Source-linked AI summary
Big Questions on Software Architecture: Report of the ICSE 2026 BoF on Software Architecture
Davide Taibi, Patricia Lago, Henry Muccini
TL;DR
SA research needs enduring questions amid rapid technological change, AI integration, and increasingly complex systems. This paper reports the first phase of a two-session community initiative, using the ICSE 2026 BoF to identify and discuss big questions before planned refinement at ICSA 2026. The discussions emphasize combining AI support with human judgment, preserving architecture knowledge, and continuing work on foundational SA problems.
Problem
Rapid technological evolution and increasingly complex, AI-assisted systems have created a need to distinguish enduring SA research questions from short-term trends while preserving foundational knowledge and human responsibility.
Method
The paper reports a community-driven two-phase initiative that collects and discusses questions at an ICSE 2026 BoF and plans to refine and extend them through an ICSA 2026 working session.
Results
The discussions highlighted AI-assisted architecting with continued human judgment and responsibility, the preservation and use of architecture knowledge, and unresolved foundational SA concerns.
Takeaways & Limitations
Future SA research is framed as needing to balance AI-enabled architecting with foundational challenges that remain open in increasingly AI-driven systems.
Takeaways & Limitations
The findings reflect ICSE 2026 BoF participants and may not represent the broader SA community; the ICSA session is intended to add further perspectives.
Abstract
from arXiv · showhide
Software Architecture (SA) is more important than ever, from the need to master the pervasive impact of technology in society and planet earth, to balancing the role of humans versus technology, flexibility versus chaos, economic convenience versus digital autonomy. Despite SA importance, we feel that the SA research community needs to reflect on what are the big questions that deserve scientific research, and what are instead technology trends or hypes. In this paper, we report the results of a community initiative aiming to identify and discuss the big questions in SA research. Organised in two sessions hosted in 2026 at the ICSE and ICSA conferences, respectively, this paper reports on the results collected during the first session, and looks forward to complementing them with the results of the second one. To be continued ...
1 Introduction
Software Architecture remains central as systems and technologies evolve, but rapid change has shifted attention toward short-term trends. The initiative therefore seeks enduring “big questions” to guide future SA research and practice.
- SA supports the design, evolution, maintainability, and quality of software systems across successive technological paradigms.These paradigms include service-oriented, cloud-native, microservice, cyber-physical, AI-enabled, and autonomous systems.
- Rapid technological evolution has raised concerns about SA’s long-term foundations, research sustainability, and enduring scientific questions.
- Modern SA research faces complexity, AI-related decision making, fragmented architecture knowledge, and continuously evolving distributed systems.
- The initiative aims to identify enduring SA “big questions” that can shape research agendas, foundational principles, and industry practices.
- The ICSE 2026 BoF provided an open forum for researchers and practitioners to discuss major challenges, unresolved problems, and future SA directions.The paper reports this first phase and introduces an ICSA 2026 session intended to refine and extend its questions.
2 Study Design
The study uses a community-driven, two-phase process to identify, discuss, and refine enduring questions for Software Architecture research. It combines the ICSE 2026 BoF’s broad discussion with a planned ICSA 2026 working session.
- Study objectives: The study aims to identify research questions that can remain relevant over time and guide future SA research and practice.
- Two-phase design: The first phase collects and discusses candidate big questions during an ICSE 2026 BoF, while the second phase refines, validates, and extends them at ICSA 2026.
- Phase 1: ICSE 2026 BoF: The BoF invited researchers and practitioners to submit position papers, peer-reviewed nine submissions, and organized them into three thematic panels.The panels covered skills and knowledge, novelty and foundations, and research agenda and directions.
- Phase 1: ICSE 2026 BoF: Moderated panel discussions and audience debate surfaced diverse perspectives, recurring concerns, tensions, and gaps rather than seeking consensus.
- Phase 1: ICSE 2026 BoF: Moderators synthesized accepted papers, session notes, and audience comments into a preliminary set of community-identified big questions.
- Phase 2: ICSA 2026: The ICSA working session will use interactive focus-group discussion to validate, reformulate, and extend the questions and identify missing perspectives and long-term implications.
- Expected outcomes: The combined process is intended to produce a community-refined set of questions that supports future discussion and long-term research priorities.
3 Results
Phase 1 discussions identified recurring concerns about architects’ judgment, architecture knowledge, representations, evidence, sustainability, security, and digital sovereignty as AI-driven systems evolve. The resulting questions emphasize preserving human understanding, control, and trustworthiness while developing new research foundations.
- Architect skills and knowledge: Generative AI expands design alternatives and preliminary artefacts, but final architectural decisions require expert human validation and judgment.The discussion also highlights the unresolved timing of architect intervention and its cost–productivity tradeoff.
- Architect skills and knowledge: Architects need foundational skills to verify AI outputs, alongside AI-specific capabilities for increasingly AI-assisted and self-evolving systems.The panel also identifies cognitive debt when architectural structure and rationale are not made explicit.
- Architectural decision-making: PoCs can quantify complex architectural decisions, communicate alternatives, and provide evidence for resolving stakeholder conflicts, but their systematic use remains undefined.Research is needed to connect PoC outcomes with architecture decisions and their rationale.
- Novelty and foundations: Architecture knowledge is often undocumented or distributed across people and deployment artefacts, motivating machine-readable representations that support reasoning across software, infrastructure, deployment, resource, and data layers.The discussions connect this knowledge loss with risks to human control and understanding in large, evolving systems.
- Novelty and foundations: Future architecture models may need to support AI-assisted reasoning, certification, runtime adaptation, and integrated analysis in highly distributed and safety-critical systems.Participants called for new representations, datasets, modelling approaches, and reasoning techniques that preserve trustworthiness and control.
- Research agenda and directions: AI-supported architecture evolution raises both research and sustainability questions, including the difficulty of architecture reconstruction, energy footprint, and designing security and digital sovereignty by design.The discussion treats sovereignty as requiring principles, artefacts, and tools that address data ownership, service autonomy, compliance, and digital autonomy.
4 Discussion
The discussions identified recurring concerns about AI’s role, foundational SA problems, architecture knowledge, and governance. They also highlighted methodological limitations of the first-phase findings and the need for a second community session.
- Recurring themes: AI was viewed as supporting architecture reasoning and development, while participants emphasized preserving human judgment, control, and responsibility.This concern recurred across architecture decision making, knowledge management, self-coding systems, and certification of AI-assisted decisions.
- Recurring themes: Participants questioned whether long-standing SA foundations, including modelling, abstraction, interoperability, and representation, receive sufficient attention.These problems remain only partially understood despite decades of research.
- Recurring themes: Architecture knowledge was presented as central to human reasoning and future AI-enabled architecting in increasingly complex and distributed systems.The discussions emphasized preserving, reconstructing, representing, and communicating this knowledge.
- Recurring themes: Participants raised governance, ethics, organisational values, and trustworthiness as requirements for transparent and accountable AI-enabled systems.They questioned how such systems can remain aligned with organisational objectives, ethical principles, and regulatory constraints.
- Implications and limitations: Future SA research was framed as balancing AI-enabled innovation with unresolved foundational challenges, while the initiative’s first phase remained subject to participant, agenda, organisation, and synthesis biases.The ICSA session is intended to refine and extend the identified questions and broaden community input.
5 Conclusion
The first phase used an ICSE 2026 BoF to identify recurring concerns for future SA research through position-paper discussions and audience debate. Its preliminary synthesis emphasizes AI, human responsibility, architecture knowledge, and foundational SA problems, while the next phase will refine the questions with the broader community.
- Study design: The ICSE 2026 BoF discussed nine accepted position papers in three thematic panels, followed by an open audience discussion.The session constituted the first phase of a community initiative on big questions for future SA research.
- Findings: Participants emphasized AI’s growing role in SA alongside the need to preserve human judgment, architecture reasoning, and responsibility.They also stressed preserving, reconstructing, representing, and using architecture knowledge in distributed and AI-assisted systems.
- Scope: The identified questions are a preliminary, non-exhaustive synthesis of concerns and open problems from the ICSE BoF discussion.Their value is providing an initial basis for broader reflection on long-term SA research directions.
- Next phase: The ICSA 2026 working session will discuss, refine, challenge, and extend the reported questions toward a community-refined research agenda.The planned outcome concerns future discussions of SA foundations, methods, and priorities.