Some of the publicly available things I have produced, or helped to produce, in 2022/23 (in addition to the other pieces on this website).
Researching the Evolving Online Ecosystem Part 1, Part 2 (with the Institute for Strategic Dialogue and CASM Technology). We looked at how developments like the metaverse, blockchain, AI etc. could make online harms - including hate speech, radicalisation, and disinformation - harder to research, and ultimately address. Part 1 outlined a framework of barriers, Part 2 empirically tested the framework on Telegram, Discord, and Odysee. Also available in German and French.
Message-based Community Detection on Twitter (responding to the invasion of Ukraine with CASM Technology). We collected tweets using the hashtags #IStandWithPutin and #IStandWithRussia and clustered them into topics using Natural Language Processing (see image at bottom of page). We discovered that much of the activity was occuring in Africa and South Asia, with potential geopolitical implications for UN votes around Russia's invasion.
Antisemitism on Twitter before and after Elon Musk's Acquisition (with CASM Technology and the Institute for Strategic Dialogue). By combining multiple pre-trained models for detecting hate speech, we found that plausibly antisemitic posts more than doubled in the period after Musk's acquisition. We also saw a substantial spike in new accounts joining (or rejoining) Twitter just after the acquisition, many of which produced antisemitic content alongside other hate speech and conspiracies.
Navigating the Crisis: How Governments Used Intelligence for Decision-making During the Covid-19 Pandemic (with Geoff Mulgan and Anina Henggeler for the International Public Policy Observatory at UCL). We examined how governments across the world used data, models, and other forms of intelligence to make decisions during Covid, and propose recommendations for future crises.
Information Warfare and Wikipedia (with CASM Technology and the Institute for Strategic Dialogue). We examined Wikipedia's defences - human and technological - against coordinated disinformation and information warfare. We also tested methods from social media analysis, in particular network-based methods with a variety of 'seed nodes', and found they were a promising approach to discovery of threats.
Articles / Commentaries
Social Media Futures: Interventions Against Online Unpleasantness (for the Tony Blair Institute for Global Change). I looked at unpleasant behaviour online which does not consitute 'online harms', and the issues and proposed solutions around its effect on public discourse.
Civil Service and Human Connection (runner-up in Cambridge University's Bennett Institute Prospect Public Policy Prize). I considered how government can develop human connections with citizens, and proposed that civil servants, not just MPs, should provide a direct link between constituents and government.
Two ways technology could help to better regulate technology (for the New PPT). Part 1 considered problems with how legislation is made to regulate technology. Part 2 considered the substantial challenges, and ideas, for how technology regution can be actually enforced.
How governments should use intelligence in a crisis: Lessons from Covid-19 (for Apolitical); The Covid Crisis: Lessons for Data Use in Government (for UK Authority); and A Civil Servant's Guide to Making Use of Collective Intelligence (for Civil Service World). Series of articles on various aspects of my work on government decisionmaking during Covid with Geoff Mulgan, Anina Henggeler, and the International Public Policy Observatory at UCL.
BBC Panorama: Elon Musk's Twitter Storm, by Marianna Spring. I was one of the researchers (with CASM Technology and the Institute for Strategic Dialogue) providing research into misogyny on Twitter for this episode.