Table of Contents
X (Twitter) and Teen Radicalization: What the Algorithm Does and What Parents Can Do
How X's recommendation algorithm works, what the research shows about teen radicalization pathways, and how parents can have productive conversations about it.
Your 15-year-old son started spending more time on X (formerly Twitter) six months ago. You’ve noticed his language has shifted—more cynical, angrier, more dismissive of people who disagree with him. He cites accounts you’ve never heard of. He seems less interested in friends from school and more interested in a community of people online who share increasingly specific grievances. You’re not sure when this happened or how fast it’s moving.
Key Takeaways
- X’s algorithm is explicitly designed to maximize engagement, and anger and outrage are among the highest-engagement emotions—creating conditions that systematically amplify extreme content.
- Research on algorithmic radicalization documents a “rabbit hole” effect: users who engage with moderately partisan content are subsequently shown more extreme content in the same direction.
- Teen boys are disproportionately represented in online radicalization research, particularly for right-wing extremist content, but research documents radicalization pathways across political ideologies.
- Parental conversations about how algorithms work are significantly more effective at building resistance than content-based restrictions alone.
- X’s post-2022 moderation changes reduced enforcement against extremist content, which the research community has documented as worsening the radicalization pathway.
How X’s Algorithm Works
X uses a recommendation system called “For You” that surfaces content from accounts users don’t follow, prioritizing content it predicts users will engage with. After Elon Musk’s 2022 acquisition, X transitioned from third-party algorithm developers to building its own recommendation system, which was open-sourced in 2023.
The system ranks content based on: predicted probability of engagement (likes, replies, reposts); weighted by the type of engagement (replies are weighted more than likes, suggesting conversation); and filtered by a set of moderation rules that, since 2022, have been applied less strictly.
The critical insight about radicalization: outrage is one of the most reliably engagement-generating emotions. A tweet that makes you angry, disgusted, or righteously indignant generates more replies, more engagement, and longer dwell time than content that makes you feel neutral or mildly positive. The algorithm learns this and surfaces more of it—not because it’s promoting extremism as a goal, but because outrage drives engagement.
The Radicalization Research: What We Know
The academic research on algorithmic radicalization has produced consistent but contested findings:
The “rabbit hole” finding: Research by Ribeiro et al. (2020) in Proceedings of the Web Conference documented that on YouTube (where the effect is more extensively studied), users who watch moderately conservative content are subsequently recommended increasingly extreme right-wing content by the recommendation algorithm. This pattern appears on X as well.
The “alternative influence network”: Research by Rebecca Lewis at Data & Society documented an interconnected network of YouTube and social media creators who form a pipeline from mainstream content to increasingly extreme viewpoints—each step seeming only slightly more radical than the last.
The Pew Research finding on X specifically: A 2023 Pew study found that among X users, the most active users see dramatically more toxic content than average users—and that the algorithm differentially surfaces this content to engaged users.
Teen boys and right-wing radicalization: Multiple research documents, including work from the Institute for Strategic Dialogue and the Global Network on Extremism and Technology, have found that teenage boys represent a disproportionate share of those recruited into online extremist communities. The pathways typically involve: initial engagement with humor or gaming content; exposure to “edgy” humor that normalizes extreme views; gradual normalization of more explicitly ideological content; and community formation around shared ideology.
| Radicalization Stage | Content Characteristics | What Algorithm Does |
|---|---|---|
| Entry | Gaming, sports, humor, pop culture | Recommends similar content |
| Engagement | ”Edgy” criticism of mainstream views | Recommends more provocative versions |
| Deepening | Explicitly ideological content | Recommends more extreme versions |
| Community | Identity formation around ideology | Recommends community accounts |
What Changed After Musk’s Acquisition
X’s post-2022 moderation changes have been documented to:
- Reinstate many accounts previously banned for policy violations, including accounts associated with extremist content
- Reduce the size of trust and safety teams
- Remove content labels that previously flagged disputed or manipulative content
- Reduce enforcement of rules against coordinated inauthentic behavior
A 2023 report from the Center for Countering Digital Hate found that hate speech on X increased significantly after these changes. The NewsGuard “Nutrient Pollution” report found X’s algorithm was pushing more misinformation than other platforms.
For parents: X in 2026 is meaningfully different—more permissive and less moderated—than Twitter was before 2022. The baseline radicalization risk from the algorithm is higher.
What Parents Can Do
Have the algorithm conversation. The single most effective intervention documented in media literacy research is helping teenagers understand how recommendation algorithms work—that they’re not showing them a neutral sample of the internet, but a curated selection optimized for engagement. “Your algorithm is showing you content to make you angry because angry people keep scrolling” is a genuinely useful frame.
Ask about specific accounts, not platforms. “Who do you follow on X?” and “Who do they recommend?” is more informative than “Do you use X?” The account ecosystem tells you more than the platform itself.
Look for ideological drift, not just offensive content. Radicalization doesn’t start with offensive content—it starts with gradual normalization. Warning signs include: increasing cynicism about all institutions, contempt for people who hold different views, a strong in-group/out-group dynamic with online communities, and the feeling that “everyone is against us.”
Use curiosity, not alarm. “That’s an interesting argument—where did you first encounter that?” opens more conversation than “That’s dangerous, you can’t believe that.” Your goal is to maintain conversational access to your teenager’s evolving beliefs so you can engage with them, not drive them underground.
For younger teens: Consider whether X is appropriate at all for users under 16. The platform’s minimum age is 13, but the radicalization research documents fastest movement in the 13–17 age range. Many parents of 13–14 year olds reasonably choose to delay X access.
What to Watch For Over 3 Months
- Has your teen’s language become more extreme or contemptuous in ways that track with X content?
- Are they spending time in communities with strong in-group language and “enemy” framing?
- Do they dismiss all mainstream sources as corrupt or biased while accepting fringe sources uncritically?
- Are they isolating from friends who don’t share their online community’s views?
- Are they being asked to recruit others, buy things, or take real-world actions by online communities?
Frequently Asked Questions
My teen says I’m censoring him if I express concern about what he reads online. How do I respond?
Engage with the concern directly. “I’m not saying you can’t read it—I’m saying that the algorithm is showing you this content because it keeps you scrolling, not because it’s true. What would you say to someone arguing the opposite?” You’re modeling critical analysis, not censorship.
What’s the difference between radicalization and just having strong opinions?
Strong opinions that emerge from thinking about different perspectives are healthy. Radicalization warning signs are specific: contempt rather than disagreement for those with different views; belief that violence is justified to achieve political goals; dehumanization of out-groups; conspiracy thinking that doesn’t respond to evidence; and community formation that requires severing outside relationships.
My teen follows a lot of accounts that seem “edgy” but not explicitly extremist. Should I be worried?
“Edgy” humor communities are documented entry points to radicalization pathways, not destinations in themselves. The question is trajectory: is the content getting gradually more extreme over time? Is it normalizing contempt for specific groups? Are the “jokes” shifting from self-aware irony to sincere ideology? These are the signals to watch.
Does this only apply to right-wing radicalization?
No. Radicalization pathways exist across the political spectrum, including left-wing extremism and various single-issue movements. However, the documented social media radicalization research has found right-wing content pathways more extensively, which may reflect both measurement choices and real differences in the volume of content produced. The algorithmic mechanics are the same regardless of direction.
Sources
- Ribeiro, M. H., Ottoni, R., West, R., Almeida, V. A., & Meira, W. (2020). Auditing radicalization pathways on YouTube. Proceedings of the Web Conference 2020.
- Lewis, R. (2018). Alternative influence: Broadcasting the reactionary right on YouTube. Data & Society.
- Pew Research Center. (2023). The biggest news stories of 2022: A look at reactions on Twitter. Pew Research Center.
- Center for Countering Digital Hate. (2023). Toxic Twitter. CCDH.
- Global Network on Extremism and Technology. (2023). Algorithmic amplification of extremist content. GNET.
- Rauch, J. (2021). The constitution of knowledge. Brookings Institution Press.
Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.
