First Person: How Autoethnography Is Changing What Accessibility Research Can Say
Disabled researchers are turning their own experiences into scholarship — and the findings challenge assumptions that third-person studies cannot reach
In 2015, a researcher with Friedreich's Ataxia published a paper about his 25-year search for a viable computer input method [Felzer & Rinderknecht, 2015]. He documented how his typing rate on a standard keyboard declined from 12 words per minute in the 1990s to 3.5 WPM over 24 years as his progressive neuromuscular disease advanced, and how a custom dual-scribe interface eventually "turned back the clock" ten years, reaching a performance level comparable to his keyboard speed a decade earlier. No usability study with external participants could have captured the emotional toll of that decline, the texture of daily negotiation with a changing body, or the precise moment when a new tool restored something that felt like agency.
That paper was an early signal. In the years since, autoethnography — first-person research where scholars systematically examine their own experiences — has become a recognised method in digital accessibility research. Across more than 20 papers published between 2015 and 2025, disabled researchers have turned the lens on themselves: documenting what it is like to travel blind [Stephens et al., 2020], navigate graduate school with a disability [Jain et al., 2020], depend on assistive technologies that fail [Fussenegger & Spiel, 2022], work in a remote mixed-ability team [Mack et al., 2021], stream on Twitch as an autistic creator [Kender & Spiel, 2023], relocate internationally with a mobility impairment [Yildiz et al., 2024], or use generative AI during a period of temporary disability [Glazko, 2025].
What makes these papers significant is not simply that they offer personal perspectives. It is that they produce concepts and findings that other methods cannot reach.
What Autoethnography Reveals That Other Methods Miss
The most cited contribution from this body of work may be the distinction between "technical sufficiency and experiential adequacy," identified by Atieh Taheri in a 2025 autoethnography of three accessibility research projects she led as a researcher with Spinal Muscular Atrophy [Taheri et al., 2025]. The concept captures something that evaluation metrics alone do not: a system can meet every functional requirement and still fail to provide the embodied sense of control and agency that makes interaction meaningful.
Taheri's MouseClicker project addressed the gap between cursor control and the physical sensation of clicking. Existing assistive systems offered cursor movement but could not replicate the immediacy and sensory affirmation of a button press — something that only registers as a problem when you live without it. Her Virtual Steps project asked what it would feel like to walk, as someone who has never walked — not as rehabilitation but as imaginative exploration. She reports that reviewers questioned this work, interpreting it through a medical model that could not accommodate curiosity about bodily experience as a legitimate research goal. Her third project used LLM-driven agents to help people recognise ableist microaggressions, drawing on real encounters from her own life.
Across all three projects, Taheri identifies what she calls epistemic friction — moments where embodied knowledge conflicts with disciplinary expectations about what counts as valid research. This friction is itself a finding: it reveals the assumptions embedded in how accessibility research is evaluated [Taheri et al., 2025].
Fussenegger and Spiel's autoethnography [2022] documents the daily realities of depending on assistive technologies from the perspective of a person with a high spinal cord injury (C6 level). The account reveals dynamics that observational studies rarely capture. AT can paradoxically both enable and diminish capabilities: the wheelchair fixing device made transfers easier but gradually caused the author to lose the ability to transfer independently without it — a loss discovered only when the device broke. AT failure can be dangerous in ways that extend far beyond inconvenience: a smart thermostat malfunction left the author unable to regulate temperature in a situation where his spinal cord injury prevented his body from self-regulating. And the social dimension of AT dependency is pervasive: he relies on others not just for setup and troubleshooting but for the emotional labour of advocating for repairs, replacements, and funding.
These are not edge cases or anecdotes. They describe structural features of assistive technology use that affect millions of people daily but rarely surface in controlled evaluations.
Travel as a Research Site
Four autoethnographies document disabled people's travel experiences, and together they reveal how profoundly inaccessible travel infrastructure remains — and how creative disabled travellers must be to navigate it.
Stephens et al. [2020] documented a 28-day cruise around Australia by Kate Stephens, a 63-year-old who has been blind all her life. She found that onboard information was frequently unavailable in accessible formats — the cruise company's Access Office ignored requests for accessible bar lists and stateroom information. When the trip was cancelled due to COVID-19, critical rescheduling information was delivered only as printed letters; Kate learned about new flight arrangements by overhearing other passengers. Staff lacked awareness of how to support different disabilities, defaulting to generic "disability assistance" — leading to absurd situations where sighted guides tried to push her in a wheelchair she did not need.
Zeidieh [2024] developed a five-pillar framework for technologically interdependent travel, drawing on over a decade of first-hand experience as a visually impaired traveller. His paper's title — "Seven Stitches Later" — references a real injury from walking into a low-hanging tree branch, underscoring the physical risks that visually impaired travellers face daily and that technology papers rarely acknowledge.
Cassidy and Branham [2024] documented airport navigation across eight round-trips through six U.S. airports, finding that the first author — a legally blind traveller — was frequently caught between systems designed for full sightedness and assistive services designed for total blindness. Automated kiosks assumed visual acuity he did not have; assistance desks assumed complete blindness he did not have either. The gap — the experience of having residual vision that does not fit either category — emerged as a central finding that challenges how airports, and accessibility more broadly, categorise visual impairment.
Yildiz et al. [2024] documented a researcher's relocation from Turkey to Germany, finding that housing platform filters reduce accessibility to a binary wheelchair-focused category that fails to represent the diversity of access needs. Language barriers compounded accessibility challenges by preventing the informal social negotiations that disabled people use daily to navigate their environments.
None of these findings would emerge from a lab study or a survey. They require the researcher to be the person doing the travelling, making the decisions, encountering the failures, and reflecting on what those failures reveal about systemic design.
Graduate School and the Academy
Three autoethnographies examine what it is like to be a disabled student or researcher in academia.
Jain et al. [2020] — a three-person autoethnography by computer science PhD students at the University of Washington, one hard of hearing, one blind, one with quadriplegia — documented the emotional labour of disability disclosure in graduate school. All three experienced feelings of inferiority, guilt about "burdening" others, and pressure to prove themselves as capable, particularly as international students navigating both cultural adaptation and disability identity simultaneously. They described "access intimacy" — having higher expectations of accessibility from close colleagues — and the disappointment when those expectations are not met. The concept of "disability as a full-time job" appeared repeatedly: managing accommodations, educating faculty about needs, and advocating for accessible practices consumed time and energy that hearing, sighted, non-disabled peers could devote to research.
Kulkarni et al. [2023] documented the experience of the first blind student to complete the introductory electrical and computer engineering course at Stanford — a required course for engineering majors involving circuits, soldering, and debugging real devices. She found circuit schematics extremely difficult to retain as mental models from verbal descriptions alone, especially circuits with hundreds of components. She could not use the CAD-based circuit simulator that sighted students used. And lab work required constant sighted assistance, creating dependence that compounded social awkwardness. Her account produced ten concrete design criteria for accessible engineering tools.
Mo et al. [2025] documented navigating the final stages of a PhD in Computer Science as a student with cerebral palsy, including physical decline caused by a cervical spondylosis in 2016. He types with one finger on his iPhone's smaller keyboard rather than a standard keyboard. He uses ChatGPT to convert pseudocode into runnable code. He relies on remote desktop software to control his Windows machine from his MacBook. His account documents how commercially available mainstream technologies — not specialised AT — often provide the most effective accessibility solutions, precisely because they are updated frequently, well-documented, and do not carry the stigma of assistive devices.
The Mixed-Ability Workplace
Mack et al. [2021] documented Microsoft Research's first all-remote intern cohort during the summer of 2020 in a group autoethnography. Eleven team members — including people who were blind, Deaf, hard of hearing, had ADHD, had chronic illness, and were non-disabled — journaled their experiences over three to four months.
The findings reveal that accessibility in mixed-ability groups is not a checklist but an ongoing negotiation. Video conferencing offered benefits like persistent text chat and multiple communication modalities, but created new barriers: screen sharing reduced video tile space, making ASL interpreters invisible; captioning delays caused conversation threading problems; a hard-of-hearing member could not be alerted he was on mute because sharing his screen blocked others' video tiles.
Power dynamics shaped who bore the cost of access: senior team members could change meeting formats, while interns sometimes masked their access needs to avoid being perceived as difficult. Non-disabled team members experienced a learning curve around "access awareness" — recognising that their default practices created barriers. The paper captures these dynamics through specific incidents that a satisfaction survey would flatten into averages.
Abramovich and Patitsas [2024] used duoethnography — a collaborative method where two researchers examine a shared topic through contrasting personal experiences — to explore the labour of doing web accessibility work. A professional consultant and a computer science professor compared their experiences and found that accessibility work is often invisible labour: it happens in the margins of other roles, is rarely resourced adequately, and the people doing it are frequently isolated within their organisations. The compliance model (meeting WCAG success criteria) is necessary but insufficient; organisations that treat accessibility as a checkbox exercise often achieve technical conformance while creating fundamentally inaccessible experiences.
Chronic Illness and Fluctuating Access
Mack et al. [2022] argued that HCI accessibility research has largely overlooked people with chronic illnesses, approaching them through a medical lens focused on symptom tracking rather than recognising their broader accessibility needs. The authors — three of whom are chronically ill — present a framework built on three tenets: viewing chronically ill people as whole individuals rather than patients, designing for consequences of illness rather than symptoms, and recognising that access is interdependent rather than individual.
Their case studies reveal how access actually works in practice. One author cannot watch videos with shaky camera movement due to motion sickness, so her friend writes motion descriptions before sharing — demonstrating how access is social and relational, not purely technical. Both authors use text-to-speech tools in ways their designers did not anticipate — one uses a screen reader visually to avoid dizziness from reading text. The paper introduces the concept of "consequence-based accessibility" — designing not for a specific diagnosis but for the functional consequences that diagnosis produces, which may fluctuate hour to hour, day to day.
Glazko [2025] documented using generative AI during a period of temporary disability that simultaneously affected verbal communication, visual processing, and manual dexterity. The author found that commercially available AI tools required no fitting, specialist training, or financial investment — a stark contrast to traditional assistive technology that requires assessment, procurement, and learning curves misaligned with uncertain temporary disability timelines. The paper raises questions about what happens when temporary users develop fluency with AI accessibility tools and then lose access when they no longer qualify as disabled.
Disability Culture and Knowledge Production
Two papers go beyond documenting individual experience to examine how disabled people's collective knowledge produces different kinds of technology.
Desai et al. [2025] studied seven disabled innovators — including the creators of NVDA (the open-source screen reader used by millions), SoundPrint (a noise-level crowdsourcing app), and DeafFriendly (a Deaf consumer review platform) — and found that their technologies were "artifacts of disability culture." NVDA's open-source model created a global community of blind developers. SoundPrint built a community of people who care about noise levels. Each technology reflected the innovator's embodied expertise — what the paper calls "crip knowledge." Ed Summers' team of blind engineers brought insight about "little things" that made their data visualisation tools successful; Rory Cooper described how wheelchair users notice "very subtle differences" from daily use that non-users cannot perceive. The innovators described moving from consumers of access to producers of it.
Brody et al. [2025] studied 12 artists with disabilities and found that they viewed tools as "interdependent relations" rather than isolated instruments — wheelchairs, human assistants, and community networks were considered creative tools as fundamental as cameras or paintbrushes. Participants almost universally rejected the notion of "adapting" tools for their disabilities, instead describing modifications as tinkering and creative expression. The paper challenges the assumption that new tools are always needed, suggesting that researchers should study existing creative adaptations before designing replacements.
What Autoethnography Produces
Across these papers, autoethnography produces several kinds of knowledge that other methods struggle to generate.
Concepts grounded in embodied experience. "Technical sufficiency versus experiential adequacy" [Taheri et al., 2025]. "Consequence-based accessibility" [Mack et al., 2022]. "Disability as a full-time job" [Jain et al., 2020]. "Access intimacy" [Jain et al., 2020]. "Epistemic friction" [Taheri et al., 2025]. These are not abstract theoretical constructs — they emerge from specific, documented experiences of living with disability and interacting with technology. They give practitioners language for phenomena they may have observed but could not name.
Failure stories that matter. Controlled evaluations tend to measure success: task completion, accuracy, satisfaction scores. Autoethnographies document failure in its full complexity — the smart thermostat that nearly caused a medical emergency [Fussenegger & Spiel, 2022], the cruise company that ignored access requests [Stephens et al., 2020], the AI writing tool that appropriated a user's voice [Bennett et al., 2024], the housing platform whose binary accessibility filter rendered a researcher invisible [Yildiz et al., 2024]. These failures are not outliers; they are structural features of how technology and disability interact.
The social dimension of access. Nearly every autoethnography in this collection describes access as something that happens between people, not between a person and a tool. Friends write motion descriptions for videos [Mack et al., 2022]. Colleagues learn to unmute a hard-of-hearing team member [Mack et al., 2021]. Strangers in airports provide information that kiosks cannot [Cassidy & Branham, 2024]. Sighted assistants handle the manual labour of inaccessible government forms [Hollier et al., 2025]. This relational picture of access challenges the individual-user focus that dominates accessibility evaluation.
Questions that researchers without disabilities do not ask. What does it feel like to walk when you have never walked [Taheri et al., 2025]? What is lost when a fixing device replaces the ability to transfer independently [Fussenegger & Spiel, 2022]? How does it feel when an AI chatbot tells you to describe yourself as "thriving in fast-paced environments" when fast-paced environments are exactly what you avoid [Moore, 2025]? These questions arise from lived experience, and the answers they produce are contributions to knowledge, not merely personal reflections.
The Method's Limits
Autoethnography is not without criticism, and the papers themselves acknowledge this. Sample sizes are inherently small — often a single person. Generalisability is limited by design; the method produces depth rather than breadth. The emotional labour of exposing personal vulnerability in academic publications is significant, and several authors describe the toll of translating painful experiences into research outputs [Taheri et al., 2025; Jain et al., 2020]. Reviewers sometimes struggle with work that does not conform to conventional validity criteria, leading to the epistemic friction that Taheri documents.
There is also a practical tension. Autoethnography is most powerful when conducted by disabled researchers, but disabled researchers face systemic barriers to entering and surviving in academia — the very barriers that several of these papers document. Hofmann et al. [2020] noted that academic conferences remain largely inaccessible despite the field's stated mission. Hayton et al. [2025] described how a visually impaired researcher running workshops had to navigate the contradiction of leading accessibility research while facing access barriers in the research process itself.
But the contribution is clear. These papers have produced concepts, frameworks, and findings that have influenced how the broader accessibility research community thinks about assistive technology evaluation, chronic illness, travel accessibility, workplace dynamics, and the relationship between disability experience and design knowledge. They demonstrate that first-person research is not a lesser form of evidence but a different kind — one that reaches places where third-person observation cannot go.
This article draws on more than 30 peer-reviewed papers reviewed at A11y Paradise (a11ybob.com). All research cited was published at ACM conferences and venues between 2015 and 2025.