engleski

Working for the algorithm: the case of Google raters

The dominance of Google for information search on the web is hard to overemphasize. Since the entrance of the company in the field in 2004 it has developed rapidly to take the leading position and become a single, dominant player (Vaidyanathan, 2011). It has diversified its business model to include such services as YouTube, the Android platform for mobile phones, Gmail, etc. However, its dominant product is web search which depends heavily on internet users and their labour (Lee, 2011). Simultaneously, search-result standards such as customer satisfaction and relevance are given precedence over fairness and representativeness (Van Couvering, 2007). The company justifies this position through a legitimizing digital discourse (Fisher, 2010) in which the algorithm, it is claimed, is constantly and cumulatively improved to provide neutral and objective search results. The claims of objectivity are the basis for maintaining the position of algorithms as ˝legitimate brokers of relevant knowledge˝ (Gilespie, 2014). In that sense we view the algorithm as an instrument for control and domination (Marcuse 1998). However, recently revealed information about globally distributed workers, the so-called raters, opens strong counter-arguments to this digital discourse. The raters are periodically provided lists of Google search queries to match them with the websites that return the highest estimated semantic relevance to each individual query. In other words, according to the instructions given by Google, the raters simply rate the relevance of search results for individual search queries. Google claims that they test new tweaks and updates to the algorithm and those human estimates and interpretations do not enter the algorithm´s search index. While it is impossible to know exactly what the actual influence on indexing is, it nonetheless allows us to make certain claims from an ideology critique position. The work of raters points to certain fractures in the technological legitimacy discourse (Fisher, 2010 ; Rieder and Sire, 2013) which claims the technological supremacy of the algorithm can answer global information needs of all internet users. Similarly to Garnham (2006) who argued against the division between base (economy) and superstructure (ideology) in critical thought, we will attempt to connect the ideological aspects of Google´s digital discourse with concrete work practices of Google raters. This will enable us to see where ideology intersects with labour and how they are connected. We will also explore how the work of raters is fitted into the company’s digital discourse and how technology mediates the relationship between the raters and the company. Taking this type of work into account enables us to analyse Google´s global power and value creation as well as to discern the influence of technology and other relevant factors in Google´s business. In the presentation we will provide some empirical data from the analysis of Google´s official documents. This type of work is task oriented and localized since raters only perform tasks for local language search queries. The raters do not work on the same query list at the same time. Unlike crowdsourcing where internet users work towards a common goal, the individual raters accept available tasks and get remunerated for the solved tasks. This process points toward a novel and specific type of fragmented, individualized and localized type of virtual work for a global multinational company.

Algorithm; search engine; economy; ideology; digital discourse; work

nije evidentirano