Link copied to clipboard
Working session · ICSA 2026 · Vrije Universiteit Amsterdam

Architecting Explainability
in Continuous Software Engineering

Governing Its Shadow Impacts

A 90-minute working session reframing explainable AI from a model-level feature to a resource-consuming architectural quality attribute — and building the governance vocabulary to manage it.

Thursday, 25 June 2026 | 10:30 AM–12:00 PM CEST
3 – 10 participants
Objective

Three questions the software architecture community has not yet answered.

Current practice treats explainability as a cost-neutral model property. This session argues it is a governed architectural concern — and asks the community to define what governing it would require.

Question 1 · Quality attribute status

Should explainability be formalised as a first-class architectural quality attribute, alongside performance, security, and maintainability?

Question 2 · Governance in continuous software engineering

How should the shadow impacts of explainability be governed within the fast-paced cycles of CI/CD pipelines and automated infrastructure?

Question 3 · Architectural tactics

Which specific architectural tactics can effectively mitigate structural redundancy in explanation artifacts while preserving systemic accountability?

Session deliverable
Shared vocabulary

Name the problem before solving it.

An agreed taxonomy for transparency debt, shadow impacts, and explanation lifecycle — established during the invited talks and refined in the working groups.

Governance canvas

Map patterns to impacts.

A completed 3×3 canvas mapping architectural pattern adaptations and lifecycle metrics to the shadow impact dimensions — published within seven days.

Research agenda

Assign the open questions.

Prioritised research questions across the three discussion tracks, with volunteer leads and evidence requirements recorded before the session ends.

Concepts

A vocabulary for explainability as an architectural concern.

These concepts form the shared language for the session. Each is grounded in the empirical findings described in the next section.

The operational shadow of explainability is the aggregate of resource costs — storage, compute, energy, maintenance — generated by explanation artifacts across the lifecycle of an AI-intensive system, independent of the explanatory value those artifacts provide.

XAI as quality attribute

More than a model output.

Explainability is a resource-consuming architectural property that introduces systemic tensions with sustainability, privacy, and maintainability — not a cost-neutral feature.

Explanation artifact

The residue of transparency.

Attribution maps, feature embeddings, model checkpoints, and saliency logs produced by XAI mechanisms. They accumulate passively and are rarely governed.

Transparency debt

Accumulated liability.

The growing architectural burden created when explanation artifacts are retained beyond their marginal explanatory value, accruing storage, energy, and maintenance costs.

Redundancy cost function

Marginal value against systemic burden.

A model linking the marginal explanatory value of an artifact to its operational cost — the basis for promotion, demotion, and pruning lifecycle decisions.

Out-of-model phenomenon

Impacts that models cannot see.

Shadow impacts manifest in storage tiers, AIOps toolchains, and governance layers — not in model accuracy metrics. Current observability practices do not surface them.

Explanation fragmentation

Coherence lost across services.

In microservice architectures, the rationale for a decision is distributed across independently-evolving components, making a coherent audit chain architecturally unviable.

Lifecycle governance model
Promote

Preserve high-value explanations.

Artifacts whose marginal explanatory value exceeds their operational cost are promoted to long-term storage with full provenance metadata.

Demote

Move stale metadata to cold tiers.

Artifacts that retain some value but are infrequently accessed are demoted to cheaper storage, reducing retrieval and compute overhead.

Prune

Remove redundant artifacts entirely.

Artifacts whose marginal value has fallen to zero — successive near-identical checkpoints, stale attribution maps — are pruned to reduce energy and storage footprint.

Three discussion tracks
Track I

Transparency debt.

Characterising and measuring the accumulation of explanation artifacts — redundancy cost functions, observability signals, and cross-dimension shadow impact interactions.

Track II

Architectural lifecycle.

Patterns and tactics for governing explanation artifacts — explanation contracts, lifecycle policies, surrogate fidelity trade-offs in edge–cloud deployments.

Track III

Governing AI-intensive systems.

Formalising XAI governance — CI/CD regression tests, ISO/IEC 25010 extension, ATAM adaptation, and the interaction with EU AI Act retention requirements.

Empirical observations

Two findings that motivate this session.

Both findings are external to model performance. Neither appears in existing architecture quality models or CI/CD governance frameworks.

Discovery 1 · ICSE NIER 2026Accepted

Redundancy as a systemic byproduct of explainability.

Longitudinal study · large-scale medical imaging pipelines · millions of samples · iterative augmentation experiments

Explanation artifacts — high-dimensional attribution maps, feature embeddings, and model checkpoints — accumulate passively across every iteration of the evolution cycle. These artifacts exhibit high informational redundancy without explicit architectural lifecycle evaluation. Storage, retrieval, and maintenance costs grow non-linearly relative to the system's utility. The marginal explanatory value of each successive artifact approaches zero while its operational cost remains constant.

Pattern observed

Architectural decay through passive artifact accumulation across every CI/CD iteration.

Cost structure

Non-linear growth in storage, retrieval, and maintenance costs relative to utility gain.

Governance gap

No current lifecycle policy triggers demotion or pruning of stale explanation artifacts.

Discovery 2 · UK–Canada collaboration

Shadow impacts and the out-of-model phenomenon.

Joint research programme · Queen's University Belfast & Concordia University · generalised from Discovery 1

Embedding explanation mechanisms triggers a cascade of hidden impacts across multiple architectural dimensions simultaneously. These impacts are external to the model — they manifest in storage tiers, AIOps iterations, and governance layers, not in accuracy metrics. In a continuous software engineering context, they accumulate invisibly because current practice treats explanations as passive outputs rather than governed architectural assets.

Technical

Computational overhead, data pipeline complexity, provenance tracking cost.

Sustainability

Increased energy footprint from artifact generation, retention, and retrieval at every iteration.

Privacy

Attribution artifacts retain training data signals; persistent retention elevates leakage risk.

Cognitive

Explanation overload for stakeholders; degraded decision quality from artifact saturation.

Organisational

Maintenance debt accumulating invisibly across teams and release cycles.

Regulatory

Compliance drift as explanation chains fragment across independently-evolving agents.

Architectural gap

What continuous software engineering practice is missing.

Cross-cutting finding · service-oriented and microservice architectures · CI/CD-deployed AI-intensive systems

In service-oriented and microservice architectures, AI evolution leads to explanation fragmentation — the rationale for a decision is distributed across multiple autonomous agents that evolve independently. No current continuous architecture practice includes explanation regression tests in CI/CD, lifecycle policies for explanation artifacts, or cross-boundary explanation contracts between services.

Missing in CI/CD

Explanation regression tests — no fidelity threshold gate exists in any standard pipeline.

Missing in microservices

Cross-boundary explanation contracts — no schema specifies what one service owes another in XAI provenance.

Missing in quality models

Explainability as a lifecycle-governed attribute — ISO/IEC 25010 and ATAM offer no mechanism for transparency debt.

Agenda

90 minutes. One working session. Two invited talks.

The invited talks establish shared vocabulary and frame the empirical evidence. The 45-minute working session converts that context into a governance canvas and a prioritised research agenda.

10:30 – 10:35
5 min

Opening remarks

Facilitators frame the session in one sentence: explainability accumulates costs that architecture does not currently govern. Session structure and track overview. Participants self-assign to one of three discussion tracks.

Problem statementTrack self-assignment
10:35 – 11:00
25 min

Working session — Research question refinement and governance canvas

Three parallel tracks refine the pre-seeded research questions, then converge in plenary synthesis onto the shared Governance Canvas.

Setup

Working sheets distributed. Each group nominates a scribe and a skeptic — the skeptic is required to challenge the first consensus reached.

Track work

Groups evaluate, refine, or reject the four pre-seeded research questions. May add one new question. Output: ranked list of 2–3 RQs with evidence requirements.

Track I · Debt measurementTrack II · Lifecycle patternsTrack III · Governance & formalization
Plenary synthesis

Each track reports in 4 minutes. Facilitator maps outputs onto the Governance Canvas. Final 3 minutes: dot vote on top cross-cutting tension and formalization position.

3 parallel tracksScribe + skeptic per groupShadow Cost WorksheetPattern Decision TreeGovernance Canvas
11:00 – 11:20
20 min

Invited talk 1 — Computing under Uncertainty on Edge-Cloud Platforms

Speaker - Dr. Daniel Balouek

Daniel Balouek is a Research Scientist at Inria, the French National Research Institute for Digital Science and Technology. His research focuses on distributed computing and complex applications, with a special emphasis on next-generation Utility Computing Infrastructures—encompassing Cloud, Edge, and beyond—as well as Urgent Science. He is a co-recipient of an Outstanding Paper Award in Artificial Intelligence for Social Impact at the AAAI Conference on Artificial Intelligence. He holds a PhD in Computer Science from École Normale Supérieure de Lyon, France. His career has spanned institutions in industry and academia across France, the USA, Japan, and India.

11:20 – 11:40
20 min

Invited talk 2 — Why Architecture Matters in AI-Driven Reliability and Observability

Speaker - Prof. Abdelwahab Hamou-Lhadj

Dr. Wahab Hamou-Lhadj is Professor and Chair of the Department of Electrical and Computer Engineering at Concordia University. He previously served as an Affiliate Researcher with the NASA Jet Propulsion Lab at the California Institute of Technology from 2022 to 2025 and continues to collaborate with the lab on advanced research initiatives. He leads a research group specializing in software observability, AI for IT Operations (AIOps), software tracing and logging, and model-driven engineering. He is a Senior Member of IEEE and a long-standing member of ACM.

11:40 – 12:00
20 min

Results Presentation

Present the outputs of the working session, including the refined research questions and the completed Governance Canvas. Discuss next steps for publication and collaboration.


Facilitators

Full Professor · Gina Cody School of Engineering and Computer Science · Concordia University, Montréal

Prof. Yan Liu

Specialises in architecture-level quality assurance for AI-intensive systems. ICSE NIER 2026 work formalises the redundancy cost function and the operational shadow of explainability.

Lecturer in Software Engineering · Queen's University Belfast

Dr. Zheng Li

Leads research in cloud-native software engineering and edge intelligence, focusing on architecture design and quality assurance for AI-enabled systems in distributed environments.