Algorithmic selection on the Internet

Operating modes, significance and effects of automatic selection processes in the information society

This project grasps and assesses an Internet phenomenon that is developing into one of the central innovations of current media change: algorithmic selection, a distinctive feature of various societal communications processes.

Algorithmic selection of information constitutes the common technological and functional core of various Internet services and is growing rapidly in importance. It is the underlying technological function of diverse applications such as news aggregators (Google News, Techmeme), search engines (Google, Bing, Yahoo), recommender systems (Amazon book recommendations, LastFM), computational advertising (Google AdSense) and reputation systems (eBay feedback). These applications automatically assess the relevance of web documents, choose the most important news articles, recommend books, songs or movies, display personalised advertising and determine the reputation of companies and scholars.

The societal implications are manifold: On the one hand algorithmic selection has become indispensable in the information society. It contributes substantially to reducing information overload by filtering the growing amount of information and by selecting and sorting information by relevance. It helps to reduce information asymmetries, lower search costs and improve market transparency. Algorithmic selection structures the supply of information, provides an orientation to users and thus allocates user’s attention – the scarce resource of the information society. On the other hand algorithmic selection often operates on the basis of intransparent processes. This results in fierce societal controversies because of potential risks such as manipulation, the violation of fundamental and personal rights, and the concentration of (market) power.

A comprehensive and systematic analysis of algorithmic selection as the core of various applications/services and its societal implications is still lacking. With the exception of search engines such applications have not been analysed as distinct phenomena, let alone as an overall distinctive feature of Internet services’ development. This research project closes this research gap with a comparative overall view and thus establishes the basis for further research. It examines the various applications of algorithmic selection as a central structuring innovation of Internet development. It shows the technological, economic and political/regulatory characteristics and effects of these new Internet services that are based on algorithmic selection. The project analyses possible generalisations of the phenomenon of algorithmic selection and shows the likely consequences for politics, the economy and communication research.

Publications

The project is supported by the Swiss National Science Foundation.