ABSTRACT
This chapter argues that accountability structures are shaped by application structures and the structures of control of algorithms. Governance and accountability are hampered, for instance, by the lack of transparency and high complexity of algorithmic systems, the fragmentation and heterogeneity of the involved industries, the increasing autonomy of the technological systems and the increasing distribution of action between humans and machines. In the context of the Internet, algorithms are abstract programme procedures, which are employed in software programmes to screen data, assign relevance to information, select elements and put them in order. The conceptions of action and agency can also be applied to (semi-)autonomously operating algorithms. The debate on algorithmic accountability includes questions about how algorithms can be controlled without the public disclosure of codes. The increase of automation, autonomy and independence of technology leads to decreasing predictability and controllability and to challenges for governance and accountability.
