A professor and several PhD students at MIT examine the challenges
and opportunities in human computation.
By Robert C. Miller, Greg Little, Michael Bernstein, Jeffrey P. Bigham,
Lydia B. Chilton, Max Goldman, John J. Horton, and Rajeev Nayak
Crowd computing is quickly becoming an essential part of the technology landscape. Crowd computing encompasses the interaction among large numbers of people facilitated by software systems and networking technology. Crowds—and by “crowds,” we literally mean a mass of people—are themselves the power that fuels sites like
Wikipedia, Twitter, Intrade, and even online labor markets like Amazon Mechanical Turk.
One way to think about crowd computing is as the human analogue to
cloud computing. Where the cloud provides access to elastic, highly available
computation, and storage resources in
the network, the crowd represents access to elastic, highly-available human
resources, such as human perception
and intelligence. Crowd computing offers the strength of software with the
intelligence and common sense of human beings.
One variant of crowd computing is human computation, which we define as
using software to orchestrate a process
of small contributions from a crowd to
solve a problem that can’t be solved by
Human computation was first pop-
ularized by Games With a Purpose
( http://gwap.com), in which the com-
putation is a side effect of a fun game
[ 8]. For example, the ESP Game asks
two players to guess words associated
with an image, scoring points when
their words agree, which makes the
game fun, but also generating useful
labels to index the image for search-
ing, which makes it an act of human
What application areas will benefit
the most from human computation?
What properties do certain problems
possess that make them amenable to a
successful solution by a hybrid human-software system? Since the end user of
such a system is also, typically, human,
we can refine this question further:
Why does a human end user need to
request the help of a human crowd to
accomplish a goal, rather than just doing it herself?
One reason is differences in capability. A group of many people has
abilities and knowledge that one single end user does not, either innately
or because of situational constraints.
For example, Viz Wiz [ 1] helps blind us-