Social Accessibility: Achieving Accessibility through Collaborative Metadata Authoring
Hironobu Takagi, Shinya Kawanaka, Masatomo Kobayashi, Takashi Itoh, Chieko Asakawa · 2008 · Proceedings of the 10th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '08) · doi:10.1145/1414471.1414507
Summary
This IBM Research Tokyo paper proposes Social Accessibility, a system that applies social computing principles to web accessibility by enabling a worldwide community of volunteers to collaboratively create external accessibility metadata for inaccessible websites. The core insight is that the traditional model — where site owners bear sole responsibility for accessibility — is too slow and often fails, leaving users with disabilities unable to access content before it becomes obsolete. Social Accessibility inverts this model by allowing any community member to fix accessibility problems on any website. The system has three participant roles: end users (screen reader users who report problems), supporters (volunteers who author metadata fixes), and site owners (who can learn from the fixes). When a user encounters an accessibility barrier, they press a shortcut key to report the problem, which captures a screen image, DOM position, and user comment. Supporters receive notifications and use a Firefox sidebar authoring tool that simulates screen reader output alongside a metadata editing interface. The metadata is designed declaratively without dependencies between entries, using XPath to target HTML elements. When any user later visits the fixed page, client-side scripts automatically retrieve and apply the metadata, transparently improving the experience.
Key findings
The pilot service was deployed internally at IBM, focusing on screen reader users. A total of 220 pages had been submitted and 410 metadata entries created for 50 domains. Supporters typically created fixes that went beyond the specific reported problem, producing general improvements applicable to similar pages. For example, one supporter who fixed an Amazon.com page created metadata for the category pages, search result pages, and shopping cart functions — addressing issues like missing alt text, inaccessible image links, heading structure, and extraneous content. The transcoding mechanism operates through client-side JavaScript and server-side processing: the browser intercepts page loads, checks the repository for metadata, and applies modifications via DOM manipulation. A key technical innovation was wildcard XPath notation that allows one piece of metadata to apply to multiple HTML elements and dynamically changing pages (e.g., online shopping sites). The system includes an incentive mechanism awarding "Supporter points" and "User points" to motivate participation, with portal pages showing rankings and activity metrics. Ten full-time metadata authors were estimated to be able to create metadata for most pages of 100 selected websites.
Relevance
Social Accessibility represents an important and prescient approach to web accessibility that anticipates many later developments. The crowdsourced fix model recognizes a fundamental reality: site owners often lack the motivation, knowledge, or resources to make their content accessible in a timely manner. By empowering the accessibility community to create fixes externally, the system provides immediate relief while also generating data that educates site owners about specific accessibility problems and solutions. For accessibility practitioners, this work raises important questions about the sustainability and quality control of community-driven accessibility efforts — challenges that remain relevant to modern accessibility overlay tools and browser extension approaches. The emphasis on making the supporter tools simple enough for non-technical volunteers to use reflects a design philosophy that accessibility work should not be limited to specialists. The system also serves as an early example of "net-wide annotation," a concept the authors predicted would grow as social computing matured.
Tags: web accessibility · social computing · collaborative authoring · metadata · web transcoding · screen reader · crowdsourcing · assistive technology