Diversity and Collaboration

Week Overview: Our group met on March 23, 2010 at 10:00pm in the Michigan League. We discussed the outlines of the chapters we read and plans for our future reading schedule.

This blog post covers Section II: Where We Are (Chapters 5 and 6) of Crowdsourcing. See the following posts for a review of the other chapters in Section II.

Diversity Trumps Ability Theorem:

• Scott E. Page. With an interest in computer programming and behavioral economics, Scott E. Page - a modern Francis Gaulton - simulated the economic actions of a collection of highly intelligent agents and a separate collection composed agents with mixed levels of intelligence. In several repeated trials, the diverse collection of agents outperformed the expected geniuses dubbed "Mensa agents." (This surprise inspired him to write a book: The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools and Societies). pg. 132
• MATLAB. In 1999, Ned Gulley proposed and implemented a computer programming competition that generated fast-paced results. As each programmer submitted a solution entry, it was judged and posted to a website. At that point, other programmers were free to use any of the previously submitted code - many choose to slightly modify code, the bulk of which was written by someone else. Since 1999, the competition has been hosted twice a year each year. "On average, Gulley notes, the best algorithm at the end of the contest period exceeds the best algorithm from day one by a magnitude of one thousand." Frequently, it is the unexpected contributors (amateurs) that provide the greatest innovation and improvements to code. pg. 138
• The Wisdom of Crowds by James Surowiecki. Jeff Howe references The Wisdom of Crowds and shares the Who Wants to be a Millionaire example. pg. 142

Harnessing collective intelligence:

1. Prediction markets --> primary ex(s):

• Iowa Electronic Market. The IEM is explained as in The Wisdom of Crowds, but to a lesser extent. pg. 161
• Policy Analysis Market. Robert Hanson, a firm believer in prediction markets, designed PAM to allow participants to bet on questions of political policies (e.g. likelihood of events in foreign affairs, direction of GDP, level of national security). One component of PAM was also a terrorism index, which gained the attention of the Terrorism Information Awareness Office. This ultimately brought about the end of PAM though its function and potential were comparable to the IEM. pg. 164

2. Crowdcasting networks --> primary ex(s):

• InnoCentive. Jeff Howe draws examples from InnoCentive throughout his work. The following story of Ed Melcarek is worth highlighting. As of 2007, Melcarek had solved 7 of InnoCentive's challenges - a commendable accomplishment. Though he earned his master's degree in physics, all of the InnoCentive challenges he has won, he selected from other categories (e.g. chemistry and biology). Each win has brought him financial support and moreover, confidence. Yet again, the unexperienced crowd member has outsmarted the challenge participants competing in their specialization. Another note - A Ph.D candidate, Karim Lakhani looked into the challenge-solution process of InnoCentive and discovered that "75% of the solvers already knew the solution to the problem." Therefore, finding the best solution is often a task of finding the right person - the solutions already exist outside the scope of internal R&D teams. pg. 150
• Netflix. In 2006, Netflix challenged the world to create an algorithm that could improve its recommendation system by 10%. Submissions were scored and their leaders were placed on a leaderboard, but their code was not displayed. However, many leaders decided to share their code without Netflix facilitation - each hoping to interact with other competitors and learn from the process. pg. 155
• Marketocracy. Ken Kam and Mark Taguchi made Marketocracy (a website) to identify savvy stock traders. Anyone could make an account and would then be given $1 million dollars in Monopoly money to invest. Though the designers planned to use it to locate the best investors, they found it could also be used to locate the best stocks. Meanwhile, investors (new and old) could test new portfolio collections without the risk of real financial failure. Marketocracy proved to be a powerful prediction market, but the algorithms behind it have been modified and seem to still be changing as downfalls are discovered. pg. 169

3. "Idea jams" --> primary ex:

• Dell IdeaStorm. Dell created IdeaStorm as a channel of innovative ideas from consumer to the company. Anyone can submit an idea and anyone can review ideas with thumbs-up or thumbs-down markers. Dell tracks the best ideas based on consumer opinions and implements them in their development (e.g. computers with pre-installed Linux OS). pg. 158