Look What Social Media Platforms Can Do!
Nine Ways to Tilt an Election Without Deleting a Single Post.
Lawfare4All Editorial ·
Imagine two candidacies competing in a presidential election. Both have millions of followers, produce content every day, give interviews, present proposals and mobilize their supporters. Their candidates remain free to speak, their pages keep working and no one receives any notice of censorship.
Yet while one campaign's posts systematically reach millions of people, the other's messages find it increasingly hard to circulate. Some of its followers stop receiving updates, certain posts vanish from recommendations and ads run into distribution obstacles.
The difference may lie not in what the campaigns publish, but in the technological infrastructure that decides who gets the chance to receive their messages.
This hypothetical scenario illustrates a very concrete problem of contemporary communication: the formal freedom to publish does not guarantee that information reaches its audience, especially when circulation depends on private platforms that run recommendation, moderation, advertising and distribution systems.
The debate about big tech usually focuses on the spread of fake news, content removal and account suspensions. But there is another, far less visible dimension that deserves attention: the technical capacity to interfere with the conditions under which political information circulates, without necessarily altering or deleting the published content.
The point is not to imagine that every drop in reach is an electoral conspiracy, but to understand the power actually concentrated in these infrastructures and to demand that its exercise be subject to transparent, verifiable criteria compatible with democratic integrity.
The Cerimedo case: the allegation that raised the alarm
Statements by lawyer Nadia Beller, former partner of Argentine political consultant Fernando Cerimedo, added a particularly serious dimension to the debate about the power of digital platforms in Brazilian elections.
According to press reports, Beller attributed to Cerimedo privileged relationships with people linked to Meta and denounced an alleged scheme to undermine the digital visibility of Lula and his supporters, against the backdrop of the consultant's political contacts with members of the Bolsonaro family in the United States.
The allegation warrants documentary and technical investigation, especially because it emerges at a time when Lula's campaign has raised questions about difficulties with the reach and distribution of content on the company's platforms. Meta denied any electoral favoritism and attributed part of the reported problems to a global technical failure.
The seriousness of the matter goes beyond the individuals involved, because it concerns the possibility that private information-distribution mechanisms could be used to interfere with the communication conditions of a presidential race.
What could a platform actually do to tilt those conditions without altering a single message, faking any content or openly suspending a candidacy?
Much more than it seems. And that is exactly what we need to show.
Trick number one: you follow, but you may not receive
Many users believe that following a person, a page or a news outlet means receiving their posts. In practice, platforms select and rank the content they display through systems that weigh different signals of interest, behavior and relevance.
Meta itself has publicly stated that its mechanisms for distributing political content have changed, including policies that reduced, and later revised, the presence of such posts in recommendations.
So this is not an imaginary possibility: the companies acknowledge that their systems shape how much political content users are shown.
This does not prove favoritism toward any candidacy, but it shows that following someone and actually receiving their messages are two different things.
What you can do: visit the profiles you care about directly, look up their recent posts and compare what was published with what spontaneously showed up in your feed. Where available, use favorites, notifications and chronological views, bearing in mind that your options vary by platform.
Number two: the post exists, but disappears from recommendations
A platform does not need to take content down to reduce its circulation, because recommendation systems can limit its exposure to people who do not yet follow the author.
This distinction is especially important for electoral campaigns, whose communication depends not only on staying in touch with known supporters, but also on reaching new audiences, presenting proposals and competing for the attention of undecided voters.
When recommendation rules change, different profiles and categories of content can experience significant shifts in reach, even if their posts remain available through direct access.
What you can do: do not confuse an available post with guaranteed distribution. If you follow a given subject, look for different sources and go directly to their channels, including official pages and news outlets outside social media.
Number three: boosting also depends on who delivers
Digital advertising adds another layer to the problem, because paying for an ad does not necessarily mean fully controlling who will receive it, when it will be shown or under what conditions it will be distributed.
Ad delivery systems combine advertiser settings, platform rules, automated criteria and audience characteristics. That combination can produce significant differences between the audience initially targeted and the one actually reached.
The electoral problem would arise if campaigns in comparable conditions were subjected to deliberately different criteria or selective interventions, artificially altering their opportunities to communicate.
What you can do: consult public ad libraries where available, check who is paying for political messages and try to distinguish organic content from content whose circulation results from paid advertising.
Number four: the clock can also work against a campaign
Few things matter as much in an election as time, because a message suspended during the final days of a campaign does not necessarily regain its relevance when it circulates again after the vote.
Ad review processes, appeals against moderation decisions and account reinstatements can consume precious time, especially when they involve events of great public impact.
A restriction later found to be unjustified may produce political effects that its reversal cannot fully repair. The same reasoning applies to court decisions on electoral communication, whose effectiveness must also be assessed in light of the moment at which they are carried out.
What you can do: when you come across news about the suspension or reinstatement of a political post, look at the dates involved, look for the justifications given and check whether there was a real opportunity to challenge the decision before the event lost its electoral relevance.
Number five: search is not a neutral portrait of the world
Internal search engines also help organize the informational experience, because the order of results influences what users find first and how likely they are to keep looking for more information.
An investigation, an interview or a news report can remain accessible and still receive little exposure in certain digital environments, depending on the criteria used to rank results.
One should not conclude that every difference in ranking is manipulation. Even so, the power to order information deserves attention when the subjects searched are electorally relevant.
What you can do: try searching the same topic on different platforms, read outlets with different editorial lines and always look for the original source of a statement, document or court decision.
Number six: notifications can make a difference
Notifications are another form of intermediation that users often overlook, believing they are fully following a given profile simply because they turned on post alerts.
In reality, alerts also depend on individual settings and on the features each service offers. On some platforms, the experience of following an account may vary depending on the type of content or the way the user interacts with it.
What you can do: check your settings regularly, do not rely solely on automatic notifications and keep other ways of reaching the sources you consider important.
Number seven: the rules may be applied unevenly
Platforms must deal with fraud, threats, illegal content and other practices that endanger users' safety. The problem is not the existence of moderation, but the possibility that equivalent criteria are applied in unjustifiably different ways.
A candidacy may have posts removed, ads rejected or accounts temporarily restricted for legitimate reasons. The democratic question arises when comparable decisions produce different treatment without verifiable justification, especially during election periods.
What you can do: before crying censorship or celebrating a removal, try to find out which rule was applied, how the platform justified it and whether an appeal was possible. Demand the same standard for the candidates you prefer and for those you disagree with.
Number eight: popularity can also be manufactured
Not every show of support found online necessarily reflects the spontaneous participation of real people, because fake accounts, automation and coordinated operations can artificially distort engagement metrics.
These practices can amplify certain messages, simulate consensus, produce artificial trends and influence public perception of how relevant certain issues or political figures are.
The legitimate use of scheduling and automated publishing tools should not automatically be confused with coordinated inauthentic behavior. Telling them apart requires investigating the identity and behavior of the accounts, their patterns of activity and the goals of the operations identified.
What you can do: do not treat likes, views or trending topics as proof of popular support. Try to identify where information comes from and watch for repetitive behavior, especially when different profiles post identical messages as if they were independent expressions.
Number nine: perhaps the biggest trick is that you don't know what you missed
The most worrying feature of algorithmic distribution lies not only in its capacity to influence what appears on screen, but in how hard it is to know the universe of content that was never shown to the user.
It is relatively easy to challenge a false post we receive. It is much harder to question relevant information whose very existence we are unaware of, especially when our daily experience comes to be managed by personalized mechanisms that constantly select what deserves our attention.
This invisibility does not, by itself, demonstrate political intent or deliberate manipulation. It does, however, reveal a structural condition that calls for transparency, oversight and the capacity for independent investigation.
The debate has already moved beyond abstract concerns. The European Union, for example, has issued specific guidelines on the risks that large platforms pose to electoral processes, covering recommendation systems, advertising, moderation, transparency and audits.
The relevant question, then, is not whether we should do away with algorithms, but who oversees the use of systems capable of shaping the conditions under which millions of citizens receive political information.
So how can you spot possible manipulation?
Distrusting platforms does not mean turning every drop in views into an accusation of fraud, because a post's reach also depends on legitimate factors such as audience interest, timing, format, competition between pieces of content, individual choices and general changes in distribution rules.
A responsible investigation should look at performance history, document what happened, identify any moderation decisions and compare similar situations, taking into account the technical explanations offered by the companies.
For ordinary citizens, however, there is an accessible step that can contribute to this process: documenting what actually happened, rather than replacing evidence with impressions.
The voter's survival guide
You do not need to understand programming, master artificial intelligence or know the details of recommendation systems to develop habits that reduce your dependence on information selected by platforms.
The first step is to abandon the idea that your timeline is a complete sample of reality, because it is the product of a continuous selection process, personalized according to mechanisms over which users have limited control.
Keep this short list:
- Go directly to sources of information instead of waiting for them all to appear spontaneously in your feed.
- Compare posts and information across different platforms and outlets.
- Consult original documents, interviews and court decisions whenever possible.
- Be especially wary of messages that perfectly confirm all your convictions.
- Before sharing an allegation, check who made it, what evidence exists and what still depends on investigation.
- If you identify a possible undue restriction, record the date, time, post address, screenshots and any communications received from the platform.
- Do not mistake a lack of reach for automatic proof of censorship or deliberate intervention.
These precautions do not eliminate the risks of informational manipulation, but they help counter its main advantage: how easily people come to treat what they receive as if it represented everything that exists.
An election does not begin at the ballot box
Electoral integrity involves much more than the security of voting machines, because political will is formed through a continuous process of exposure to information, debates, experiences and interpretations of reality.
Platforms do not need to directly determine each citizen's vote to exert significant influence over that environment. Their capacity to select, recommend, boost and restrict content makes transparency and accountability mechanisms indispensable, in proportion to the public importance of what they do.
If these capabilities were deliberately weaponized to produce selective political advantages, the controversy would go beyond the private management of content and reach the very conditions of equality in democratic competition.
It is in this territory that information warfare may meet lawfare, especially when the strategic use of legal instruments and the asymmetric management of digital visibility begin to produce converging effects on certain participants in the political contest.
There is no need to imagine conspiracies to recognize the problem, just as there is no need to blindly accept the explanations of companies that hold control over the very data needed to verify their own practices.
Voters have the right to know the rules that affect the circulation of political information, and society has the right to demand that relevant suspicions be investigated through independent procedures.
After all, democracy depends not only on ensuring that everyone can cast a vote, but also on protecting the conditions under which each citizen freely forms their choice.
The first step to stop being played on social media is to understand that what you see is not necessarily everything they are letting you see.
Lawfare4All | Democracy, communication and power.