Human-Based Computation

Human-based computation is a technique where parts of a computational process are outsourced to a person or group of people. This results in a symbiotic interaction between humans and computers in which the abilities and costs of each are analyzed. The process is different than traditional computation in that the roles are reversed. A computer program asks for the solution to a problem and people are tasked to solve the problem. The computer then collects, interprets and integrates the information.

Human-based computation originated for early work on algorithms of interactive evolutionary computation by Richard Dawkins. The early programs reversed the usual interaction between humans and computers. The growth of the internet has enabled the research on this technique to shift towards computers asking large crowds of humans for solutions rather than a single person or small group of internal workers.

Classes of computation have been used to describe the division of labor and the roles of humans and computers in this method. These classes are referred to by two letter abbreviations. The first letter of the abbreviation identifies the type of agent performing the task and the second refers to the type of selection agent. There are several methods of human-based computation. One is the HH2, of which an example is a wiki. A wiki enables multiple users to edit web content in a continuous manner. An extremely popular wiki is the online encyclopedia, Wikipedia. Another method is the HC, which involves interactive online games that gain knowledge from people in entertaining ways. Computerized tests are also HC and include CAPTCHAs that make use of problems that are easy for humans to solve and difficult for computers, which is beneficial in implementing effective security measures.

There are several incentives used to get humans involved in computation projects. Motivations include monetary compensation on websites such as Amazon Mechanical Turk, curiosity, volunteerism, desire to be entertained or share knowledge, online recognition or reputation, fun and the desire to influence the final results of the system. Some projects include a combination of these incentives. Human-based computation organizations are usually more productive and robust than traditional computation organizations. These organizations give people the chance to be spontaneous and creative in constructive ways without endangering the functionality of the whole project, leading to a new distribution of efforts called crowdsourcing.