The reading for this class was “Searching for Analogical Ideas with Crowds,” a 2014 paper from Carnegie Mellon University researchers Yu, Kittur, and Kraut. It looks at how to improve solutions to problems by leveraging the previous ideas of many others. The authors argue that finding analogical problems to the task at hand can serve as inspiration for better solutions. Further, solutions to problems with an analogous underlying structure to the task serve as better inspiration then solutions to problems with merely surface-level similarities. For example, take the problem of preventing a cup from falling over. The authors argue that problems concerned with keeping objects upright provide a far better basis than problems concerned with cups, liquids, and other similar surface features. The authors test their hypothesis by giving study participants problems to solve. Most participants were given access to a database of products over which they could search for relevant solutions. The problem was phrased in different manners for different groups of participants; some were directed to look for problems with similar structure to the task at hand, some were directed to surface feature similarities, and others were not given any particular direction at all. Finally, some participants were not even given access to the product database. The results of the experiment were rather unsurprising: those looking for similar structured problems found more analogous problems than those looking for problems with similar features or looking with no direction at all. This ordering was the same for the evaluation of the final ideas proposed by the participants.
The result of highest quality solutions coming after searching for analogously structured problems does not surprise me, but it was nice seeing research done to quantitatively show this. It makes perfect sense that looking at a problem from an abstracted vantage would allow for creative and successful solutions. One issue that I do take with the paper is in their results for percentage of analogous solutions found during the search. I personally don’t find this result to be particularly significant. In the initial phrasing of the problem, participants in the “Use Schemas” group were given a problem with an abstract description of the mechanism necessary to solve it. They were told to find problems with a similar pattern to use as inspiration. In the “Use Features” group, participants were asked to find problems with similar features to the problem at hand for inspiration. Is it any wonder that the first group found more problems that shared an underlying structure with the assigned task? The fact that the qualities of solutions were practically the same between both approaches further increases my caution in interpreting significance of the results.