Aggregate and filtered content
In addition to the level of complexity of a task, and the distribution of the costs, crowdsourcing projects can be organized based on the relation between contributions to achieve the final goal. Some projects break up the process into small parts, such as micro-tasks and voting, which are later aggregated to create a final product (Prpic et al., 2015). A well-known example of this is the crowdsourcing Zooniverse platform,6 where the crowd is asked to tag, match, classify, or identify characteristics of images and videos from large datasets generated by space telescopes or cameras in natural parks. The data collected is then aggregated and analyzed to develop new insights about galaxies or species behavior. The individual contributing members of the crowd do not need to interact, and the task started by one can be continued by another at any time. The platform Zooniverse also provides humanities content such as hand-written texts and asks the crowd to transcribe them, following the same principle.
A similar example is the Foldit project developed by the University of Washington. The crowd plays with protein molecular structures to identify images (e.g., density maps). Results are more accurate interpretations of protein structures than those from automated algorithms, and the crowd can process images faster than previous cryo-electron micrographs or X-ray crystallographs (Bkoep, 2019). The aggregated micro-tasks have served to publish a series of scientific papers by the university team, who kindly acknowledge the contributions of the Foldit players (estimated to surpass the 57,000)7
In contrast, crowdsourcing projects can rely on one individual to create a finished product, which may be then selected by the requesting firm and adopted for production. This is the case with many idea generation projects where firms outsource the creation of an innovation or the anticipation of predictions (Hossain & Kauranen, 2015). An example is NASA’s $30,000 award to a solution for the prediction of solar flares (Nam, 2011).
Management of the crowd
Crowdsourcing projects can allow centralized or decentralized authority, open (e.g., interoperability, future reuse of the gathered information) or closed participation (e.g., anonymous participation, peer review), and a range of relationships (from unknown participation to full collaboration). Activities can be objective, as transcribing a text, or subjective, as voting for “the best of.” Firms may request funds to finance a project, devise a tournament type crowdsourcing (giving a prize for the best idea), call for virtual labor (made up of a series of micro-tasks), or engage in an open collaboration (such as a wiki) (Prpic, Taeihagh & Melton, 2015).
Not all crowdsourcing projects are collaborations between crowdsourcee and crowdsourcer. Projects such as Google’s CAPTCHA tap into the crowd to do a job without encounters or relationships between actors. Platforms such as del.icio.us or YouTube are not clear in their crowdsourcing nature, as the call, the benefits, the tasks, the nature of the participation, and the role of the initiator are not clearly identified (Estelles-Arolas & Gonzalez-Ladron-De-Guevara, 2012). Such aggregators utilize the crowd’s input, including content, behavior, and ratings, to prioritize content output without calling the crowd for a greater goal. Most crowdsourcing projects, however, do call the crowd.
Attracting the right crowd and aligning efforts to the project’s goal is essential, though not as easy as one would think. Crowds respond to information about the firm or platform where the project takes place as well as to signals from other participating members of the crowd (Belleflamme et al., 2019). Breaking up problems into tasks doable for individuals outside of the firm requires sufficient organizational capacity from the firm in order to return usable results, which explains the emergence of specialized crowdsourcing platforms with a history of constructing crowds, offering paid services to intermediate coordination (Prpic et al., 2015).
Consistent and good quality participation from the crowd requires a clear central idea, as well as transparent governance structure. An important point is the ownership of the final product. An example of a jointly owned product is open source software (OSS), free to use and developed by a crowd generally driven by altruistic motivations, which often outperforms commercial products. Well-known examples include Linux, Mozilla Firefox, OpenOffice, and Ubuntu. OSS is dependent on the active crowd to assess, improve, and evaluate the software to increase its quality, and to keep it usable (Wexler, 2010).
Another example is the free online encyclopedia Wikipedia, with over 49 million articles (6 million of which are in English), edited by 200,000 contributors every month, available in 307 languages, and viewed by over 21 billion individuals across the world on a monthly basis.8 The project is one of many products coordinated by the Wikimedia Foundation as part of a free knowledge infrastructure aiming at giving access to all the knowledge of the world/ It has become an important global information resource, a socially constructed common good (Hess, 2012).
