Research Brief: Searching for a Google that Goes Off on a Tangent

WHAT DID THE STUDY LOOK AT?

For organizations whose lifeblood is innovation, valuable ideas are born of new information. To be useful, novel information must have a relationship to existing organizational knowledge. “It must be far enough away to qualify as novel,” the researchers write, “but it must be close enough that it can be understood and exploited.” Managers searching for these kernels of potential gold often turn to the internet and its popular search engines such as Google or Bing. But in doing so, they often run into one of three challenges: locating novel information to begin with, recognizing its value, and actually applying it when the novel information is “outside existing individual and organizational mental models.” Typical search engines churn out results that are directly related to the search query, which may be too tidy for those searching for novel information. This study looks at the advantages of using a search engine that finds results that are just over the horizon from the initial search terms.

HOW WAS THE STUDY DESIGNED?

The group designed and created a prototype of a search engine called Athens 2.0 that returns results indirectly related to the original search. Athens uses two waves of generalize-search-cluster to produce concepts connected to the original key terms in non-obvious ways. To see how well Athens performs as a “novel information discovery” tool, the researchers placed 99 business students in a hypothetical situation: they were told they worked for a company and that their boss had asked them to generate up to five ideas, derived from their search results, that were new opportunities to leverage the company’s existing expertise. One-third of the group used a search engine called Clusty, one-third used Google, and the remainder used Athens. Two independent judges with experience in the high-technology industry evaluated the resulting ideas based on how novel and how radical the participants’ ideas were. 

WHAT DID THE STUDY FIND?

• Google and Clusty each returned results that were directly related to the search. Because the relationship between the original search topic and the search results were so obvious, this approach rarely stimulated new ideas. Clusty performed no better than Google at discovering novel information. 
• Based on the perceptions of users and assessments of independent evaluators, the ideas generated from the indirect search engine Athens resulted in newer and more radical ideas than those from Google or Clusty. 
• Athens’s edge seemed to come from combining the clustering of search results with results indirectly connected to the original search by way of an intermediate concept.
• Using an indirect search engine is more effective because individuals, having found new information, try to make sense of the non-obvious connection between their search and the results; this often translates into a potentially valuable idea.  
• As for how well the search tool works in the real world, the researchers conducted a second study with a high-tech company to test drive the tool under actual conditions. Participants saw promise, commenting on the speed with which they were able to generate ideas and the quality of those ideas vis-à-vis typical brainstorming processes.

WHAT DO I NEED TO KNOW?

One implication of this study is that the search tool developed by the researchers could help employees locate game-changing insights more quickly than if they relied on serendipity. “This could make the innovation process more efficient and effective for organizations,” the researchers write. An important takeaway is that when engaging in research and development, organizations should consider using non-conventional search practices. Novel ideas are best generated when individuals look at a broad scope of data indirectly related to their search and are then forced to make sense of the results. 

Title: Individual Exploration, Sensemaking and Innovation: A Design for the Discovery of Novel Information

Authors: Tracy A. Jenkin (Queen’s School of Business), Yolande E. Chan (Queen’s School of Business), David B. Skillicorn (Queen’s School of Computing), Keith W. Rogers (Queen’s School of Business)

Published: Decision Sciences, forthcoming

— Brittany Harris