WebAnywhere: Experiences with a New Delivery Model for Access Technology
Jeffrey P. Bigham, Wendy Chisholm, Richard E. Ladner · 2010 · Proceedings of the 2010 International Cross Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/1805986.1806007
Summary
This follow-up paper describes the evolution of WebAnywhere two years after its initial release, documenting how it expanded from a web-based screen reader for blind users into a broader platform for delivering access technology. Released publicly in June 2008, WebAnywhere attracted approximately 600-700 unique visitors per week, with users from over 90 countries. The system grew in three surprising directions beyond its original scope. First, language support became the most-requested feature — non-English-speaking blind users had far fewer options than English speakers since commercial screen readers and alternatives like Serotek SA-to-Go defaulted to English only. An open source contributor initially added Cantonese support, leading to a generalized localization framework that expanded to over 30 languages using the eSpeak open source TTS engine. Second, the system attracted unexpected sighted users — people with low vision, web developers using it to evaluate page accessibility (appreciating the visual highlight that tracked what was being read), special education teachers, and people with learning disabilities. Features for these users included mouse support, content highlighting, and a high-contrast magnified view of text being read. Third, speech recognition was added via the WAMI Toolkit, allowing users who found keyboard shortcuts difficult to control WebAnywhere with approximately 30 voice commands like 'next,' 'previous heading,' and 'click.'
Key findings
The paper identifies three key features of WebAnywhere's delivery model that enabled its reach: being free (removing cost barriers and allowing unrestricted distribution), requiring no installation (working on any platform with a browser), and enabling low-cost distribution and automatic updates (users always get the latest version). The server infrastructure proved surprisingly resilient — a single server could handle nearly 20 simultaneous TTS requests not served from cache, and in practice most user requests were served from the extensive cache since users spend most time reading rather than generating new requests. The main technical challenges were: JavaScript-heavy pages breaking when rewritten by the proxy (JavaScript was both the hardest to rewrite correctly and the most disruptive when it failed), and the proxy being exploited to bypass internet filters — the team noticed unusually high traffic from China and from US high school students circumventing school web filters. A key pragmatic finding was that many people who contacted the team were unaware that screen readers existed at all, highlighting that awareness and distribution are as important as technical capability. The authors proposed future research directions including visualizing blind user browsing trails for developers, leveraging browsing history to teach effective strategies to novice users, and delivering ARIA support to older browsers.
Relevance
This paper offers important lessons about assistive technology delivery that extend well beyond WebAnywhere itself. The finding that the most-requested feature was language support — not more navigation commands or better performance — reveals how accessibility solutions designed in English-speaking contexts can inadvertently exclude the global majority of disabled users. The unexpected adoption by sighted users (developers, teachers, people with learning disabilities) demonstrates that when access technology is made frictionless to try, it finds broader audiences than anticipated, supporting the principle that accessibility features benefit everyone. The challenges with proxy-based architecture and JavaScript rewriting foreshadowed the increasing difficulty of building overlay-style accessibility tools as web applications grew more complex — a tension that continues in debates about accessibility overlays today. For the accessibility field more broadly, the paper's core argument — that getting working technology to users who need it is as important as building better technology — remains underappreciated. Many people with disabilities still lack awareness of or access to assistive technology that could help them, making distribution and education as critical as innovation.
Tags: screen readers · blindness · low vision · web-based assistive technology · text-to-speech · speech recognition · open source · digital equity · internationalization · cloud computing · AT delivery
Standards referenced: WAI-ARIA