Algorithms Shift Polarization. Why Does Policy Still Miss the Real Problem?

A landmark study provides some of the clearest evidence to date that platform architecture can drive political hostility. But policy responses reveals how wide the gap remains between what research shows and what gets addressed, writes Matthias J. Becker, the AddressHate research scholar at NYU’s Center for the Study of Antisemitism, a postdoctoral researcher at the University of Cambridge, and lead of the “Decoding Antisemitism” project.

A breakthrough study published in Science in November provided some of the clearest causal evidence to date that social media algorithms shape political polarization. Researchers built a browser extension that independently reranked users’ X feeds in real time, demonstrating that algorithmic choices directly altered users’ political attitudes by amounts comparable to several years' worth of polarization change in long-term US surveys.

For researchers who study online polarization and radicalization, the findings confirmed what extensive observational evidence had long suggested: platform architecture—what gets amplified, how content is ranked, which posts surface first—can directly shape democratic outcomes. For the first time, independent researchers demonstrated algorithmic causation without requiring access to platform infrastructure—a milestone in platform accountability research.

Yet the policy conversation remains largely elsewhere.

Around the same time the Science study appeared, lawmakers from both parties welcomed X's new location feature, which displays the country from which an account operates. Rep. Don Bacon (R-NE) said the feature revealed how "foreign interests are trying to spread" antisemitic ideas in the United States, while Sen. James Lankford (R-OK) told Jewish Insider that "foreign adversaries have spent years flooding social media with hate-filled and antisemitic propaganda to divide Americans.” The contrast is instructive: research identifies mechanisms; policy focuses on symptoms.

This pattern repeats across digital governance. Policy attention gravitates toward solutions that fit existing narratives about external threats, require minimal structural change, and avoid examination of the architectures that enable affective polarization and radicalization. The result is a widening gap between what we know works and what we're willing to demand.

What the research shows

The Science study, led by researchers Tiziano Piccardi, Martin Saveski, and Chenyan Jia, represents both a methodological and substantive breakthrough. Working without platform permission, they built infrastructure to intercept and rerank X feeds in real time, using large language models to identify content expressing "antidemocratic attitudes and partisan animosity" (AAPA)—including partisan hostility, support for undemocratic practices, opposition to bipartisanship, and biased evaluation of facts.

1,256 US participants received either increased or decreased exposure to AAPA content for one week during the 2024 presidential campaign. Reducing AAPA exposure led participants to feel warmer toward the opposing party by 2.11 degrees on a 100-point scale. Increasing exposure caused a symmetrical cooling of 2.48 degrees. These are not huge shifts for any single individual, but applied to millions of users over months or years, they translate into substantial structural tilts toward hostility.

Critically, 74% of participants reported noticing no impact on their experience. Algorithmic effects operate below conscious awareness, shaping attitudes without users recognizing the intervention.

The study demonstrates what years of correlational research could not: that algorithmic ranking decisions directly shape political attitudes. The content remained the same. The users remained the same. Only the ranking changed—and polarization moved accordingly.

Three drivers—and why only one gets policy attention

To understand why this evidence matters—and why policy responses miss the mechanism—we need a framework for how phenomena such as antisemitism and antidemocratic hostility actually spread online.

My research, analyzing over 300,000 web user comments across major news channels using qualitative discourse analysis supported by computational methods, reveals three distinct but interconnected drivers:

1. Malicious actors. Foreign and domestic campaigns deliberately inject polarizing content. This is the driver dominating political rhetoric and policy attention.
2. Algorithmic amplification. Platform ranking systems, optimized for engagement, disproportionately surface emotionally intense content. Outrage drives clicks. Hostility generates shares. The Science study provides causal evidence for this mechanism.
3. Communication conditions. The socio-technical environment of online spaces—anonymity that removes social accountability, mutual reinforcement that validates extreme positions, omnipresent hate speech that normalizes extremity—fundamentally shapes how users behave and what content they produce.

The Science study isolates the second driver and proves it matters causally. Yet policy interventions typically address only the first—and often in ways that miss even that mechanism.

Why geographic solutions miss all three mechanisms

The X location feature exemplifies how policy attention focuses on one driver while ignoring the others entirely. The feature addresses exclusively malicious actors, specifically foreign ones, assuming that knowing where an account is located helps identify bad actors and discount their content accordingly.

This assumption is flawed on multiple levels. Geographic origin tells us nothing about intent, coordination, or impact. An account in Eastern Europe may be an ordinary person with opinions about American politics. An account in Iowa may be part of a domestic influence operation. Location and malicious intent are orthogonal categories.

More fundamentally, focusing on foreign accounts ignores where most high-engagement polarizing content actually originates. Research by the Decoding Antisemitism project found that antisemitic discourse surged to 36-38% of comments on major UK news outlet YouTube channels following the October 7 attacks (see also DA’s Discourse Report 6). After the antisemitic hate crime in Washington in May 2025, such content averaged 43% across major English-language news channels, reaching 66% on some outlets."

The idiom, references, and interaction patterns indicate domestic participation. High-profile domestic actors—politicians, media figures, influencers—routinely traffic in strategically ambiguous statements that seed problematic interpretations. Digital intermediaries sharpen and amplify these messages. Comment sections collapse the ambiguity into explicit hate speech.

Foreign interference exists and deserves scrutiny—but in the datasets we examined, domestic dynamics overwhelmingly drove the volume and intensity of antisemitic expression. Even if the location feature perfectly identified every foreign account, it would leave the dominant source of polarizing content completely untouched.

What lies beyond the Science study's design

The Science research provides strong causal evidence demonstrating that changing how content is ranked changes political attitudes, even when the content itself remains constant. This is precisely the kind of mechanistic evidence that should drive policy—and represents a major methodological achievement in studying algorithmic effects without platform cooperation.

The study's design also clarifies what questions remain open for complementary research: how communication conditions shape the content that algorithms then amplify. The AAPA posts that participants encountered were produced by users operating within social media platforms where specific communication conditions prevail—including anonymity that can enable transgressive expression, mutual reinforcement that validates extreme positions, and normalized hostility that makes extremity seem ordinary.

Communication conditions contribute to the production of radicalized content. Algorithms amplify it. This material reaches users who are further radicalized by those same conditions, producing more extreme content in a self-reinforcing cycle. The Science study isolates one mechanism in a much larger system, but amplification does not operate in isolation.

What evidence-based accountability would require

If policymakers took the Science findings seriously, they would demand fundamentally different interventions—primarily around transparency and independent research access.

The Science researchers had to build their own infrastructure to study algorithmic effects without platform cooperation—a remarkable methodological achievement, but one that shouldn't be necessary. The European Union's Digital Services Act already establishes precedent for requiring platforms to provide researcher access and conduct systemic risk assessments of their algorithmic systems. Similar frameworks could enable independent auditing of how ranking decisions shape democratic outcomes, without prescribing specific algorithmic outputs.

Understanding communication conditions requires systematic, multimodal analysis of online hostility—its lexicon, metaphors, frames, and escalation patterns over time. These patterns must then be connected to specific platform affordances that amplify conflict and facilitate mutual reinforcement. This empirical foundation would inform platform design choices around friction mechanisms and context provision. And malicious actor detection should focus on behavioral patterns—coordination, inauthentic amplification, systematic manipulation—rather than geography, since domestic actors often matter more than foreign ones in driving high-engagement polarizing content.

What unites these interventions is the need for transparency that enables independent research. We cannot regulate what we cannot measure, and we cannot measure what platforms keep opaque.

Why the gap persists, and how to close it

The location feature requires almost nothing from platforms: add a label, done. No change to business models, no transparency into proprietary systems, no examination of how design enables radicalization. Real accountability would require transparency into recommendation systems that threatens competitive advantage, independent researcher access that creates legal risks, and redesigning communication affordances that requires significant engineering resources. The contrast between what is easy for platforms and what is necessary for democracy explains the persistent gulf between research and regulation.

We now have causal evidence that algorithmic amplification matters. We also have extensive empirical evidence that communication conditions—anonymity that removes social accountability (Suler, 2004; Rösner & Krämer, 2016), mutual reinforcement that validates extreme positions (Bail et al., 2018), and omnipresent hate speech that normalizes extremity (Bilewicz & Soral, 2020)—radicalize ordinary users who produce the content that algorithms amplify. And we have clear data showing that domestic actors and participatory dynamics drive most high-engagement polarizing content.

Against this evidence, focusing on account geography represents a fundamental misunderstanding of how online radicalization operates. Online antisemitism and other forms of antidemocratic hostility are not primarily imported threats. They are produced through the interaction of algorithms that reward polarization, communication conditions that enable radicalization, and malicious actors—domestic and foreign—who exploit these dynamics.

If we're serious about addressing online polarization and radicalization, we need interventions aimed at mechanisms, not geography. We need transparency into algorithmic systems and research into how platform affordances shape political and social outcomes. The Science study demonstrated what is possible when research operates independently of platform gatekeeping. The policy response demonstrates how far we remain from demanding the accountability that evidence requires.