The many challenges of tracking hate

To develop effective responses and to mitigate the impact of hate speech on our societies, we need to monitor and analyse this phenomenon. We need to better understand the dynamics of hate speech, including the motivations behind its spread. This is particularly important in a context of crisis and fear.

But the lack of a universal definition of hate speech makes it hard to study this phenomenon on a global scale. So do the varied social, cultural and historical contexts associated with hate speech. Further difficulties include the diversity of national laws, specific local and linguistic subtleties, and the variety of online communities and forms of online hate speech, to name just a few. Another challenge is that victims may not consistently report hate speech to relevant authorities, for various reasons starting from a lack of awareness. Researching online hate speech is also complicated by the non-transparency of proprietary algorithms and lack of access to data held by private companies.

In the digital world, Internet companies must be able to counter the spread of online hatred on their platforms, to comply with international human rights laws. Although social media companies have started reporting the amount of flagged and/or removed content on their platforms, their efficiency has been called into question. Their reliance on automated detection (rather than human intervention) to identify and remove hate speech is prone to error and may infringe on users’ freedom of expression. This highlights the need for more transparency about how hate speech policies are implemented and evaluated.

The United Nations Strategy and Plan of Action on Hate Speech identifies priority areas for combating hate speech and several UN entities are doing research to recognize, monitor and analyse hate speech trends.