Popularity contests and crowd-sourcing

By Tamara Wilhite
An IE in IT

Crowd-sourcing is the open source form of project management and work distribution. Crowd-sourcing has grown from idea generation to software testing, database duplication checking, and social media campaigns. The ability to reach out to untold multitudes to tap their time and talent can save time and money for projects. It also raises conundrums for several types of crowd-sourcing projects.

Examples include:

  • Idea generation contests allow anyone to submit ideas to improve a product or recommend a new marketing campaign. However, when the same crowd-sourcing method is used to determine the best idea or top priorities of problems to fix, the ideas that rise to the top are not necessarily the worst product bugs or most effective campaign. The voting mass falls to whoever has the largest active network of respondents. Voting is based on numbers, and the number of votes is often biased to who screams loudest to the largest audience first. While a squeaky wheel gets the grease, but that doesn’t mean it is the best investment of maintenance money.
  • Voting schemes are also prone to bias akin to popularity contests. Whether or not voting is anonymous, voting for the most important problems to solve or best idea may be determined by the supposed repercussions. Thus votes may be swayed by what the supervisor says, not what employees think is the best idea. A top manager’s vote may be followed by the management chain, leaving good ideas in the lonely wilderness in favor of what the boss thinks. Like all the cheer leaders voting for the Captain because of the possible penalties of being discovered for voting for the competition, voting for the boss’ idea is a painless means of gaining approval or be interpreted as a measure of team solidarity, regardless of the idea’s merit.
  • Once a crowd has gathered, it tends to grow, regardless of how good the other presentations may be. For example, it is common for idea submission databases to list the ideas with the most votes at the top. Just as Google rankings see a 10% or greater drop for each row down from the first result and negligible results for those who get second page, the earlier comers to the voting booth have a disproportionate impact in this presentation model. The first ideas to get a few votes rise to the top, regardless of their actual value. They then stay near the top as subsequent visitors tend to vote on the top five or ten ideas. Like fighting to be one of the top names on a ballot because no one bothers to flip it over, initial bias tends to remain throughout the rest of the contest.