When building complex systems—whether they are software applications, physical buildings, or artificial intelligence models—designers face a fundamental choice: do they consider the diverse needs of end-users at the very beginning, or do they build a core system first and patch it later?
Historically, accessibility and user experience (UX) research have been treated as late-stage evaluations. We build the software, construct the museum, or train the model, and then run a user study or add an assistive layer right before release. This "afterthought" paradigm creates a fundamental engineering problem: architectural lock-in. By the time a system is evaluated, its foundational data structures, physical layouts, or algorithmic frameworks have crystallized. Retrofitting accessibility at this stage forces developers to build fragmented, brittle workarounds—often called "island solutions"—resulting in massive technical debt and suboptimal user experiences.
Proactive architectural design shifts accessibility and user feedback from a post-production checklist to a foundational constraint. By integrating universal design principles into the initial infrastructure, systems become inherently inclusive, highly efficient, and significantly cheaper to maintain.
The Mechanics of Architectural Lock-In
To understand why proactive design works, we must first understand why reactive design fails. In the software development lifecycle, the cost of modifying a system scales exponentially over time.
[Conceptualization] ---> [Core Architecture] ---> [Post-Production / Release]
| | |
High Freedom Lock-In Imposed High Cost to Retrofit
Low Cost to Pivot Brittle Workarounds Massive Technical Debt
When accessibility is treated as an add-on, developers are forced to map inclusive features onto an underlying architecture that was never designed to support them. Bi et al. observe that this approach frequently leads to severe integration challenges and costly refactoring post-release. The same pattern emerges in assistive technology itself; Vaidyanathan et al. show that treating user experience improvements as secondary additions rather than core design requirements prevents the creation of seamless, cohesive tools for daily autonomy.
This lock-in effect is not unique to software; it is a structural reality across multiple domains:
- Physical Spaces: In civic architecture, Yoon et al. demonstrated that incorporating universal design during the initial blueprint phase yields significantly higher accessibility ratings than attempting to renovate older, inaccessible structures.
- Virtual and Interactive Environments: In Collaborative Virtual Environments (CVEs), García et al. found that delaying the integration of user awareness and feedback mechanisms forces expensive, late-stage system redesigns. Similarly, in game design, Shmatko et al. argue that treating UX research as a late-stage evaluation rather than a foundational tool creates a permanent misalignment between a designer's intent and the player's cognitive experience.
- Computer Vision and Graphics: Bylinskii et al. highlight that conducting user studies late in the research cycle completely castrates their utility; they can no longer inform early design choices or proactively identify user needs.
Infrastructure over Island Solutions
The core insight of proactive design is that accessibility should be a native property of the system's infrastructure, not a standalone application or outsourced technicality.
When accessibility is outsourced or delayed, developers create "island solutions"—specialized, isolated tools designed for specific disabilities that sit outside the main infrastructure. As Weber et al. argue, these island solutions are economically and socially inefficient. Instead, the authors advocate for infrastructure-level middleware that integrates accessibility features directly into operating systems and web frameworks. When the underlying platform natively understands diverse sensory and cognitive inputs, every application built on top of it inherits accessibility for free.
In digital environments like e-learning, this infrastructure-level thinking prevents repetitive, manual adjustments. Sutherland et al. propose a unified methodology that evaluates content, platforms, and interaction simultaneously. By baking standards into the initial creation of educational materials, institutions eliminate the long-term technical debt of creating individual, reactive accommodations for every new student.
Aligning Assumptions Through Early Participatory Design
Proactive design is not just about technical frameworks; it is about changing when and how human feedback enters the loop. When developers build complex infrastructure in a vacuum, they rely on unverified assumptions about user behavior.
For example, Jensen et al. studied the development of smart energy systems and found a massive disconnect: developers assumed householders would optimize for economic efficiency, whereas users actually engaged with the technology based on novelty and playfulness. Because human-computer interaction principles were integrated late in the engineering cycle, the system became rigidly technology-centered, making it impossible to pivot based on real-world household dynamics.
To prevent this, proactive design leverages participatory methods during the earliest phases of conceptualization. Even in highly technical domains like artificial intelligence, non-technical stakeholders can fundamentally shape system architecture if engaged early enough. Kuo et al. utilized a "comicboarding" method to gather feedback from frontline workers and unhoused individuals on AI systems used in homeless services. The authors note that deploying such feedback mechanisms during conceptualization and prototyping—rather than after deployment—allows for substantial, structural design modifications that align the AI's behavior with human needs from the outset.
Ultimately, shifting accessibility from a reactive patch to a proactive foundation changes the organizational ownership of inclusion. As Tuominen et al. observe in the context of live events, moving the responsibility of access away from external, third-party retrofitters and embedding it directly into the core planning phase ensures that inclusion is treated as a fundamental creative value rather than a secondary technical layer.
Key Foundational Work
Enhancing Collaborative Manipulation Through the Use of Feedback and Awareness in CVEs (2009) — contributed 1 concept elements to this direction.
Assessing the accessibility of E-learning (2012) — contributed 1 concept elements to this direction.