Stop Hate Speech

In times of unprecedented digital interaction, we are witnessing a worrying increase in hateful and sexist comments in online media. Our colleague Paolo Rüegg interviewed Sasha Rosenstein, community manager for Stop Hate Speech, a platform aimed at identifying and reducing the number of such comments.

PR- What is Stop Hate Speech?

SR- Stop Hate Speech is an initiative by alliance F, the parent organisation of the Swiss Women’s Association. The project started because there is too much unnoticed and unmentioned sexism in the comment sections of online media. These articles and comments, commonly referred to as hate speech, make it much harder for women to expose themselves. In practice, this could mean women fearing to run for any position with public visibility, such as -for example- a political position. Based on these grounds, Sophie Achermann, the initiator of the project, developed the concept for a civil technology project which we called Stop Hate Speech.

In a nutshell, we are developing an algorithm that finds hate speech online and we have formed a community that responds to it using so-called ‘counterspeech’. As the algorithm is trained to identify what people perceive as hate speech, establishing what this perception is was the first stage of the project.

We created an assessment app – a bit like Tinder – where you could swipe left or right for comments that were potentially hateful. When users found a comment degrading, they could give more information such as which group of people was degraded and which specific words were used. In this way, they fed the perceived degradation to the algorithm, which we called Bot-Dog, which then could search for similar comments on the Internet.

PR- How did you select the group that initially identified hateful online content?

SR- It was a big aspiration of us to have a diverse audience so we reached out to the groups we wanted to represent. For instance, we contacted Pro Senectute [biggest Swiss organisation for the elderly], used the universities’ networks, and promoted Stop Hate Speech in newsletters with less academic audiences. Finally, we asked people to provide their educational level, nationality and sexual orientation in an anonymous manner.

© Visuals from Stop Hate Speech website

PR- What does “confronting hate speech with counterspeech” mean?

SR- Counterspeech is crucial, because the biggest part of hate speech cannot be prosecuted legally. Overall, the laws in Switzerland are very mild. We have laws against racial discrimination and for protecting people against incitement of violence, but sexism and many others are not among them. As a consequence, very little forms of derogatory online content are legally relevant.

  • Calling out hate speech, explicitly saying that a comment is discriminatory or even prosecutable
  • Entering a conversation, by countering with facts or using humour to disqualify their statements
  • Showing support by saying something like: “I think you made bad experiences with […], but you should not generalise this.”

But I always underline the following: it is not about changing the opinion of those people willing to provoke; it is about the silent readers. When people consume content on social media or on news platforms, they often look at the comments below the article or post: if there are only hateful comments below an article, you likely do not want to be a part of this discourse anymore. This is what Stop Hate Speech is trying to prevent.

When an article is posted, our algorithm has learned to generate a potential hate speech risk based on the content and the choice of words in the header and text. This makes it much easier for us to be the first ones to comment. If we manage to comment there before any hateful comments arrive, we can already try to set the tone. This increases the hurdle for people who write hateful comments reducing the overall amount of hate speech.

“We are not a digital police, but we are a grassroots movement that tries to make the online space a bit safer.”

Sasha Rosenstein

PR- How does the collaboration across Switzerland’s different language regions happen?

SR- I work very closely with Morgane Bonvallat who manages the Romandie. We developed the community-platform concept together and we are helping each other to implement Stop Hate Speech in the different regions. Of course, this means that we need to run two different algorithms: this makes similar efforts in countries with one predominant language such as Germany a bit easier. Needless to say, we also want to launch Stop Hate Speech in the Italian speaking part of Switzerland.

PR- How do you keep your community engaged and motivated?

SR- On the one hand through chats where users can communicate with each other. We try to constantly expand our knowledge base, providing information about commonly appearing discrimination topics. But we always stress the fact that – in terms of perceived degradation – our community members are in most cases the experts. We constantly invite our community to share their knowledge, whether it is about islamophobia, transphobia, ableism or anything else.

We also host various online workshops. The idea is to bring together people who have gathered different discrimination experiences in order to broaden the capacity of understanding and responding. Last winter, just before everything shut down again, we had planned a live event at Kosmos about online racism. The workshop was targeted at people who, for instance, went to protest for the BLM movement, but had then no idea how to act in these discourses as a white person. This is very much the approach with our project: sharing knowledge and expanding horizons.

Sarah Ebling

          Sarah Ebling holds a professorship in Accessibility Studies at Zurich University of Applied Sciences (ZHAW) and is a senior researcher at the University of Zurich. Her research focuses on natural language processing in the context of disabilities and special needs, specifically, sign language technology and automatic text simplification. Her groups’ contributions involve artificial intelligence techniques with a strong emphasis on user involvement. She is involved in various international and national projects and leads a large-scale Swiss innovation initiative entitled „Inclusive Information and Communication Technologies“ (2022-2026;