Trusts, co-ops, and crowd workers: Could we include crowd data workers as stakeholders in data trust design?

Richard C Gomer,Elena Simperl

doi:10.1017/dap.2020.21

Abstract

AbstractData trusts have been proposed as a mechanism through which data can be more readily exploited for a variety of aims, including economic development and social-benefit goals such as medical research or policy-making. Data trusts, and similar data governance mechanisms such as data co-ops, aim to facilitate the use and re-use of datasets across organizational boundaries and, in the process, to protect the interests of stakeholders such as data subjects. However, the current discourse on data trusts does not acknowledge another common stakeholder in the data value chain—the crowd workers who are employed to collect, validate, curate, and transform data. In this paper, we report on a preliminary qualitative investigation into how crowd data workers themselves feel datasets should be used and governed. We find that while overall remuneration is important to those workers, they also value public-benefit data use but have reservations about delayed remuneration and the trustworthiness of both administrative processes and the crowd itself. We discuss the implications of our findings for how data trusts could be designed, and how data trusts could be used to give crowd workers a more enduring stake in the product of their work.

Highlights

Datasets that are of economic and social value often have complex provenance, arising from a sequence of actions spanning multiple stakeholders
We argue that crowd workers are a recipient stakeholder group who are affected, negatively or positively, by the ways in which datasets are curated and exploited, and that there is scope for novel governance models such as data trusts or data co-ops to attend to the needs of crowd workers
The results of Survey 2 suggest that crowd workers expect to be remunerated fairly in return for their time and effort, even where their work contributes to a public benefit; and we suggest that this factor probably sets a lower bound on what they are prepared to accept in exchange for granting access to a crowd-produced dataset

Summary

Introduction

Datasets that are of economic and social value often have complex provenance, arising from a sequence of actions spanning multiple stakeholders. Data trusts have been proposed as a means of increasing the availability of economically or scientifically valuable datasets (Hall and Pesenti, 2017; House of Lords, 2018) or allowing data subjects to disrupt current “feudal” approaches to data governance (Delacroix and Lawrence, 2018). There are compelling arguments, beyond ethics, for treating a work force fairly and—even within a dominant system of free-market capitalism—most countries have, albeit to varying degrees, adopted worker protection rules that speak to the social value of secure and rewarding employment. There are three reasons that compel us to consider data trusts as a means of disrupting current crowd work practices. Granting crowd workers a stake in a dataset is potentially a pragmatic means of recognizing their work and improving their remuneration via an ongoing return. Not all forms of crowd work produce such a delineated output to which a stake could be attached

Objectives

Methods

Results

Discussion

Conclusion