IT business practices, reflecting a shift towards new benchmarks for Scrum team efficiency. These benchmarks emphasize the data-driven formalization of production models, automation facilitated by AI tools, and noticeable cost optimization within engineering teams, which collectively align with the evolving demands of the modern IT landscape. The article details a series of management strategies designed to adapt to these shifts, including the formalization of sprint goals and processes within Scrum, management of technical debt, and the cost-efficient organization of development teams.

An exploratory study on automatic identification of assumptions in the development of deep learning frameworks

ContextStakeholders constantly make assumptions in the development of deep learning (DL) frameworks. These assumptions are related to various types of software artifacts (e.g., requirements, design decisions, and technical debt) and can turn out to be invalid, leading to system failures. Existing approaches and tools for assumption management usually depend on manual identification of assumptions. However, assumptions are scattered in various sources (e.g., code comments, commits, pull requests, and issues) of DL framework development, and manually identifying assumptions has high costs (e.g., time and resources). ObjectiveThe objective of the study is to evaluate different classification models for the purpose of identification with respect to assumptions from the point of view of developers and users in the context of DL framework projects (i.e., issues, pull requests, and commits) on GitHub. MethodFirst, we constructed a new and largest dataset (i.e., the AssuEval dataset) of assumptions collected from the TensorFlow and Keras repositories on GitHub. Then we explored the performance of seven non-transformers based models (e.g., Support Vector Machine, Classification and Regression Trees), the ALBERT model, and three decoder-only models (i.e., ChatGPT, Claude, and Gemini) for identifying assumptions on the AssuEval dataset. ResultsThe study results show that ALBERT achieves the best performance (f1-score: 0.9584) for identifying assumptions on the AssuEval dataset, which is much better than the other models (the 2nd best f1-score is 0.8858, achieved by the Claude 3.5 Sonnet model). Though ChatGPT, Claude, and Gemini are popular models, we do not recommend using them to identify assumptions in DL framework development because of their low performance. Fine-tuning ChatGPT, Claude, Gemini, or other language models (e.g., Llama3, Falcon, and BLOOM) specifically for assumptions might improve their performance for assumption identification. ConclusionsThis study provides researchers with the largest dataset of assumptions for further research (e.g., assumption classification, evaluation, and reasoning) and helps researchers and practitioners better understand assumptions and how to manage them in their projects (e.g., selection of classification models for identifying assumptions).

Applying SOLID principles for the refactoring of legacy code: An experience report

The presence of technical debt in legacy systems is an inevitable consequence of years of development. Metrics play a significant role in informing the prioritisation process of maintenance activities to reduce this debt. However, it is important to note that not all metrics are equally important or readily available in real industrial settings.This paper summarises an experience report of refactoring activities performed at a Dutch partnering company, aimed at identifying, prioritising and repaying parts of the architectural technical debt accumulated in two decades of development.Given the size of the refactoring task, a data-driven prioritisation was necessary, and based on the impact that the maintenance activity would have on the base system. However, the metrics available from the monitoring of the system formed a limited set, and were not always focused on architectural aspects. Even so, the impact analysis was performed and resulted in the selection of a subset of components that needed urgent maintenance.The refactoring of the identified components was centred around the well-known SOLID design principles, particularly the Dependency Inversion (DI) principle. Additionally, a set of recurring actions was established into ‘refactoring patterns’ and systematically applied to more than 5,000 source, header and custom domain language files.This work, albeit limited to the period where the activity was planned for, was well received by the industrial collaborator. The patterns have proven very valuable in the process of maintaining such a large project scope. The data-driven approach and the identified patterns have helped the team navigate this large space and consistently refactor similar architectural issues that fall under the same category.

Comparative Evaluation of Software Composition Analysis Tools in Context of Technical Debt Reduction

Comparative Evaluation of Software Composition Analysis Tools in Context of Technical Debt Reduction

Reducing Technical Debt through Strategic Leadership in Retail Technology Systems

In the fast-paced and highly competitive landscape of retail, the management of technical debt has become a critical challenge for organizations striving to maintain robust and scalable technology systems. Technical debt, the accumulated cost of prioritizing quick fixes or suboptimal solutions over more sustainable and efficient options, can significantly hinder an organization's ability to innovate, adapt, and grow. This paper explores the role of strategic leadership in effectively reducing technical debt within retail technology systems. By examining key leadership strategies, such as fostering a culture of continuous improvement, investing in long-term technology planning, and emphasizing cross-functional collaboration, the paper highlights how retail organizations can systematically address and mitigate technical debt. Strategic leadership in the retail sector involves a comprehensive understanding of both business and technological imperatives. Leaders must recognize the implications of technical debt not only in terms of immediate financial costs but also in the broader context of business agility, customer experience, and competitive advantage. The paper emphasizes the need for leaders to adopt a proactive approach, prioritizing technical debt reduction as a strategic objective that aligns with the organization's overall goals. One of the central arguments of the paper is that effective technical debt management requires a shift from reactive to proactive leadership. This involves anticipating potential issues, encouraging innovation, and making informed decisions that balance short-term gains with long-term sustainability. The paper discusses various leadership practices that contribute to this shift, including the implementation of robust governance frameworks, the adoption of agile methodologies, and the promotion of a learning-oriented organizational culture.

Reducing Technical Debt through Strategic Leadership in Retail Technology Systems

In the fast-paced and highly competitive landscape of retail, the management of technical debt has become a critical challenge for organizations striving to maintain robust and scalable technology systems. Technical debt, the accumulated cost of prioritizing quick fixes or suboptimal solutions over more sustainable and efficient options, can significantly hinder an organization's ability to innovate, adapt, and grow. This paper explores the role of strategic leadership in effectively reducing technical debt within retail technology systems. By examining key leadership strategies, such as fostering a culture of continuous improvement, investing in long-term technology planning, and emphasizing cross-functional collaboration, the paper highlights how retail organizations can systematically address and mitigate technical debt.Strategic leadership in the retail sector involves a comprehensive understanding of both business and technological imperatives. Leaders must recognize the implications of technical debt not only in terms of immediate financial costs but also in the broader context of business agility, customer experience, and competitive advantage. The paper emphasizes the need for leaders to adopt a proactive approach, prioritizing technical debt reduction as a strategic objective that aligns with the organization's overall goals.One of the central arguments of the paper is that effective technical debt management requires a shift from reactive to proactive leadership.

Hearing the Voice of Software Practitioners on Technical Debt Monitoring: Understanding Monitoring Practices and the Practices' Avoidance Reasons

Hearing the Voice of Software Practitioners on Technical Debt Monitoring: Understanding Monitoring Practices and the Practices' Avoidance Reasons

Technical debt, a double-edged sword that can damage moral responsibility

Technical debt, a double-edged sword that can damage moral responsibility

Technical Debt Monitoring Decision Making with Skin in the Game

Technical Debt Management (TDM) can suffer from unpredictability, communication gaps and the inaccessibility of relevant information, which hamper the effectiveness of its decision making. These issues can stem from division among decision-makers which takes root in unfair consequences of decisions among different decision-makers. One mitigation route is Skin in the Game thinking, which enforces transparency, fairness and shared responsibility during collective decision-making under uncertainty. This article illustrates characteristics which require Skin in the Game thinking in Technical Debt (TD) identification, measurement, prioritisation and monitoring. We point out crucial problems in TD monitoring rooted in asymmetric information and asymmetric payoff between different factions of decision-makers. A systematic TD monitoring method is presented to mitigate the said problems. The method leverages Replicator Dynamics and Behavioural Learning. The method supports decision-makers with automated TD monitoring decisions; it informs decision-makers when human interventions are required. Two publicly available industrial projects with a non-trivial number of TD and timestamps are utilised to evaluate the application of our method. Mann–Whitney U hypothesis tests are conducted on samples of decisions from our method and the baseline. The statistical evidence indicates that our method can produce cost-effective and contextual TD monitoring decisions.

Monitoring of Artificial Intelligence in Hospitals.

Monitoring of artificial intelligence (AI)-based algorithms is necessary for safe implementation and will be required in upcoming regulations. This study investigates the potential for monitoring of AI in hospitals. First, by reviewing regulatory requirements and state of the art of monitoring. Then, by conducting a gap analysis of ISO42001, containing industry agnostic requirements harmonized with the EU AI Act. The analysis illustrates the need for comprehensive monitoring capable of capturing deviations in input, performance drifts and unintended interactions. However, hospitals often suffer from a technical debt, and the gap analysis provides qualitative indications on implementation challenges, including data quality, infrastructure and limitations in continuous improvement.

Navigating social debt and its link with technical debt in large-scale agile software development projects

Agile methodologies have emerged as transformative paradigms in the ever-evolving software development landscape, emphasizing iterative development, customer collaboration, and adaptability. As the scope and complexity of projects and organizations expand, applying agile principles within the context of Large-Scale Agile Development (LSAD) encounters distinctive challenges. The majority of challenges encountered in LSAD, technical and non-technical, are attributed to the accrual of social debt. However, a conspicuous gap remains in understanding and addressing social debt in LSAD. This study aims to fill this void by investigating social debt in LSAD through an in-depth industrial case study with a leading Nordic company specializing in telecommunications software and services and focusing on producing secure 5G network solutions. The study investigates the causes of LSAD’s social debt and examines its impacts on secure 5G telecom software development. By addressing these objectives, this research sheds light on a critical aspect of LSAD’s social debt, caused by 3C challenges(communication, coordination and collaboration), social confines challenges, community smells challenges, and organisational social challenges in the telecom sector that have been underrepresented in the existing literature.

Local and Global Explainability for Technical Debt Identification

Local and Global Explainability for Technical Debt Identification

A rule-based decision model to support technical debt decisions: A multiple case study of web and mobile app startups

ContextSoftware startups are immature software organizations that focus on the development of a single software product or service. This organizational context accumulates a lot of technical debt to cope with constraints such as limited resources and product-market fit uncertainty. While some research has explored technical debt in startups, there is no study that investigates how software startups should make technical debt decisions throughout the startup evolution stages. ObjectiveThe objective of this study is to understand how technical debt decisions are made, and how such decisions should have been made in hindsight. MethodWe conducted a multiple embedded case study to investigate technical debt decisions in five web/mobile app startups. For each case, we interviewed the case founder and developer (a total of 17 participants across cases). In addition, we collected some public documents about the five startups. The data were analyzed using qualitative data analysis techniques. ResultsWe developed a rule-based decision model that summarizes the logic to effectively make technical debt decisions throughout the startup evolution stages. In addition, we evaluated the model by conducting follow-up interviews with three participants. ConclusionThe study provides a decision model that reflects actual practice, and is designed to help software teams in startups when making technical debt decisions throughout the startup evolution stages.

Reliable proactive adaptation via prediction fusion and extended stochastic model predictive control

Proactive self-adaptation has emerged as a vital approach in recent years, aiming to preemptively address potential goal violations or performance degradation, thus improving the system’s reliability. However, this approach encounters specific challenges in prediction and decision-making, including issues such as erroneous predictions and adaptation latency. Addressing these issues, our study presents an innovative framework that leverages evidence theory to improve prediction accuracy and employs stochastic model predictive control (SMPC) for devising reliable adaptation strategies. We further refine the decision-making process by incorporating a latency-aware system model and a novel utility model inspired by the technical debt metaphor into the SMPC. Our framework’s effectiveness is validated through experiments conducted on a cyber–physical system exemplar DARTSim, demonstrating notable improvements in prediction accuracy and system reliability within dynamic environments.

Optimizing regression testing with AHP-TOPSIS metric system for effective technical debt evaluation

Regression testing is essential to ensure that the actual software product confirms the expected requirements following modification. However, it can be costly and time-consuming. To address this issue, various approaches have been proposed for selecting test cases that provide adequate coverage of the modified software. Nonetheless, problems related to omitting and/or rerunning unnecessary test cases continue to pose challenges, particularly with regard to technical debt (TD) resulting from code coverage shortcomings and/or overtesting. In the case of testing-related shortcomings, incurring TD may result in cost and time savings in the short run, but it can lead to future maintenance and testing expenses. Most prior studies have treated test case selection as a single-objective or two-objective optimization problem. This study introduces a multi-objective decision-making approach to quantify and evaluate TD in regression testing. The proposed approach combines the analytic-hierarchy-process (AHP) method and the technique of order preference by similarity to an ideal solution (TOPSIS) to select the most ideal test cases in terms of objective values defined by the test cost, code coverage, and test risk. This approach effectively manages the software regression testing problems. The AHP method was used to eliminate subjective bias when optimizing objective weights, while the TOPSIS method was employed to evaluate and select test-case alternatives based on TD. The effectiveness of this approach was compared to that of a specific multi-objective optimization method and a standard coverage methodology. Unlike other approaches, our proposed approach always accepts solutions based on balanced decisions by considering modifications and using risk analysis and testing costs against potential technical debt. The results demonstrate that our proposed approach reduces both TD and regression testing efforts.

Modelling the quantification of requirements technical debt

Modelling the quantification of requirements technical debt

Technical debt in AI-enabled systems: On the prevalence, severity, impact, and management strategies for code and architecture

Context:Artificial Intelligence (AI) is pervasive in several application domains and promises to be even more diffused in the next decades. Developing high-quality AI-enabled systems — software systems embedding one or multiple AI components, algorithms, and models — could introduce critical challenges for mitigating specific risks related to the systems’ quality. Such development alone is insufficient to fully address socio-technical consequences and the need for rapid adaptation to evolutionary changes. Recent work proposed the concept of AI technical debt, a potential liability concerned with developing AI-enabled systems whose impact can affect the overall systems’ quality. While the problem of AI technical debt is rapidly gaining the attention of the software engineering research community, scientific knowledge that contributes to understanding and managing the matter is still limited. Objective:In this paper, we leverage the expertise of practitioners to offer useful insights to the research community, aiming to enhance researchers’ awareness about the detection and mitigation of AI technical debt. Our ultimate goal is to empower practitioners by providing them with tools and methods. Additionally, our study sheds light on novel aspects that practitioners might not be fully acquainted with, contributing to a deeper understanding of the subject. Method:We develop a survey study featuring 53 AI practitioners, in which we collect information on the practical prevalence, severity, and impact of AI technical debt issues affecting the code and the architecture other than the strategies applied by practitioners to identify and mitigate them. Results:The key findings of the study reveal the multiple impacts that AI technical debt issues may have on the quality of AI-enabled systems (e.g., the high negative impact that Undeclared consumers has on security, whereas Jumbled Model Architecture can induce the code to be hard to maintain) and the little support practitioners have to deal with them, limited to apply manual effort for identification and refactoring. Conclusion:We conclude the article by distilling lessons learned and actionable insights for researchers.

A Systematic Mapping Study Exploring Quantification Approaches to Code, Design, and Architecture Technical Debt

To effectively manage Technical Debt (TD), we need reliable means to quantify it. We conducted a Systematic Mapping Study (SMS) where we identified 39 quantification approaches for Code, Design, and Architecture TD. We analyzed concepts and metrics discussed in these quantification approaches by classifying the quantification approaches based on a set of abstract TD Quantification (TDQ) concepts and their high-level themes, process/time, cost, benefit, probability, and priority, which we developed during our SMS. This helped identify gaps in the literature and to propose future research directions. Among the abstract TDQ concepts discussed in the different quantification approaches, TD item, TD remediation cost, TD interest, and Benefit of remediating TD were the most frequently discussed concepts. They were also supported by some form of measurement. However, some TDQ concepts were poorly examined, for example, the benefit of taking TD. It was evident that cost concepts were more frequently quantified among the approaches, while benefit concepts were not. Most of the approaches focused on remediating TD in retrospect rather than quantifying TD to strategically use it during software development. This raises the question of whether existing approaches reliably quantify TD and suggests the need to further explore TD quantification.

The Nature of Technical Debt in the Development of Descriptive Models for MBSE

AbstractModel‐Based Systems Engineering (MBSE) is the growing practice of systems engineering (SE) in which descriptive models replace documents as the embodiment of SE knowledge. These descriptive models capture SE information in place of documents, are developed in a similar manner to software source code, and are encoded and used in machine‐to‐machine applications as data. This paper describes how the Technical Debt concept widely used in the software domain—rework deferred to the future for expediency—needs to be modified for the domain of descriptive models. Technical Debt is often associated with agile development practices of software or descriptive models, which emphasize the rapid creation and iterative evolution of a Minimum Viable Product (MVP). Consciously applying appropriate modeling principles and practices is essential to make informed decisions during the modeling process to prevent the accumulation of excessive model Technical Debt—particularly during this iterative evolutionary process—which can require substantial rework to correct. The paper establishes a foundation for characterizing the Technical Debt implications of key model architecture and implementation decisions that are made explicitly or implicitly by modelers when developing descriptive models. To illustrate the model Technical Debt concept, several examples of modeling principles pertaining to model purpose and implementation are described along with their implications on model Technical Debt.

Modeling Principles to Moderate the Growth of Technical Debt in Descriptive Models

AbstractModel‐Based Systems Engineering (MBSE) is the next step in the evolution of systems engineering (SE) in which documents are replaced by descriptive models as the authoritative embodiment of SE knowledge. These descriptive models possess a unique combination of the features and characteristics of documents, software, and data. Conscientiously applying appropriate modeling principles and practices is essential to make informed decisions to moderate the accumulation of model technical debt. This paper briefly explains the concept of technical debt and its adaptation to the domain of descriptive models for MBSE. The paper then describes 18 foundational modeling principles that model architects should consider when making architectural and implementation decisions about their models and describes some of the key model technical debt tradeoffs that result when these principles are not followed. These principles address commonly observed challenges regarding model federation architecture, the selection and use of model layers, the modeling of the domain, and the semantics of modeling constructs. Applying disciplined model architecting practices to conscientiously manage the accrual and payoff of technical debt can make the difference between an enduring model that provides substantial value throughout the life cycle of the modeled entity and a model whose value collapses under the weight of uncontrolled technical debt.