AUTHOR
Engin Bozdag and Jeroen van den Hoven
ABSTRACT
Almost 50 years ago individual users at terminals communicated over telephone lines with a central hub where all the computing was done. The shift back to this model is currently under way. Data and programs are being swept up from desktop PCs and corporate server rooms are installed in “the computer cloud”. When you create a spreadsheet with Google Docs, major components of the software reside on unseen computers, whereabouts unknown, possibly scattered across continents (Hayes, 2008). This paradigm, known as Cloud Computing, allows users to “outsource” their data processing needs to a third-party (Jaeger et al., 2008). Thus, the computing world is rapidly transforming towards developing software for millions to consume as a service, rather than to run on their individual computers (see Buyya et al, 2009). Cloud computing changes the way software is designed and it is becoming ubiquitous. Major IT providers such as Google, Microsoft, Sun, and IBM are all offering cloud services. Intel has launched a vision for the next decade. Government agencies recently have started to use cloud applications and it is expected that a significant part of all the financial, economic and logistic transactions will be performed (semi-) automatically “in the cloud” (Buyya et al, 2009).
Cloud Computing changes the way software is defined, developed, marketed, sold and used. In Cloud Computing, the computer is valued as a gateway to computing services and resources in distant places. It is no longer what is on your desk-(top) or on the server in the basement of your office that counts, but rather which services, facilities and resources you have access to. The software is no longer a digital commodity that you install on your local machine, but rather a service offered by providers that you access and share with other users.
The internet and the web have become full fledged social environments; they facilitate and enhance known and traditional social phenomena by means of social software (e.g. social networking sites and collaborative software). In this way a mesh of computer networks, social software and interacting human persons has come in to being. Most cloud services not only allows the user to store and process a data in remote servers, but it also allows the user to share this data. This leads to an intertwine of cloud services and social computing.
While social phenomena such as crowdsourcing and large-scale social computing projects , cooperative computing are receiving attention, another form of social computing is emerging in cloud services that are storing a large amount of user data. Search engines or other social cloud services such as Facebook cannot solely rely on algorithms to analyze and modify this data, since these may perform imperfectly or may work with bad data generated by spammers, abusers, non-complying users. These services will also use human reviewers, to analyze, understand, filter, remove ,modify, add, sort, categorize the data if necessary. They will use the analyzing capacity of humans, to make judgements when an algorithm cannot decide. For instance a website scoring high in the search engine could be engaged in bad practices, such as spam, violation of TOS of the search engine. This violation might go undetected by the algorithm, but can be detected by a human controller, because of her epistemic and moral capabilities. However, the website in question can be a source of knowledge valuable to many people, leading to a moral dilemma on the part of the human controller, who is in charge to support or supplement the algorithms. If the human agent’s decision on the website is fully left to her own discretion, then this may lead to biases in the search engine results and the computation will differ per human agent, a problem that unassisted algorithms don’t have.
In this paper, we discuss the risks involved in using people as processing units, much like computer processes or subroutines, to provide a cloud service. We argue that, while designing a social computing application, a cloud service provider should have clear policies to minimize this individual judgment, without compromising the functionality it adds to the overall performance of the social computing application.
REFERENCES
[Buyya et al., 2009] Buyya, R., Yeo, C., Venugopal, S., Broberg, J., and Brandic, I. (2009). Cloud computing and emerging IT platforms: Vision, hype, and reality for delivering computing as the 5th utility. Future Generation Computer Systems, 25(6):599–616.
[Hayes, 2008] Hayes, B. (2008). Cloud Computing. Commun. ACM, 51(7):9–11.
[Jaeger et al., 2008] Jaeger, P., Lin, J., and Grimes, J. (2008). Cloud computing and information policy: Computing in a policy cloud? Journal of Information Technology & Politics, 5(3):269–283.