
What’s the best way to respond to hateful comments on the internet? A study has now tested three possible counter-strategies on Twitter: a humorous reaction, a reference to possible consequences and a comment that arouses empathy for those affected. The result: While humor and warning of consequences are mostly ineffective, encouraging empathy can actually lead to the author of the hate comment deleting it and posting less hate speech in the following weeks. The findings can help counteract hate speech on the Internet more effectively in the future.
Hateful comments, known as hate speech, have become a global problem thanks to social media. Victims are often sexual, ethical, or religious minorities. Studies have shown that hate speech is harmful to the mental and physical health of those affected, promotes violence and polarizes public opinion, which harms democratic discourse. Platforms like Facebook and Twitter try to delete hate comments, but only recognize and remove a small part. People and organizations that campaign against hate speech on the Internet try instead with targeted counter-speech, the so-called counterspeech.
Algorithm detects hate comments
“But although counterspeech is increasingly being used, there have not yet been any scientific studies on the effectiveness of various strategies,” writes a team led by Dominik Hangartner from ETH Zurich. His research team therefore carried out a field experiment to investigate which messages are suitable to induce the authors of hate speech to post fewer hate comments in the future. To do this, the researchers first developed an algorithm that was able to find hateful comments through machine learning. With the help of this algorithm, they identified 1,350 English-speaking Twitter users who regularly expressed themselves racist or xenophobic.
The researchers assigned some of these people to a control group that received no intervention. In the case of the others, they responded to a hate tweet with one of three frequently used counterspeech strategies: humor, also in the form of memes, warnings about possible consequences or messages intended to arouse empathy for those affected. “Humor and memes are supposed to change and de-escalate the dynamics of communication,” explain the authors. “The warning of the consequences reminds the sender of the hate speech that family and acquaintances can also see his public messages. Empathy aims to see the victim as a person and to remind the sender that people can be hurt by their behavior. “
Empathy works best
The researchers posted the randomly assigned counter tweets with the help of a bot within 24 hours of the original hate comment. Over the next four weeks, they watched how many more hateful comments the writer posted, the general tone of his utterances, and whether he deleted the original hate tweet.
The result: only responses that generate empathy with the people affected by the hate speech are suitable to induce the hate writers to change their behavior. Such a comment could, for example, read: “Using such language is simply unnecessarily hurtful to immigrants.” After such a message, the likelihood that the commenter deleted his hate tweet increased by 8.4 percentage points compared to the control group and over the next four weeks, the “empathy-treated” authors posted an average of 1.3 fewer xenophobic tweets. Humorous counterattacks or hints such as that the sender’s family and friends can see what he is posting, on the other hand, had no clear effect.
Project against hate speech
“We certainly haven’t found a general-purpose remedy for hate speech on the Internet, but we have found important information about which strategies might work and which don’t,” says Hangartner. “While a focus on empathy appears to be a promising direction for designing future counterspeech strategies, our results also dampen expectations of the effectiveness of some of the most common counterspeech interventions.” Future studies may clarify whether certain empathy-based formulations are more effective than others and what effect repeated interventions have.
The researchers are working with the Swiss women’s association alliance F, which has initiated a project against hate speech. This cooperation between science and practice is intended to help ensure that the research results are realistic and quickly find practical application. “The research results make me very optimistic,” says Sophie Achermann, managing director of alliance F and co-initiator of the Stop Hate Speech project. “For the first time we have results that can empirically prove the effectiveness of counterspeech by means of an experiment carried out under real conditions.”
Source: Dominik Hangartner (ETH Zurich) et al., Proceedings of the National Academy of Sciences, doi: 10.1073 / pnas.2116310118