Crowdsourcing to Solve Big Problems Gary M. Olson Department of Informatics
A Bit of Background n My background is in cognitive psychology –BA U of Minnesota 1967 –MA Stanford U 1968 –PhD Stanford U 1970 n Career: – US Navy – Michigan State U, Dept. of Psychology – U of Michigan, Dept. of Psychology, then School of Information 1983ish got into field of Human-Computer Interaction –Same year I married Judy Olson, with whom I have worked with since 1994 joined the new School of Information –2008-present UC Irvine, Dept. of Informatics n Interests: –Human-Computer Interaction –Computer Supported Cooperative Work –Information Visualization
Crowdsourcing n Having members of the general public do a small thing that can be aggregated into something large and significant n Examples –Christmas bird count –Clickworkers –Galaxy Zoo –Amazon’s Mechanical Turk
Galaxy Zoo n Images from the Sloan Digital Sky Survey n Are galaxies spiral or elliptical –Essentially impossible for computer image software to determine n So people are doing it –Initial pass – 1.25 million galaxies classified n New wave looking at other characteristics –Galactic mergers –Supernovae –Solar storms
Hanny van Arkel n Dutch schoolteacher n Discovered a new galaxy type –Using Galaxy Zoo –Called “Hanny’s Voorwerp”
Crowdsourcing n Having members of the general public do a small thing that can be aggregated into something large and significant n Examples –Christmas bird count –Clickworkers –Galaxy Zoo –Amazon’s Mechanical Turk
Climate CoLab n A project centered at MIT –Thomas Malone as leader n I’m involved in it as well n An attempt to raise the level of public discussion of climate change issues
Some Recent Findings n As of September, 2014 –220,000 unique visits –24,000 registered members n Surveys of users –Two demographic surveys –One on effects
Challenge: Climate Change is an Example of a “Wicked Problem” n Can such a problem be successfully solved via crowdsourcing? n Existence proof: Several successful contests n Analogies: other “wicked” problems are being approached –Writing Wikipedia articles –Large scale contests like Innocentive n Climate Colab is an evolving research project –With the possibility of an important social impact
Wisdom of Crowds
Why Does This Work? n Surowiecki – draws on the larger literature on markets –Cognition –Coordination –Cooperation
Criteria for Success n Diversity of inputs – each contributor has their own unique input n Independence – each contributor’s input is independent of others n Decentralization – each contributor draws on their own analysis n Aggregation – there is a way to merge all the contributions into a collective decision
Failures of Collective Action n Homogeneity – everyone thinks the same; Groupthink n Centralization – Columbia shuttle disaster ignored inputs from engineers n Division – 9/11 Commission Report faulted isolation of information n Imitation – using past decisions n Emotionality – peer pressure, etc.
An Old Idea Princeton University Press, 2005
Thank you – Questions? n