Towards a Model of Understanding Social Search

Brynn M. Evans, Ed Huai-hsin Chi

TL;DR

This paper challenges the traditional view of web search as a solitary activity, proposing instead a canonical social model that accounts for human interaction before, during, and after the search act. By analyzing 150 real-world search incidents, the authors demonstrate that social interactions—ranging from brainstorming keywords with a colleague to sharing results with a client—are integral to how we navigate the information landscape.

What Problem Does This Paper Address?

For decades, search engine research has focused on the "lone wolf" model: a single user interacting with a search box to satisfy an information need. While library science has long acknowledged that people consult friends and colleagues, technical research into web search has largely overlooked these social dimensions.

The authors argue that by ignoring the social context, we miss opportunities to design better search tools. They seek to bridge this gap by integrating existing theories of sensemaking (how we make sense of new information) and information seeking into a unified model that explicitly includes social touchpoints.

Methods

The researchers conducted a survey of 150 users via Amazon’s Mechanical Turk, employing the critical-incident technique. Instead of asking general questions about search habits, they prompted users to describe their most recent search act in detail. This approach minimizes recall bias and provides a snapshot of actual behavior.

The survey captured:

  • Context: What prompted the search?
  • Pre-search: Did the user talk to anyone before starting?
  • Process: What steps were taken?
  • Post-search: What happened after the information was found? Was it shared?

The participants represented a diverse range of professions, including education, finance, and healthcare, with varying levels of job expertise.

Table 1. The most frequently occurring professions and job roles reported by users in our sample.

Datasets

The primary dataset consists of the qualitative and quantitative responses from the 150 survey participants. The authors also categorized these search acts using established taxonomies (e.g., Broder’s taxonomy of web search).

Table 2. Information reported about duration of the search act and level of job experience.

Key Insights and Ideas

  • Search is a Social Lifecycle: Social interaction isn't just an "add-on" at the end; it happens throughout the process. The authors present this in a comprehensive model (Figure 1).
  • Informational Search is the Social Hub: The study categorized searches into three types: transactional (doing something), navigational (going somewhere), and informational (learning something). While transactional and navigational searches are often solo, 59.3% of searches were informational, and these were significantly more likely to involve social interaction.
  • Requirement Refinement: Before even touching a keyboard, 42% of users interacted with others to "influence the information need," gather keywords, or establish guidelines for their search.
  • The "Evidence File": During complex informational searches, users often engage in a "foraging and sensemaking" loop. Social interactions here serve as a feedback mechanism—for example, a programmer brainstorming with a colleague to decide if a specific API (the "evidence") meets their needs.
  • Distribution as the End Product: Search rarely ends with just finding a link. 58.7% of users shared their results with others, turning the search product into a social currency.
Figure 1. Canonical social model of user activities before, during, and after a search act.

Key Results

  • Pre-Search: 31.3% of searches were prompted by an external request (e.g., from a boss or client).
  • During Search: In informational tasks, 39.3% of users had social experiences specifically to improve their search schemas or assess others' opinions.
  • Post-Search: 47.3% of users organized their findings (bookmarking, printing, etc.), and 58.7% distributed them. Interestingly, 47.5% of self-motivated searchers shared their results, suggesting we share not just out of obligation, but because we think others will find the information valuable.
  • Efficiency: Most searches (44.7%) were brief, lasting less than 5 minutes, yet still managed to incorporate social elements.

Limitations and Future Work

The authors acknowledge that their sample from Mechanical Turk, while diverse, may not capture the full spectrum of global search behavior. Additionally, the study relies on self-reporting, which can be subject to subjective interpretation.

For future work, the authors suggest that search engines should move beyond the simple query box. Potential "social search" features could include:

  • Integrating instant messaging directly into search interfaces.
  • Revealing domain-specific experts within a community who can advise searchers.
  • Displaying "search trails" or successful keyword combinations from previous users to provide high-level feedback on complex topics.

Summary updated

Abstract

Search engine researchers typically depict search as the solitary activity of an individual searcher. In contrast, results from our critical-incident survey of 150 users on Amazon's Mechanical Turk service suggest that social interactions play an important role throughout the search process. Our main contribution is that we have integrated models from previous work in sensemaking and information seeking behavior to present a canonical social model of user activities before, during, and after search, suggesting where in the search process even implicitly shared information may be valuable to individual searchers.
BibTeX
@inproceedings{evans2008towards,
  author = {Brynn M. Evans and Ed Huai-hsin Chi},
  title = {{Towards a Model of Understanding Social Search}},
  booktitle = {SSM},
  year = {2008},
  url = {https://openreview.net/forum?id=pqYfOX4XBSh},
}
Figure