Digital wildfires (spread of dubious and dangerous information, hate speech and rumours, via social media) can seriously challenge the capacity of traditional media, civil society and government to report accurately and respond to events as they unfold. But how people communicate in these digital social spaces is not well understood; users may not fully understand how these spaces ‘work’ as channels of communication and so what constitutes appropriate and responsible behaviour may be unclear. The challenge then is to develop appropriate ways of governing these spaces and how to apply and use them responsibly.
The research project, led by Dr Marina Jirotka, will attempt to address this challenge by framing the study in a programme of work known as Responsible Innovation in ICT and by developing a methodology for the study and advancement of the responsible governance of social media.
A key question is to what extent do people in these spaces ‘self-regulate’ their behaviour? If this is evident then there is a case for exploring how self-correction mechanisms may be amplified so that false rumours are identified more quickly.
The legitimacy of new governance mechanisms may be enhanced if they respect and build on such existing self-governance techniques.
Ethics statement submitted as part of the Je-S proposal
Ethical considerations are paramount in this proposed study. The team are already fully cognisant of the challenges online methods pose to social science researchers and have already composed an ESRC-funded guide to using and repurposing multi-modal data. A key objective of the current ESRC project is to expand this guide to include data harvested from social media sites.
In the proposed project we will engage with issues of harm, informed consent, privacy, civil liberties and data reuse and repurposing. The computational technology to be used in the proposed study affords users of social media complete anonymity and confidentiality. Only characteristics of the users will be recorded for analysis and all data will be stored on password protected university systems.
We will expand this guide to include data used to inform the new tool and model methodologies. We will continue to thoroughly engage with issues of harm, informed consent, privacy, civil liberties and data reuse and repurposing. The harvesting techniques used by COSMOS can collect information on the gender, location and account identifier of tweeters. This information is stored on password and firewall protected systems. All data that are presented to researchers are pre-anonymised by the system, maintaining confidentiality. Quantitative data outputs are presented at aggregate level meaning no identifying information is presented.
Qualitative data outputs can include data that renders tweeters identifiable. Twitter’s terms of service state that users of their data should not alter content in any way, and that tweets should remain as the tweeter intended. This requires tweets to be published in full and leads to important considerations regarding informed consent and the privacy of users. However a number of academic journals require tweets to be anonymised as a condition of publishing – in contradiction to Twitter’s terms of service. During the project we will engage with Twitter and other stakeholders in order to identify a best practice approach to the publication of tweets and the assessment of potential risks involved in publication. We will seek an approach that balances the rights and privacy of users with the needs of researchers.
The fieldwork will be conducted in accordance with Oxford University’s Research Governance Framework. Data collection and storage will follow the data protection policies of participating institutions, copies of which can be supplied. All data will be de-identified and raw data will be stored on a non-networked encrypted hard drive. The RA will undertake social science ethics training ahead of data collection. We will also obtain ethical approval from Oxford, Cardiff and Warwick Universities prior to the project commencement.