Dr. Agnieszka Marzęda from the University of Warsaw’s Faculty of Journalism, Information and Book Studies has examined how trolls operate on social media and the different strategies they use to disrupt online conversations.
She analyzed content from platforms including Facebook, Twitter/X, Instagram, and YouTube. Her work resulted in a classification of trolling, showing that it is not a single, simple phenomenon, but a whole set of practices ranging from jokes and provocation to manipulation, disinformation, and ideological activity.
Dr. Marzęda presents her findings in the article “Diversity and complexity of online trolling: An extended classification and analysis of social consequences,” published in the journal Media Biznes Kultura. Her interest in the topic was sparked by a conference on cyberbullying.
“I listened to various presentations and came to the conclusion that everyone sees it a bit differently. The terms that are often used are not always clearly defined. I felt something was missing,” the researcher explains.
The dark side of the internet
Before we turn to the different forms of trolling described by the author, it is worth clarifying what trolling is not. Although it is often lumped together with hate speech or cyberbullying, it follows a slightly different logic.
Hate speech is direct: it insults and demeans, driven by hostility, prejudice, or the desire to attack another person.
Cyberbullying refers to online harassment, usually targeted at a specific individual and often repeated over time.
Trolling works differently. Its aim is manipulation and provocation: introducing controversial, absurd, or false content into a conversation in order to trigger emotions, pit participants against each other, or take control of the discussion.
It does not always target one person. More often, trolls cast multiple “baited hooks” and wait to see who reacts. The fishing metaphor is no accident here. One popular explanation of the term traces it back to “trolling for fish,” a fishing technique in which bait is dragged behind a boat. Combined with the figure of the troll from Norse mythology – a malicious creature lurking to harm humans – it is easy to see why the term stuck so well in internet culture.
Puppets and flames
Trolling rarely exists in a vacuum. It thrives in environments where identities can be easily hidden, roles quickly switched, and a single voice can appear to be many. That is why it fits so well within social media, where a comment can be posted in seconds and a fake account created in under a minute. It feeds on anonymity.
“We fight for freedom of speech, but it is not accompanied by civic courage. People express themselves online in a certain way, knowing they cannot be identified. On top of that, there is sockpuppeting, meaning the creation of multiple fake accounts. Some politicians even use this to support their own arguments in debates and wage political battles. As long as anonymity is tolerated, we are not even beginning to fight trolling. Unfortunately, we have accepted it as part of online culture. That is a mistake. I am in favor of full transparency: if you have something to say, say it under your own name,” stresses Dr. Marzęda.
This is crucial because anonymity and the ease of creating fake accounts enable the proliferation of trolling strategies. In Dr. Marzęda’s classification, the phenomenon clearly unfolds on two levels.
The first consists of broad categories: interaction-based and social behavior trolling, disinformation and manipulation-based trolling, ideological and cultural trolling, and thematic trolling.
The second level includes more specific variants, such as destructive, provocative, humorous, narcissistic, political, health-related, educational, or technology-related trolling.
This becomes easiest to see in the first group – interaction-based and social behavior trolling. It includes actions that directly disrupt conversations, provoke users, or attempt to draw attention to the troll themselves.
One of its destructive forms is the well-known phenomenon of spam. Imagine following an interesting discussion only to be forced to wade through a flood of random characters, repetitive meaningless posts, or images entirely unrelated to the topic. Annoying? Yes. Frustrating? Absolutely. And the troll has achieved exactly what they intended: they have disrupted the flow of conversation.
Another variant is provocative trolling, including “flaming.” The goal is to “set fire” to a discussion and escalate emotions. Statements such as “The pandemic is a hoax” or “Vaccines are harmful” may appear on health forums and beyond. Among medical professionals, they often provoke strong irritation or even anger. And once again, the troll gets exactly what they came for. The problem is that when such provocation concerns health, its consequences do not end in online arguments. False beliefs can spread further and influence the decisions of people seeking information or help online. And wrong decisions can have serious health consequences.
“This may be stirring the pot, but I think trolling does not have to be purely negative, although it is usually perceived that way. While hate speech is always harmful, trolling can also take neutral forms and even serve an entertaining function. I am thinking, for example, of memes, which I personally enjoy,” notes Dr. Marzęda.
Another humorous form includes giving absurd advice on forums (for instance, on cooking: “the best way to cook pasta is to throw it onto the roof and wait for rain”) or initiating rhyming challenges. If this draws attention, the troll has achieved their goal.
Trolls in the age of AI
As technology evolves, so does trolling – especially its disinformation-driven forms, which can be significantly amplified by artificial intelligence.
It is no longer just about false bomb threats or fabricated celebrity gossip. Fake news, made-up stories, and manipulated images or videos are now part of the landscape. We are witnessing a whole new range of possibilities opening up for trolls. AI is a powerful tool for creating deepfakes – images and videos that can appear entirely authentic.
“In theory, we are aware of how artificial intelligence works, and yet we still get fooled. One solution could be systemic tools for marking the credibility of content, for example in web browsers. Some companies are already developing software that assesses the level of truthfulness of a given material. A helpful approach could be a color scale: red for false, yellow for questionable credibility, and green for true. But will the tech giants, who profit from engagement driven by controversy, actually implement this? It’s doubtful. And without that, nothing will change,” predicts the researcher.
What we do to our own minds
“It is crucial that we understand how broad the spectrum of trolling really is. Only then can we learn to recognize it. That is difficult, because it can be extremely subtle and nuanced, and it often hides behind seemingly spontaneous, natural expressions. My research is only a starting point for further discussion,” explains Dr. Marzęda.
This discussion is not limited to internet scholars. The classification of trolling should also be used by media literacy educators, communication and PR professionals, forum and platform moderators, institutions dealing with disinformation, public communication, and reputation management.
Because if we cannot recognize which type of trolling we are dealing with, it becomes much harder to respond appropriately. Hate speech is not moderated in the same way as absurd jokes, spam requires different tools than false health advice, and political provocation cannot simply be dismissed as “rude comments.”
The stakes go far beyond comment sections. As Dr. Marzęda points out, trolling can lead to stress, anxiety, reduced self-esteem, withdrawal from online activity, and on a broader scale to social divisions, erosion of trust, increasing polarization, poorer quality of public debate, and the spread of disinformation.
That is why this typology is not just an academic catalogue of online behavior. It helps us name the mechanisms that shape our emotions, relationships, and the way we participate in the digital world.
“This is an important study because it answers the question: what are we allowing to happen to our own minds,” concludes Dr. Agnieszka Marzęda.
The text was originally published in Polish on the Serwis Naukowy UW website on June 8, 2026.
