Authoring accessible media content on social networks
Letícia Seixas Pereira, José Coelho, André Rodrigues, João Guerreiro, Tiago Guerreiro, Carlos Duarte · 2022 · Proceedings of the 24th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '22) · doi:10.1145/3517428.3544882
Summary
This paper investigates why user-generated visual content on social media remains overwhelmingly inaccessible to blind and visually impaired users, despite platforms offering tools to add alternative descriptions. The researchers conducted a two-phase study: an online questionnaire completed by 258 social network users (25% reporting a disability, including 34 blind and 12 low-vision participants), followed by semi-structured interviews with 20 participants (7 blind, 13 sighted). The questionnaire, available in English, French, Portuguese, and Spanish, gathered data on social media usage patterns, content posting and sharing behaviors, and awareness of accessibility features. Interviewees were also asked to practice accessible content authoring on their own social networks for two weeks before the interview, posting at least three media items with alternative descriptions. The research addressed three questions: what motivates users to create accessible content, what barriers they encounter, and what requirements exist for better support. The study examined practices across Facebook, Twitter, Instagram, and WhatsApp, finding that each platform takes a different approach — Twitter provides a dedicated alt text input field, Facebook and Instagram rely on machine-generated descriptions with optional user editing, and WhatsApp had no built-in support at the time. Visually impaired participants were more likely to use desktop devices for posting (47% vs. 25% for sighted users), and far less likely to post image content (2% vs. 5%). Blind participants showed much lower engagement with image-centric platforms like Instagram, with GIFs and memes being particularly unpopular among visually impaired users who reported never sharing these formats.
Key findings
The study revealed a stark awareness gap: 74% of all participants believed that other people don't provide alternative descriptions simply because they don't know it's possible. Among sighted participants not providing alt text, 22% didn't know it was possible and 15% didn't know where to write one. Even among those who did provide descriptions, only 25% used the platform's built-in accessibility feature — 23% instead embedded descriptions in the post text itself. The majority of sighted participants (56%) had no awareness of accessibility practices adopted by people around them. Platform discoverability was a major problem: accessibility features were described as "hidden," hard to find, and inconsistent across mobile and desktop interfaces. Facebook's alt text editing feature was particularly difficult to locate, with one blind participant unaware it existed despite being a frequent user. Sighted users who did engage were primarily motivated by personal connections to people with disabilities or by believing it was "the right thing to do," but many felt it was an additional burden that disrupted their spontaneous posting routine. Blind participants reported an exhausting advocacy burden — constantly having to remind contacts to post accessibly, feeling like "the activist" in their social circles. Machine-generated descriptions were criticized for lacking contextual information and the author's intent, with both groups emphasizing that human-authored descriptions are essential for conveying meaning beyond what AI can detect.
Relevance
This research provides crucial evidence for anyone working on social media accessibility or content authoring tools. The findings demonstrate that simply adding an alt text feature is insufficient — platforms must integrate accessibility into the core authoring workflow, make features discoverable, provide guidance on writing good descriptions, and standardize approaches across platforms. The paper highlights a particularly important tension: the burden of advocating for accessible content falls disproportionately on disabled users who are already excluded by inaccessible content. For practitioners building content management systems or publishing tools, the key takeaway is that accessibility features must be visible, integrated into the default workflow (not hidden in advanced settings), and supported with education about why and how to write effective descriptions. The researchers recommend hybrid approaches combining AI-generated descriptions with user editing, standardized accessibility interfaces across platforms, and proactive education strategies. The study also surfaces an uncomfortable finding about stigma — some sighted users questioned whether blind people even want visual content, revealing assumptions that reinforce exclusion from image-centric platforms.
Tags: social media accessibility · alt text · user-generated content · image descriptions · digital inclusion · content authoring
Standards referenced: WCAG · ATAG