MSNBC Host Says X Is HELPING Liberals

Smartphone showing social media apps with text background. — Photo: Shutterstock

The core misunderstanding in debates about Musk-era X is simple but consequential: visibility is not the same thing as persuasion, and neither guarantees electoral payoff. What we can say with confidence is that X’s recommendation system makes right-leaning political content more visible to many users, including those who do not seek it out; what we cannot say—because the best evidence cuts the other way—is that this lopsided exposure somehow advantages liberals by turning the platform into a self-defeating spectacle for the right.

The Short Version

X’s algorithm disproportionately recommends right-leaning political posts, including to users who do not follow right-wing accounts.
Controlled experiments link the algorithmic feed to modest shifts toward more conservative political opinions, not a liberal backlash.
Musk-era product and policy choices reduced moderation frictions and raised the salience of fringe and conspiratorial content.
The leap from exposure imbalances to liberal electoral advantage is unsupported; the strongest causal evidence points the other way.

What the system currently rewards: exposure, alignment, and high-salience voices

Independent audits conducted during and after the 2024 election scrutinized how X’s recommender populates the “For You” feed. A multi-week observational analysis found that right-leaning content appeared more often than left-leaning content, including for accounts without explicit right-wing follow graphs; a Sky News sampling, for instance, reported that political posts in “For You” skewed over 60 percent right versus roughly a third left, and that politically neutral accounts saw about twice as much right as left material. Complementing this, a six-week audit using 120 monitoring accounts during the election period found the system amplifies politically aligned and high-popularity accounts, reducing exposure to opposing viewpoints—a pattern consistent with engagement-optimization dynamics rather than viewpoint neutrality. Together, these audits support a descriptive claim: for many users, X is more likely to recommend right-leaning political posts than left-leaning ones.

Mechanistically, none of this is mysterious. Modern recommender systems optimize predicted engagement under constraints; when political content that is emotionally moralized, identity-charged, or otherwise provocative earns more replies and shares, the ranking model learns to surface more of it. Structural changes after the ownership transition—shrinking moderation headcount and tooling, deprioritizing some legacy integrity signals, and restricting outside research access—further reduced friction on what circulates and what can be independently verified. The result is a feed ecology that tilts toward loud and aligned political voices, and that makes fringe content more visible than it once was.

The strongest causal evidence: algorithmic feeds nudge opinions rightward

Where commentary runs ahead of the evidence is in asserting that this environment boomerangs against the right and therefore benefits liberals. The best-identified evidence points in another direction. A randomized experiment tracking users switched from a chronological to an algorithmic feed for several weeks found two critical effects: engagement increased and political opinions shifted modestly toward more conservative positions, including on issue priorities and evaluations tied to investigations involving Donald Trump and the war in Ukraine. The same research reported no significant change in affective polarization or self-identified partisanship, which cautions against grand claims of sweeping realignment—but on the narrow question of attitudinal direction, the conservative shift is the main finding. If the platform’s recommender makes conservative and right-coded content more salient and if the algorithmic feed causally nudges attitudes right, the claim that this visibility reliably advantages liberals is not supported by current best evidence.

Observational reporting aligns with that interpretation. Multiple analyses describe elevated reach for right-leaning media and influencers after the transition to Musk-era policies and product mechanics; fact-checking outlets, by contrast, saw engagement erode. Toxic or high-conflict posts also appear to receive disproportionate algorithmic lift—again, a hallmark of engagement-driven ranking. These are not proofs of electoral consequence, but they are consistent with a platform whose design choices and incentives expand right-coded exposure and influence rather than neutralize it.

Fringe salience, conspiracy memes, and the moderation–verification pivot

X is not just surfacing more mainstream conservative content; it has also heightened the visibility and political usability of conspiratorial and fringe material. A peer-reviewed analysis of 132 image-based posts shared by the owner between early 2023 and late 2024 documents how conspiracy memes gained traction under the new governance regime, aided by changes to verification (which recalibrated credibility heuristics), moderation enforcement, and recommendation. This is an important distinction for understanding the feed ecology: fringe content benefits from the same engagement feedback loops as mainstream political posts, and changes in verification and reply-ranking can amplify its spread by altering who gets exposure in the most trafficked conversational slots.

These dynamics do not, by themselves, prove electoral benefit to either coalition. They do, however, clarify why the platform can feel like an accelerant for sensational claims and intra-right radicalization even as it increases the overall visibility of right-leaning discourse. Both can be true: the algorithm promotes aligned and provocative content, and that lift includes the movement’s least disciplined voices.

Why “more exposure hurts the right” is an evidence-weak leap

The intuition behind the “boomerang” claim is politically familiar: flood the zone with extreme rhetoric and swing voters recoil. But the bridge from visibility to persuasion runs through audience psychology, message quality, and counter-messaging capacity—and the data we have do not cross that bridge. The multi-platform literature has repeatedly found that large changes in what people see often produce small or null changes in durable attitudes; Facebook and Instagram field experiments in 2020, for example, moved exposure and engagement substantially without producing commensurate shifts in polarization or core beliefs over a three-month window. Against that backdrop, the X-specific randomized study stands out by detecting not a liberal backlash but a conservative tilt from algorithmic exposure.

This does not make the platform a mind-control device. Effects are modest, conditional, and likely heterogeneous across users and topics. It does mean that a confident claim—more right-wing content on X helps liberals—is at odds with the strongest available causal evidence. Observational audits documenting exposure imbalance, while valuable, cannot supply the missing persuasion link on their own.

Attribution, uncertainty, and what would actually settle the argument

One final analytical caution: attributing effects uniquely to Musk’s ownership is hard without internal logs and versioned model histories. Engagement-driven curation, election-cycle salience, network cascades, and exogenous news events all shape what rises in feeds. That is why the key empirical gaps are straightforward to state and difficult to fill: longitudinal, preregistered panel studies that tie verified exposure and randomized feed conditions to repeated measures of attitudes and validated voter-file outcomes; transparent, replicable audits spanning election and non-election periods; and, where feasible, regulatory access to ranking-code and experiment archives to identify deliberate boosts for political categories or accounts. Until those arrive, the prudent read is this: X’s recommender currently advantages right-coded political visibility; algorithmic exposure modestly shifts opinions in a conservative direction; and the notion that this ecology reliably benefits liberals is a rhetorically attractive overreach, not an evidence-backed conclusion.

Bottom line

Treat X as what it is: an engagement-optimized political media system that presently recommends more right-leaning content and that, in at least one well-identified experiment, nudges opinions rightward. If the claim is that this arrangement does liberals a favor, the burden of proof remains unmet—and the best current study points the other way.