Karsten Donnay

With armed conflicts at a historic high and attacks on civilians rising ¹ , understanding the evolving nature of conflict is a critical research priority. The current paradigm in conflict research relies heavily on text-based data, using fatalities as the primary—and often sole—proxy for violence intensity. Although these data have expanded our ability to study armed conflict, they exhibit inherent limitations due to uneven human reporting ^2–6 . War damage assessments based on satellite data ^7–13 offer a complementary perspective. Satellite-derived data have their own limitations, but these arise from different mechanisms, creating distinct, complementary strengths that can be leveraged through data integration. Here we propose three concrete approaches to integration: improvement, enrichment and fusion. Each bridges a different gap in the underlying data sources. We use case studies from Ukraine and Myanmar to illustrate how integration can be implemented in practice and the novel analytical insights that emerge. Prioritizing data integration enables a paradigm shift away from fatality-centric research towards a broader spectrum of violence, revealing the complexity of conflict dynamics.

In this paper, we contribute to the debate on generative artificial intelligence (GenAI) in scientometrics. We argue that moving from a trial-and-error approach to an explainable and actionable use requires a principled understanding of strengths and weaknesses of GenAI as compared with other techniques and with human judgment. To this end, we introduce a conceptual framework based on the distinction between the semantic dimensions of texts, i.e., the meanings attributed to words, and their pragmatic dimension, i.e., their embedding within communicative situations. We leverage this framework to interpret the results of applications of GenAI in scientometrics and to provide guidance to users. Specifically, we conclude that key parameters to be considered are the nature of the task, the level of granularity of the analysis and whether the goal was descriptive, inferential or evaluative. These parameters lead to different strategies for using GenAI and human–machine integration. Finally, we suggest that, by generating large amounts of scientific language, GenAI might affect textual characteristics used to measure science, such as authors, words, and references. We argue that careful empirical work and theoretical reflection will be essential to remain capable of interpreting the evolving patterns of knowledge production in the age of AI.

Propaganda plays a key role in maintaining power in authoritarian regimes. Previous research finds that overt, crude, and heavy-handed messaging, so-called hard propaganda, can be used to effectively convey government strength and deter citizens from joining anti-regime protests in relatively stable autocratic regimes like China. Yet, it is unclear if this is also true in more contested and unstable autocratic contexts. In these settings, citizens are more likely to question such messaging and prior beliefs of government strength vary more widely. We explore the perception of hard propaganda in one such difficult test case for hard propaganda: the crisis-ridden Maduro regime in Venezuela. We measure perceptions of hard propaganda among the public using an online survey that featured a choice experiment in which respondents chose between and rated different propaganda images against more neutral political communication. Our results show that respondents perceived hard propaganda images as stronger compared to neutral political communication. This holds true---contrary to our pre-registered expectations---regardless of whether respondents overall perceived the government as strong or weak. Moreover, respondents reported a lower willingness to join anti-government protests but, at the same time, had a greater motivation to challenge the regime. These results support and extend prior findings on the effectiveness of hard propaganda in deterring anti-regime activities to the case of contested and unstable autocracies. But they also suggest that this kind of messaging erodes regime legitimacy providing the first evidence outside of the Chinese case of the pathology of hard propaganda.

Cross-document coreference resolution (CDCR) identifies and links mentions of the same entities and events across related documents, enabling content analysis that aggregates information at the level of discourse participants. However, existing datasets primarily focus on event resolution and employ a narrow definition of coreference, which limits their effectiveness in analyzing diverse and polarized news coverage where wording varies widely. This paper proposes a revised CDCR annotation scheme of the NewsWCL50 dataset, treating coreference chains as discourse elements (DEs) and conceptual units of analysis. The approach accommodates both identity and near-identity relations, e.g., by linking “the caravan” - “asylum seekers” “those contemplating illegal entry”, allowing models to capture lexical diversity and framing variation in media discourse, while maintaining the fine-grained annotation of DEs. We reannotate the NewsWCL50 and a subset of ECB+ using a unified codebook and evaluate the new datasets through lexical diversity metrics and a same-head-lemma baseline. The results show that the reannotated datasets align closely, falling between the original ECB+ and NewsWCL50, thereby supporting balanced and discourseaware CDCR research in the news domain.

The OpenSky Network has grown into one of the largest community-driven air traffic surveillance infrastructures worldwide. While its data are widely used in research, comparatively little is known about the individuals who maintain the sensor network. In late 2024 to early 2025, we conducted the first large-scale user survey of OpenSky Network members to better understand the demographics, motivations, and barriers associated with participation. This paper presents key findings from more than 500 responses, highlighting who contributes to OpenSky and why, and what prevents others from becoming active feeders. The collected data show that participants are predominantly well-educated, above-median-income males from Western countries, with an approximate average age of 50 years. Their primary motivation is contributing to research, despite limited knowledge of the specific research projects. Cost is the main barrier to participation, particularly in underrepresented regions, whereas disinterest and environmental concerns deter adoption in other areas.

Hate speech is widely seen as a significant obstacle to constructive online discourse, but the most effective strategies to mitigate its effects remain unclear. We claim that understanding its distribution across users is key to developing and evaluating effective content moderation strategies. We address this missing link by first examining the distribution of hate speech in five original datasets that collect user-generated posts across multiple platforms (social media and online newspapers) and countries (Switzerland and the United States). Across these diverse samples, the vast majority of hate speech is produced by a small fraction of users. Second, results from a pre-registered field experiment on Twitter indicate that counterspeech strategies obtain only small reductions of future hate speech, mainly because this approach proves ineffective against the most prolific contributors of hate. These findings suggest that complementary content moderation strategies may be necessary to effectively address the problem.

The aim of this paper is to review the use of GenAI in scientometrics, and to begin a debate on the broader implications for the field. First, we provide an introduction on GenAI's generative and probabilistic nature as rooted in distributional linguistics. And we relate this to the debate on the extent to which GenAI might be able to mimic human 'reasoning'. Second, we leverage this distinction for a critical engagement with recent experiments using GenAI in scientometrics, including topic labelling, the analysis of citation contexts, predictive applications, scholars' profiling, and research assessment. GenAI shows promise in tasks where language generation dominates, such as labelling, but faces limitations in tasks that require stable semantics, pragmatic reasoning, or structured domain knowledge. However, these results might become quickly outdated. Our recommendation is, therefore, to always strive to systematically compare the performance of different GenAI models for specific tasks. Third, we inquire whether, by generating large amounts of scientific language, GenAI might have a fundamental impact on our field by affecting textual characteristics used to measure science, such as authors, words, and references. We argue that careful empirical work and theoretical reflection will be essential to remain capable of interpreting the evolving patterns of knowledge production.

In authoritarian regimes, governments regularly use overt and heavy-handed visual propaganda to effectively signal regime strength and deter protests. Can democratic leaders also use this so-called hard propaganda to project strength, or does this kind of authoritarian-style messaging potentially backfire because of societal norms for leaders’ behavior? Focusing on a rare instance in which outright visual hard propaganda was used in a democratic setting, we study how US citizens perceived its use by the Trump administration during the 2020 Black Lives Matter protests. In a preregistered online survey fielded from June 12 until June 16, 2020, we exposed participants from both sides of the political spectrum to randomized pairs of real-world propaganda images. This allowed us to explore how partisanship influenced perceptions of hard propaganda during the height of the protests. Our findings reveal that, compared to standard political messaging, hard propaganda can communicate greater strength to both government supporters and opponents in a democracy. Yet, in contrast to autocratic settings, it fostered opposition among Trump opponents. Trump supporters, however, did not find such propaganda any more or less appropriate than standard political communication, consistent with an increased acceptance of authoritarian practices within polarized democracies.

Efforts to curb online hate speech depend on our ability to reliably detect it at scale. Previous studies have highlighted the strong zero-shot classification performance of large-language models (LLMs), offering a potential tool to efficiently identify harmful content. Yet for complex and ambivalent tasks like hate speech detection, pre-trained LLMs can be insufficient and carry systemic biases. Domain-specific models, fine-tuned for the given task and empirical context could help address these issues but, as we demonstrate, the quality of data used for fine-tuning decisively matters. In this study, we fine-tuned GPT-3.5 using a unique corpus of online comments annotated by diverse groups of coders with varying annotation quality: research assistants, activists, two kinds of crowd workers, and citizen scientists. We find that only annotations from those groups of annotators that are better than zero-shot GPT-3.5 in recognizing hate speech improve the classification performance of the fine-tuned LLM. Specifically, fine-tuning using the two most high quality annotator groups -- research assistants and Prolific crowd workers -- boosts classification performance by increasing the model's precision without notably sacrificing the good recall of zero-shot GPT-3.5. In contrast, low quality annotations do not improve or even decrease the ability to identify hate speech.

Propaganda plays a key role in maintaining power in authoritarian regimes. Previous research finds that overt, crude, and heavy-handed messaging, so-called hard propaganda, can be used to effectively convey government strength and deter citizens from joining anti-regime protests in relatively stable autocratic regimes like China. Yet, it is unclear if this is also true in more contested and unstable autocratic contexts. In these settings, citizens are more likely to question such messaging and prior beliefs of government strength vary more widely. We explore the perception of hard propaganda in one such difficult test case for hard propaganda: the crisis-ridden Maduro regime in Venezuela. We measure perceptions of hard propaganda among the public using an online survey that featured a choice experiment in which respondents chose between and rated different propaganda images against more neutral political communication. Our results show that respondents perceived hard propaganda images as stronger compared to neutral political communication. This holds true – contrary to our pre-registered expectations – regardless of whether respondents overall perceived the government as strong or weak. Moreover, respondents reported a lower willingness to join anti-government protests but, at the same time, had a greater motivation to challenge the regime. These results support and extend prior findings on the effectiveness of hard propaganda in deterring anti-regime activities to the case of contested and unstable autocracies. But they also suggest that this kind of messaging erodes regime legitimacy providing the first evidence outside of the Chinese case of the pathology of hard propaganda.