Crowdsourcing is a hot topic. Many individuals hold massive processing power in their hands. Ear-phone is an interesting proof-of-concept project. It shows how smartphones from the general public can collectively contribute to public well-being. This particular case covers noise pollution.

The authors have done extensive research on potential solutions to overcome various difficulties. However, many problems are so new that they have not been solved on smartphone platforms before. The technical problems include context classification, and reconstruction of the signals with missing samples. After careful comparison, the authors choose a suitable algorithm by its empirical performance.

As the authors point out, they did not perform real-life experiments as they are highly expensive. Also, the paper has unfortunately left out the kind of incentive programs to encourage more participation. This is a very important aspect for any type of crowdsourcing. The authors cite many references and dismiss them as not applicable instead of employing relevant prior work. In another instance, the authors cite a survey to determine the likely contexts (where people carry their phones). However, they cherry-pick the conclusion to eliminate the phone belt as outdated. As for the phone platforms, I personally would like to see more popular ones, such as Android and iPhone. Although Android is briefly mentioned in the beginning, all of the numerical results for performance and power draw on Nokia.

In the end, one may still question whether crowdsourcing is a feasible solution based on such small-scale experiments. At least the paper contributes enough to be thought provoking.