Why people troll on the Internet: Psychology, strategies and the right way to deal with them

An everyday scene in the comment columns under a video, podcast or blog post: A community exchanges arguments on a subject, questions are asked and answered objectively. Suddenly a post appears that completely misses the topic, contains an absurd insinuation or places a subliminal trickery. Within a few hours, the mood has turned throughout the discourse room. Instead of talking about the original thing, dozens of people are highly emotionally debating the provocation.

This phenomenon is an integral part of digital media culture. In order to counter it effectively, no blanket scolding about ‘evil trolls’ helps. It requires an understanding of the interplay between individual motivation, group dynamics and the technical architectures of our platforms.

Central guiding thesis:

Trolling gains power not because the provocation is particularly persuasive, but because it binds attention, shifts rules, and forces others to respond.

1. What is trolling and what is not?

The term troll has become widely used in public debates. It is often used as an argumentative abbreviation to devalue inconvenient criticism, sharp counter-opinions or simply rude behavior. This is problematic because it devalues the real, factual discourse.

A precise definition

It's not just the content of a single post that makes it trolling. Rather, the decisive factor is the recognizable pattern of behaviour: Are answers included and arguments examined – or are they merely a starting point for the next provocation?

For classification, a clean demarcation helps:

• Criticism: Targets content, actions, or arguments. Even if it is presented harshly, unpleasantly or emotionally, a factual reference remains recognizable.
• polemics and satire: Use stylistic means such as aggravation, exaggeration or irony, but represent a clear, comprehensible position in terms of content.
• Insult: A direct, usually affective attack on a person who does not necessarily follow a long-term disruptive strategy.
• Trolling: A recurring communicative behavior pattern in which provocation, confusion or thematic derailment are in the foreground and factual attempts at clarification repeatedly run into the void.
• Hate speech and threats: May be criminally relevant depending on the wording, context and circumstances. Whether a specific statement is punishable depends on the relevant criminal offences and must be assessed on a case-by-case basis.

2. Trolling or misunderstanding? The three-level model

A key problem in practice is: We can't look people in the head. The inner intention behind a comment is not unequivocally recognizable from the outside. A poorly formulated sentence, irony without context or a technical misunderstanding can seem like a provocation at first glance.

In order to avoid premature convictions and to protect legitimate criticism, the evaluation is recommended along three levels:

Plaintext

1. Is the statement only inconvenient or factually wrong? (No trolling) 2. REACTION → How does the person react to a factual clarification? 3. MODEL → Does dodging and provoking repeat itself systematically?

1. The content level: Is the article merely critical, unconventional or steeply thematic? Dissenting opinion, even if ill-founded, does not in itself constitute trolling.
2. The response level: What matters is what happens after an objective answer. Does the person respond to arguments? Does it acknowledge facts or correct misunderstandings? Or does she completely ignore the answer and immediately put the next needlestick?
3. The model level: Trolling rarely appears in the first step, but in the course. Is the behavior permanently destructive? Are new post-war locations constantly being opened up? Are mutual agreements ignored?

Principle: It is not the single provocative sentence that decides whether it is trolling, but the recurring pattern of provocation, reaction and re-escalation.

A case study from practice

How this behavioral dynamic happens in reality can be illustrated by a typical sequence in a comment column:

A commentary appears under a podcast episode: “It is a pity that you treat the subject so superficially and conceal important facts.”

1. The reaction: The moderation team responds objectively, asks for specific points and offers suitable supporting documents.
2. The evasion: The author does not address the facts, but unfoundedly requests more and more specific evidence (Sealioning).
3. The escalation: When these are also delivered, he declares all sources to be biased, takes up a legacy of the moderation team (Derailing) and accuses the editors of suppressing uncomfortable voices.
4. The victim role: After the removal of the non-topic contributions, the author publishes screenshots of the deletion and presents himself as a victim of alleged censorship.

If one analyzes this course with the three-level model, it becomes clear: The first comment was not yet proof of trolling. Only the systematic refusal of a factual clarification, the constant postponement of the requirements and the targeted victim staging reveal the destructive pattern.

3. The Toolbox of Provocation: Typical troll strategies

Those who know the typical patterns of digital interference experiments can classify them faster and lose less energy to them.

📌 info box: How do you recognize strategic trolling?

• Derailing: The actual topic is deliberately replaced by a non-specialist, emotionally charged accusation.Example: In a technical discussion about the audio quality and compression of a podcast format, someone repeatedly writes: “Interesting that you are talking about sound quality, but never about why your moderator deleted a specific comment three years ago.”
• Sealioning: The seemingly polite but endless reclaiming of evidence and explanations with the aim of exhausting the counterpart by sheer perseverance. Sealioning is not available for a single demand, but only when delivered answers are systematically ignored.
• Concern Trolling: Alleged concern is used as a rhetorical camouflage to subtly devalue people, content, or projects.Example: “I only mean well with you, but with this topic you are ruining your reputation.”
• Moving the Goalposts: As soon as a required condition or document source has been delivered, a new, even more difficult requirement is immediately made.Model: "Show me a source." → Source is called. → ‘One source is not enough’ → Three other sources will be provided. → ‘These are all courtesy studies.’
• Dogpiling: A targeted or spontaneous mass attack in which numerous accounts interact simultaneously with a single person or moderation. This can give the impression that the attacking position represents an overwhelming or representative majority opinion – although often only a particularly active and mobilised group becomes visible.
• Sock puppets: A single person uses multiple second or fake profiles to artificially fuel discussions, self-consent, or fake a supposedly broad and independent critique.
• Ironic camouflage (‘just fun’): After a clear crossing of the border, one’s own intention is retroactively ridiculed in order to present the backlash as ‘hypersensitive’ or ‘humorless’.

4. Why people troll: Personality meets situation

In psychological research, trolling is often associated with the personality traits of so-called Dark tetrade (Sadism, narcissism, psychopathy, Machiavellianism). The 2014 study by Buckels et al. shows statistical correlations – especially with regard to so-called everyday sadism, i.e. the tendency to feel pleasure in the frustration or helplessness of others.

However, no hasty conclusions can be drawn from this:

• These are: statistical correlations, Not for compelling reasons.
• An online comment can be No clinical diagnosis or remote diagnosis.
• Not every person who trolls online has a personality disorder – and not every person with such traits shows destructive online behaviour.

The Underestimated Impact of the Situation

In addition to stable personality traits, situational factors can play a significant role. The study by Cheng and colleagues (2017) shows that negative mood and an already destructive discussion environment can promote problematic behavior even in people who had not previously been noticed by trolling behavior:

1. Boredom and the need for attention: A steep provocation generates immediate visible resonance – often much faster and more reliable than an elaborately researched technical article.
2. Group dynamics and follow-up: If the basic tone in a comment column is already aggressive or cynical, the personal inhibition threshold for newcomers drops drastically.
3. Frustration and imitation: Those who feel unfairly treated themselves or observe that others succeed with aggressive methods are more likely to resort to similar means.

5. The accelerator I: The Psychology of Online Inhibition

Why is it so much easier to provoke others online than in a personal conversation? The American psychologist John Suler described this phenomenon in 2004 as the Online disinhibition effect (Online inhibition effect). It can be traced back to six key factors:

• Dissociative anonymity: The feeling that the action on the net is separated from one's own everyday identity (“This is not me, this is just my profile”).
• Invisibility: The other person is not physically present. At the same time, nonverbal signals such as facial expressions, gestures, tone of voice and immediately visible injury are missing.
• Asynchronicity: Communication takes place on a time-delayed basis. One throws a comment into the room and can immediately turn away without having to endure the immediate reaction.
• Solipsistic introjection: Since voice, facial expressions and immediate reactions are missing, an image of the other person is created in one's own head. His statements are associated with an imagined voice and personality, which only partially has to correspond with the real person.
• Dissociative imagination: Some people experience the digital space like a world separate from everyday life or a game. After logging out, what is said there no longer seems to belong to one's own real identity and responsibility.
• Minimization of status and authority: Traditional hierarchies, social statuary features and the authority of moderators are more easily receding into the background in the digital space.

Important differentiation: Online inhibition is not just toxic. She also has an benevolent side (benign disinhibition): It allows people to talk more openly about fears or strokes of fate, seek support and reduce stigma. The digital distance takes inhibitions in both directions.

6. The Accelerator II: Platforms, Interaction and Algorithms

Trolling does not occur in empty space. It encounters technical infrastructures which, through their architecture, favour certain behaviours.

Many social networks use referral systems, the assessment of which can be influenced, inter alia, by interactions such as clicks, comments, likes, redirects or length of stay (cf. Gillespie, 2018; EU Regulation 2022/2065 / DSA, Art. 27). However, which signals are actually crucial and how they are weighted differs from platform to platform.

Plaintext

[Provocative Commentary] ↓ [Many Outraged Responses and Discussions] ↓ [The System Registers High Activity] ↓ [Additional Visibility Can Be Created]

A factual contribution may receive ten responses, while a provocation triggers two hundred responses. Depending on the respective recommendation system, this high activity can generate additional visibility – even if most responses contradict the original statement.

This creates a systemic paradox: Outraged contradictions and factual counterarguments can also technically increase the scope of a provocation.

7. The indignation trap: The Asymmetric Discourse Game

One of the most effective patterns in trolling is the Indignation trap. It exploits the dynamics of emotional reaction and moderation in a targeted manner.

The Mechanics of the Trap:

1. The impulse: A targeted provocation is set.
2. The outrage: Community members react emotionally, invest time and try to refute the statement.
3. The decontextualization: The provocative person picks up a single, sharply formulated backlash and releases it from its original context.
4. The victim role: The author of the provocation now presents himself as a victim of aggressive attacks.
5. The accusation of censorship: If the moderation finally intervenes and removes the disturbing contributions, this is staged as evidence of alleged suppression of opinion.

This game is highly asymmetrical: The trolling account only takes a few seconds for a provocative throw-in, while the opposite side has to invest minutes or hours to limit the damage, explain matters and enforce rules.

8. The consequences for those affected and silent readers

The damage caused by trolling is not limited to those directly involved. It also influences the many silent readers who observe the climate of a community without commenting on it themselves.

When observers experience factual participants being wiped out, provocations tolerated, and forums dominated by perpetual disputes, they draw consequences:

• Withdrawal and self-censorship: Constructive and expert voices discontinue their contributions because the argumentative effort is disproportionate to the yield.
• Postponement of the group standard: The level of aggression is gradually increasing. What used to be considered rude becomes the new standard.
• Erosion of trust: Confidence in fair rules of conversation and the platform's ability to act is declining.
• Binding of moderation capacities: Moderation teams spend their time arranging deletion debates rather than actively shaping the community.

Key findings: A community is shaped not only by the visible contributions, but also by the people who no longer participate because of the prevailing tone.

9. Responding with system: The extended decision matrix

While the classic ‘don’t feed the troll’ rule is correct in its approach, it is too short-lived as a universal solution. Anyone who leaves false information, insults or attacks on others uncommented, leaves the public space to the loudest.

A step-by-step, situational approach is required:

Situation	Meaningful measure	objective
I don't know if it's criticism or trolling.	Provide one-off factual answers and ask specific questions	Checking Good Intention
Obvious provocation without reach	Do not reply; Hide or remove if necessary	Don't pay attention
High visibility misinformation	Correct once-in-a-lifetime factual for fellow readers	Providing orientation for the community
Repetitive derailment (derailing)	Formulating the limit of conversation and moderating other non-topic contributions	Protecting the discourse space
Dogpiling against community members	Limit attacks, protect those affected and temporarily restrict comments	Stop the spiral of escalation
Insult / violation of rules	Consistently delete comments, warn or block users	Enforcing domestic law
Targeted hate speech / threat	Securing evidence (screenshots), reporting and reviewing legal action	Protection and prosecution

Concrete text modules for moderation and community

Here are proven reaction patterns for practice:

• In case of unclear criticism:"Thank you for your feedback. I am happy to answer the specific question on the subject: [Short answer]. Please stay directly with the topic of the article for further contributions.”
• In case of repeated derailment:“This point has already been answered at this point. We will remove further contributions that derail the original topic again, without further discussion, in order to keep the space for the actual exchange.”
• In case of misinformation (address of the participants):‘For all participants on classification: The claim that […] is not substantiated in this form. The available data show instead: […].’
• In the case of personal attacks:“Content criticism and counter-arguments are welcome here at all times. However, personal attacks, insults and reductions are against our netiquette and will be removed.”

10. What Affected People Can Do Specifically

Anyone who becomes the target of trolling, targeted provocations or a dogpile is often under considerable pressure to act. Each new notification creates the feeling of having to react immediately, explain yourself or defend your reputation. However, it is precisely this compulsion to react that often forms part of the dynamics of the attack.

1. Do not respond in the first affect

An emotionally formulated response can later be detached from the context and used against the affected person. Therefore, it makes sense to create a distance first: Turn off notifications, leave the comment section for some time and ask a second person for a sober assessment. This does not mean that you are right about the attack – it simply means determining the timing and form of your own reaction.

2. Distinguish between clarification and justification

A factual clarification can be important for the participants. An endless justification against a person who does not want to accept an answer, on the other hand, consumes above all time and strength. A proven structure is:

‘The claim is false. For the classification of all participants, we put the facts right: […]. We will not go into any further detail on repetitions of the same allegation that has already been answered.”

3. Securing evidence in an orderly manner

Relevant content should be documented in case of repeated attacks, threats, identity abuse or coordinated campaigns. This includes not only individual screenshots, but also, if possible, date, time, username, platform and the full conversation context. An orderly documentation shows whether there is a recurring pattern, an escalation or a targeted campaign.

4. Organize support

Those affected do not have to cope with a major escalation alone. Trustees or team members can view comments, arrange evidence, take on moderation tasks, or help distinguish between legitimate criticism and targeted attacks. Especially with a dogpile, it is relieving not to have to constantly check your own account yourself.

5. Controlling digital access

Blocking, muting, restricting and deleting are not moral defeats. These are technical tools that people use to control access to their communication space. No one is obliged to make a stage permanently available to a person simply because he or she refers to his or her attacks as a ‘debate’.

6. Do not downplay in serious attacks

Concrete threats, reenactments, disclosures of personal data (doxing) or systematic attempts at intimidation no longer belong in the category of ordinary Internet provocation. In such cases, the safety of the data subject shall take precedence over any public discussion.

11. Moderation and censorship: An important distinction

As soon as posts are deleted or accounts are blocked, the word ‘censorship’ quickly falls. Here a clear conceptual distinction is necessary:

• State censorship in the constitutional sense: A preventive state control in which a publication is pre-examined, approved or prevented (cf. Art. 5 (1) sentence 3 GG; BVerfGE 33, 1).
• Private Moderation: Operators and platforms can decide on the basis of their rules which contributions remain in their communication room. This is not to be equated with state pre-censorship. Nevertheless, moderation decisions can be unfair, inconsistent or critical.

Anyone who runs a forum, podcast channel or blog provides a space for exchange. Household law, however, is not a license for arbitrariness: Good moderation should be transparent, comprehensible and applied as uniformly as possible. This is the only way to maintain community trust.

12. Frequently Asked Questions (FAQ)

Is every provocation on the Internet trolling?

No. Even harsh criticism, polemics, misunderstandings and rude comments are not automatically trolling. Decisive is the recurring behavior after a factual clarification.

Should trolls always be ignored?

Not always. For small provocations with no reach, ignoring can make sense. Visible misinformation, attacks on third parties or rule violations, on the other hand, should often be briefly classified or moderated in order to give orientation to the community.

Is deleting comments censorship?

Private moderation within the framework of domestic law is not state censorship. Platform operators are allowed to set rules for their discourse space, but should enforce them transparently and comprehensibly.

Why do people react so strongly to trolling?

Provocations often address moral indignation, a sense of justice, or a desire to immediately correct false claims. This gives the impression that one has to react immediately – a dynamic that can be exploited in a targeted manner in strategic trolling.

Conclusion: Drawing boundaries is not a defeat

Trolling lives on the assumption that any provocation expressed has a right to a content stage. That is not the case.

A healthy digital discourse space does not arise by itself. It requires active care, clear lines and a willingness to set boundaries. The consistent suppression of interference attempts is not a sign of weakness or lack of ability to debate, but the basic prerequisite for appreciative communication on the network to take place at all.

Literature and sources

• Buckels, Erin E.; Trapnell, Paul D.; Paulhus, Delroy L. (2014): Trolls just want to have fun. Personality and Individual Differences, 67, pp. 97-102.
• Federal Constitutional Court (1972): BVerfGE 33, 1 – Broadcasting decision / Concept of censorship for Art. 5 GG.
• Cheng, Justin; Bernstein, Michael; Danescu-Niculescu-Mizil, Cristian; Leskovec, Jure (2017): Anyone Can Become a Troll: Causes of Trolling Behavior in Online Discussions. Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing (CSCW 2017), pp. 517-530.
• European Union (2022): Regulation (EU) 2022/2065 (Digital Services Act / DSA), Art. 27 & 34.
• Gillespie, Tarleton (2018): Custodians of the Internet: Platforms, Content Moderation, and the Hidden Decisions That Shape Social Media. Yale University Press.
• Suler, John (2004): The Online Disinhibition Effect. CyberPsychology & Behavior, 7(3), pp. 321-326.

📚 Further book recommendations

If you want to deal more intensively with trolling, digital aggression, online inhibition and dealing with destructive discussions, you will find well-founded backgrounds and practical strategies in these four German-language books.

---

🔥 1. Hatred on the Net

Ingrid Brodnig

Ingrid Brodnig explains how hate comments, targeted provocations, bullying and false claims are changing digital discussions. Her concrete recommendations for dealing with aggressive rhetoric and escalating comment columns are particularly helpful.

Particularly suitable for: Trolling, hate speech, counter speech and community moderation.

→ View on Amazon

---

🧠 2. Cyberpsychology – Life on the Net

Catarina Katzer

The book describes how the Internet influences our thinking, perception and social behavior. It provides an understandable psychological framework for anonymity, digital identities, online inhibition and changed forms of communication.

Particularly suitable for: Online inhibition, anonymity and psychology of digital communication.

→ View on Amazon

---

🌐 3. Against the Cruelty

Ingrid Brodnig

This book explores how public debates are becoming increasingly aggressive through emotionalization, hate speech, fake news, and rhetorical manipulation. At the same time, it shows how to recognize such strategies and react meaningfully to them.

Particularly suitable for: Outrage traps, manipulation, platform dynamics and debate culture.

→ View on Amazon

---

🛡️ 4. Objection! – Counter conspiracy myths and fake news

Ingrid Brodnig

The focus of this book is on practical counter-talk. It explains how to correct false claims and manipulative argumentation patterns without grinding into endless justification loops.

Particularly suitable for: Moving the goalposts, fact-resistance and factual counter-talk for fellow readers.

→ View on Amazon

---

Note: The marked links may be affiliate links. If a purchase is made through such a link, we may receive a small commission. For you, the price does not change.

Glossary: Terms related to trolling, net culture and moderation

C

• Concern Trolling A rhetorical tactic in which a person protects alleged concern or benevolence (“I only mean well with you …”) to subtly unsettle, devalue or slow down a project, topic or person.

D

• Derailing A disruptive strategy in digital discussions in which the actual core topic is deliberately distracted and displaced by unrelated, highly emotional accusations or past conflicts.
• Dogpiling A spontaneous or deliberately coordinated mass attack in which a large number of accounts simultaneously interact with a single person or moderation. This creates the impression of an overwhelming majority opinion and builds up massive social pressure.
• Doxing (also Doxxing) The malicious research and publication of a person's private, personal data (such as home address, telephone number, employer or clear name) on the Internet in order to intimidate or expose them to real dangers.
• Dark tetrade A model from personality psychology that describes four interrelated characteristics: Narcissism (self-indulgence), Machiavellianism (manipulation), psychopathy (lack of empathy) and everyday sadism (joy at the frustration of others). With regard to trolling, there are statistical correlations here, but no compelling causalities.

E

• Indignation trap An asymmetrical discourse dynamic: A targeted provocation triggers emotional outrage among those involved. The provoked person is then picked out a sharp backlash to portray them as aggressive and push the original provocateur into the victim role.

H

• Domestic law (Virtual domestic law) The right of operators of digital spaces (blogs, forums, social media channels) to set rules for exchange (Netiquette). Enforcing domestic law by deleting or blocking infringing posts is a moderation measure and not state censorship.

M

• Moving the Goalposts An argumentative evasive tactic: As soon as a required condition or document source has been delivered, a new, even more difficult requirement is immediately made, so that no proof is ever accepted as sufficient.

O

• Online disinhibition effect (online disinhibition effect) A term coined by John Suler for the observation that people in the digital space reduce inhibitions faster than in face-to-face communication due to factors such as anonymity, insynchrony, invisibility and the lack of non-verbal signals.

S

• Sealioning A special form of trolling in which a person, under the guise of apparent politeness, endlessly demands evidence, arguments and explanations. The aim is not to clarify facts, but to exhaust the other person by sheer perseverance and to tie them up in time.
• Sock puppet (sock puppet) A second or fake profile used by a person to artificially fake majorities in their own discussions, to prove themselves right, or to criticize from a seemingly neutral perspective.

T

• Trolling A recurring communicative behavior pattern in the net, in which provocation, confusion, thematic derailment or the triggering of emotional reactions are in the foreground and factual clarification attempts systematically run into the void.