Investigating the Appropriateness of Social Network Question Asking as a Resource for Blind Users
Erin L. Brady, Yu Zhong, Meredith Ringel Morris, Jeffrey P. Bigham · 2013 · Proceedings of the 2013 Conference on Computer Supported Cooperative Work (CSCW 2013) · doi:10.1145/2441776.2441915
Summary
This paper explores whether social networking sites (SNSs) are a viable alternative to paid crowdsourcing for answering blind users' visual questions. The research combines three methods: a survey of 191 blind adults about their social networking habits and attitudes toward SNS-based Q&A, a log analysis of over 40,000 questions asked through VizWiz Social (an iPhone app with 5,000+ users that lets blind people send photo-based questions to either crowd workers or social network contacts), and a month-long field experiment with 23 blind VizWiz Social users that introduced monetary costs for anonymous crowdsourced answers to simulate real-world pricing. VizWiz Social offered five answer sources: Mechanical Turk web workers, IQ Engines (automated image recognition), email to specific friends, Facebook posts, and Twitter posts. The study was motivated by the hypothesis that friendsourcing could offer free, personalized, higher-trust answers compared to paid anonymous crowd workers, potentially making visual Q&A more sustainable and better quality.
Key findings
The survey revealed that 92% of 191 blind respondents used social networking sites — significantly higher than the 66% general population rate — with Facebook (80%) and Twitter (52%) most popular. Twitter adoption was especially notable at 3.5x the general population rate, likely due to its accessible text-based interface. However, blind users had smaller networks (median 100 Facebook friends vs. 130 general population; median 45 Twitter followers) and only 55% viewed SNS Q&A as effective. In the VizWiz Social log analysis, only 15% of users ever tried a social source, and of 3,116 questions in one month, just 5% (156) were sent to social networks. Social sources had poor response rates — only 3 of the questions sent to Facebook and Twitter received answers, with a median response time of 2 hours 55 minutes, compared to 100% response rate and 98-second median for web workers. In the field experiment, even when crowdsourced answers cost money (/bin/zsh.05-/bin/zsh.25 per question), all 12 post-study questionnaire respondents still preferred crowdsourced over friendsourced answers. The primary barrier was perceived social cost: users worried about appearing dependent, burdening friends, cluttering others' feeds, and being unable to reciprocate. As one participant stated, "Web workers are completely anonymous, and there is sometimes no reason to think they are actually assisting with a disability related question."
Relevance
This study provides critical evidence that the seemingly obvious solution of leveraging social networks for accessibility assistance does not work in practice due to deeply felt social and psychological barriers. The finding that blind users strongly prefer anonymous crowd workers over friends — even when friends are free and crowd workers cost money — challenges assumptions about the value of personalized, trusted answer sources. For accessibility practitioners and designers of visual assistance services, this research highlights several important considerations: blind users' desire for independence and anonymity often outweighs the benefits of personalized help; response time matters enormously for real-time accessibility needs (2+ hour social network delays are impractical for identifying a product at the store); and smaller network sizes among blind users reduce the practical effectiveness of friendsourcing. The implications extend beyond visual Q&A to any accessibility service considering social network integration — the social costs of publicly requesting disability-related assistance are a significant design constraint that must be addressed through features like private messaging, urgency signaling, or reciprocity mechanisms.
Tags: social media accessibility · blind users · crowdsourcing · friendsourcing · visual question answering · social costs