Algorithms are no longer a background technical issue
Sitra’s March 2026 intervention on the influence of social-media algorithms over young Europeans’ worldviews addresses a democratic problem that is often discussed too narrowly. The question is not simply whether teenagers and young adults spend too much time on TikTok, Instagram, YouTube or other large platforms. It is who decides which political claims, cultural conflicts, news sources and social identities become visible—and according to which incentives.
For many younger Europeans, the algorithmic feed is not an additional channel beside television, newspapers, schools and local civic life. It is the layer through which those institutions are discovered, judged or ignored. A local newspaper’s investigation may never appear in a young person’s feed, while a short, emotionally charged video about the same subject can receive repeated distribution. This does not mean young people are passive or incapable of critical thought. It means the conditions under which they exercise judgement are shaped by private ranking systems designed primarily to maximise attention, retention and advertising value.
Sitra’s seven recommendations therefore matter beyond a debate about online safety. They point towards a basic European question: can democratic societies retain meaningful public debate when the infrastructure for discovering information is largely governed by opaque commercial systems?
Why recommendation algorithms can shape political outlooks
Recommendation systems decide what to place next: another video, a suggested account, a trending topic, a group, an advertisement or an item in a personalised news feed. They use signals such as viewing time, likes, shares, comments, follows, location, language and the behaviour of similar users. The resulting feed can feel personal and neutral because it responds to individual choices. Yet it also actively structures future choices.
Engagement is not the same as public value
Content that creates strong reactions is often easier to measure than content that improves understanding. Anger, fear, outrage, tribal identity and conflict can produce comments and repeat viewing. Careful reporting, uncertainty, nuance and policy explanation frequently demand more attention while generating less immediate interaction.
This does not require an algorithm to be explicitly designed to promote extremist or misleading material. A system optimised for engagement can nevertheless amplify polarising narratives because they are effective at holding attention. The danger is cumulative. Repeated exposure can make a particular frame seem dominant, normal or universally accepted, even where it represents a loud minority.
For young citizens who are still developing political habits and media routines, that repetition may influence what issues seem urgent, whom they view as credible and whether democratic institutions appear capable of solving problems.
Personalisation can reduce shared reference points
European democracy depends on disagreement, but it also needs some common factual ground. When different people receive radically different streams of information, they may not merely reach different conclusions; they may begin from incompatible versions of what happened.
This fragmentation is especially consequential across Europe’s multilingual and national media landscapes. A recommendation engine can connect people across borders, which is valuable. But it can also elevate simplistic, decontextualised claims faster than reporting that explains the legal, historical or local reality behind them. A debate about migration, climate policy, agriculture, defence or elections may be reshaped by content produced outside the affected country, with little accountability to its public-interest standards.
The European alternative is not censorship or algorithmic business as usual
A credible European response should avoid a false choice between unrestricted platform power and state control of speech. Removing illegal content is necessary, but it is not enough. Much harmful democratic impact comes from lawful content that is misleading, manipulative, relentlessly repetitive or rewarded by design.
Likewise, demanding that platforms simply “be transparent” is insufficient if disclosure comes as dense technical reports that ordinary users, journalists and regulators cannot act on. Transparency must lead to contestability: people need a real ability to understand, challenge and alter how information reaches them.
The European Union already has relevant tools, including the Digital Services Act, which establishes duties for very large online platforms around systemic risks, advertising transparency, researcher access and recommender-system choices. Sitra’s focus is a reminder that enforcement and practical usability now matter as much as adopting rules. Rights that exist only in legal text will not change a feed designed by default to keep users scrolling.
The reported recommendations should be read as a call for a package of changes rather than a single technical fix. The most useful benchmark is whether an ordinary young user can make informed choices without needing to be a data scientist or digital-rights lawyer.
Give users meaningful control over their feeds
Platforms should offer genuinely accessible alternatives to profiling-based recommendations, including chronological or subscription-led feeds. These options must be prominent, persistent and easy to select—not buried in a settings menu or reset after an app update.
Users should also be able to see plain-language explanations of why a particular post is recommended: for example, whether it is based on their viewing behaviour, a followed account, local popularity or paid promotion. That explanation will not reveal every model parameter, but it can expose the logic of distribution and help users recognise when a feed is steering them into a narrow loop.
Researchers with strong privacy and security safeguards need reliable access to data that can reveal systemic patterns: how civic information travels, whether certain groups are disproportionately targeted, and whether design changes increase harmful exposure. Independent audits should assess not just the removal of illegal posts, but the effects of recommendation, autoplay, notifications and frictionless sharing.
This is essential because platforms possess the most complete evidence about their systems. Without external scrutiny, the public is asked to trust claims that cannot be properly tested.
Europe should not assume that a better setting toggle will solve a weakened information ecosystem. Local journalism, public-service media, fact-checkers, educators and civil-society organisations need sustainable ways to reach younger audiences without adopting the same outrage-driven format that platforms reward.
That can mean funding media literacy, supporting local reporting, building interoperable public-interest digital spaces and ensuring that trustworthy sources can be found without paying for visibility. The goal is not to give governments authority over “approved” views. It is to make evidence-based, accountable information competitive in the spaces where attention is actually allocated.
Actions young users, families and educators can take now
Regulation moves slowly, but people do not have to wait for a new law or platform redesign.
First, young users can deliberately diversify their information inputs. Follow local and national news organisations, specialist reporters, public institutions and people who disagree respectfully. Search for an issue directly instead of relying only on the next recommended clip.
Second, review feed and advertising settings at regular intervals. Disable unnecessary notifications, limit autoplay where possible and choose chronological or following-based feeds when available. These small changes reduce the platform’s ability to turn every idle moment into another recommendation cycle.
Third, before sharing emotionally intense political content, check the original source, date, country and evidence. A clip may be authentic while being used to support a false conclusion. Asking “what is missing from this video?” is often more useful than asking only whether it is real.
For teachers and youth workers, media literacy should include platform design, not just fact-checking. Students need to understand how ranking, influencers, monetisation and targeted advertising affect visibility. For policymakers and platform professionals, the lesson is equally direct: democratic resilience cannot be delegated to individual self-discipline while interface design rewards impulsive consumption.
FAQ
Not necessarily. Most recommendation systems are designed to optimise measurable business objectives such as engagement, watch time or retention. However, those objectives can unintentionally favour divisive or emotionally intense material, creating polarising effects even without an explicit political intention.
Does choosing a chronological feed solve the problem?
It helps because it reduces personalised ranking, but it is not a complete solution. Trends, influencers, advertising, search results and the accounts a user already follows still shape exposure. A chronological feed works best alongside diverse sources and deliberate verification habits.
What can the EU do beyond removing illegal posts?
The EU can enforce obligations concerning systemic risks, recommender-system choices, advertising transparency and data access for vetted researchers. It can also support independent media, digital literacy and public-interest online infrastructure, while protecting freedom of expression and privacy.
Why is this issue particularly important for young Europeans?
Young adults often receive political information through social platforms first, while forming civic habits and identities. If their discovery environment repeatedly rewards sensationalism over context, it can weaken informed participation, trust and the ability to discuss disagreement constructively.
Fuente: sitra.fi — Tue, 10 Mar 2026 07:00:00 GMT
