(p.9) As recounted by Schwartz (2015) in a piece tellingly titled ‘Human pretending to be computers pretending to be human’, in 1770 Wolfgang von Kempelen presented in Vienna to Empress Maria Theresa a sort of robot (see picture) that could beat humans at playing chess; he called it the ‘Turk’. The ‘Turk’ toured Europe and evoked contrasting responses, as some were ready to admit and welcome that machines were surpassing humans but many opposed this view. Although the Turk included a ‘labyrinth of levers, cogs and clockwork machinery’ obviously it was not using any algorithm and was operated by a person...

(p.12) As put by economist Timothy Taylor in his blog (Taylor, 2015), the use of the ‘sharing economy’ expression to refer to various commercial platforms is a ‘triumph of public relations artistry’ (he would rather use ‘the matching economy’). This consideration clearly applies to digital labour markets and it would suffice to observe their revenues and market evaluation; unlike community-based ‘time banking’ digital platforms where ‘true’ sharing can occur, commercial initiatives such as TaskRabbit fail to cater for less advantaged members of the community (Thebault-Spieker, et al., 2015). The same goes for the ‘crowd working’ or ‘crowd employment’ labels resounding the...

(p.55) The enactment of some form of regulation to establish the proposed portability of benefits would already represent a positive step forward to ensure more dignified conditions for workers in digital labour markets; various analysts, however, consider it insufficient in view of the facts that earnings are at times too low in the absence of any minimum wage rules, the flow of work is unstable and no employment benefits exists, there are clear information and power asymmetries, no protection against privacy violations, and various forms of information or reputation-based ethnic and gender discriminatory mechanisms occur unregulated.

(p.64) A far more effective means of control was the ‘algorithmic management’ enabled by platform-based rating and reputation systems (Lee et al., 2015; Rosenblat and Stark, 2016). Workers were rated by their clients following the completion of tasks. Workers with the best scores and the most experience tended to receive more work due to clients’ preferences and the platforms’ algorithmic ranking of workers within search results. This form of control was very effective, as informants stressed the importance of maintaining a high average rating and good accuracy scores. Whereas Uber’s algorithmic management ‘deactivates’ (dismisses) workers with ratings deemed low (Rosenblat and...