Don't feed the trolls 2.0 - How Lula Starved the Algorithm
Brazil revived one of the oldest rule of the internet and turned cognitive sovereignty into a campaign strategy
Once upon a time, when the internet was still a cozy, slightly chaotic village populated mostly by nerds, we lived by one golden rule: "Don't feed the trolls."
Back in the days of early discussion boards and forums - Usenet, IRC, phpBB forums - it was an infallible weapon against digital toxicity. Trolls were a noisy minority, and starving them of attention kept the digital air relatively clean. And attention was distributed by humans, not machines. The logic was beautifully simple: no reaction meant no satisfaction, and the troll would eventually crawl back under their bridge to sulk. Ignoring someone actually made them disappear (Yes, 4chan existed. It was the sewer, and everyone knew where the sewer was.).
But then came what I call the Great IT Divide. The internet morphed from a specialized tool for business and tech enthusiasts into a sprawling global metropolis, inviting everyone to the party. And with this massive influx of humanity, the volume of trolls skyrocketed.
DocIsInDaHouse
The Problem: How Trolls Became the Fuel of the Attention Society
As the digital population boomed, algorithmic social media platforms like Facebook and X took over the town square. These platforms operate on a business model that treats our cognitive focus as the most valuable currency in the world. Their algorithms quickly discovered a dark truth about human psychology: nothing keeps our eyes glued to a screen quite like outrage.
Suddenly, "feeding the trolls" wasn't a rookie mistake anymore; it was the entire business model. The algorithms actively encouraged us to fight, argue, and quote-repost our enemies, leading us directly into an exhausted Attention Society.
DocIsInDaHouse
The political camp that mastered this new reality first, and most effectively, was the extremist far-right movement. They turned "angertainment," fake news, and perpetual outrage into weapons of mass distraction. They didn't just play the algorithmic game; they cherished it.
A Real-World Lesson: The 2026 Brazilian Election
Nowhere was the danger of this algorithmic trap clearer than in the 2026 Brazilian presidential election. On one side, you had the challenger, Flávio Bolsonaro, backed by a massive, highly efficient digital ecosystem (often dubbed the "Cabinet of Hate") that was already used by his father. It was perfectly optimized to trigger algorithmic rage and flood the timelines of Facebook, X, Tiktok and Co. with divisive content.
On the other side was Luiz Inácio Lula da Silva. Lula is a classic grassroots politician who historically didn't even use a smartphone. Initially, his campaign was drowning because his supporters were falling victim to Brandolini’s Law—better known as the Bullshit Asymmetry Principle.
Brandolini’s Law states: "The amount of energy needed to refute bullshit is an order of magnitude bigger than to produce it."
In the Attention Society, this law is the death sentence for truth. A troll can invent a fake scandal in 30 seconds (low energy). A well-meaning supporter then spends 30 minutes researching and writing a well-reasoned quote-tweet to debunk it (high energy).
The problem? The algorithm doesn't read the truth. It only registers engagement. By trying to fight the bullshit, the defenders were actively powering the machine that distributed it.
After nearly losing in the first round, Lula made a bold move. He brought in brilliant digital strategists who truly understood this battlefield. Instead of trying to out-outrage the opposition on public timelines, they decided to socially hack these algorithms. To win, they resurrected the internet's oldest rule, updated for the 21st century: Don't feed the algorithm.
The Strategy: Moving to Dark Social
Lula's digital team took the fight underground. They utilized direct messaging channels like WhatsApp and Telegram. These platforms represent what marketers call "Dark Social", they are private, chronological, and, crucially, out of reach of algorithmic distortion.
Through highly organized, decentralized network groups (like "Exército do Lula"), they coordinated millions of volunteers with one strict directive: stop arguing with trolls on Facebook and Twitter. Do not engage with their outrage bait. Instead, share our positive narratives and fact-checks directly with your friends, family, and peers via direct messages.
They successfully bypassed Brandolini's Law. By refusing to debunk publicly, they starved the trolls of their precious algorithmic engagement while rebuilding a digital grassroots movement based on genuine, unfiltered human connection.
Scenarios for the Future: Reclaiming Cognitive Sovereignty
This tactical masterclass from Brazil isn't just a political anecdote; it offers a blueprint for how we can reclaim our cognitive sovereignty in an era of digital exhaustion. We don't have to be passive data points.
If we want to stop algorithms from dictating our social and political discourse, here are a few scenarios and solutions we can explore:
- The Political "Post, Don't React" Shift
Campaigns don't have to leave the public square – they have to stop fighting in it. Think of the algorithmic feed as a billboard, not a battlefield: you put your message up, you walk away, you never answer the graffiti. All the energy that used to go into viral "gotcha" moments, quote-reposts and real-time rebuttals flows instead into something slower and far more resilient: direct-messaging networks built on existing human trust – family groups, neighbourhood chats, union and church lists.
The division of labour is simple. Public feeds carry the narrative. Private channels carry the conversation. Fact-checks travel peer-to-peer, from a cousin, not from a campaign account – which is the only way to beat Brandolini's Law, because a message from someone you trust needs no algorithmic amplification to land.
The price of admission is patience. This approach builds no overnight virality and offers no dopamine of the perfect dunk. It requires a higher baseline of trust, months of groundwork, and leaders willing to lose the daily outrage war in order to win the trust war. Few campaigns have that discipline. The ones that do get something the algorithm cannot take away: a community that doesn't disappear when the feed changes its ranking. - Architectural Platform Regulation – because abstinence alone doesn't scale
There is a catch in the Lula story that we shouldn't gloss over: the same WhatsApp that carried his team's calm fact-checks in 2026 carried the Bolsonaro bulk-messaging machine in 2018. Dark social is algorithm-free, but it is also moderation-free and measurement-free. It doesn't filter for truth; it filters for trust – and trust can be abused just as efficiently as outrage. "Don't feed the algorithm" only worked in Brazil because someone else was simultaneously starving the feeder: the Superior Electoral Court's aggressive takedown regime against disinformation.
That points to the societal lever. If a campaign has to flee the public square to escape the outrage machine, the problem is the square, not the campaign. Options we could explore range from mild to radical:None of this is censorship; it is plumbing. The far right wins when the pipes reward rage. Change the pipes and the "Lula Strategy" stops being a heroic exception and becomes an ordinary way to run a democracy – one where the public square is safe to post in again, and the group chat is for your family, not your campaign.- Chronological by default. Platforms must offer – and default to – unranked, time-sorted feeds. Europe's DSA already requires at least one non-profiling option; the fight is over the default, because nobody changes defaults.
- A "quote-repost circuit breaker." Friction on the exact mechanics that feed trolls: rate-limit quote-posts of flagged content, or show the original without counting the engagement. Starve the metric, not the speech.
- Engagement-neutral ranking in election windows. For the final weeks before a vote, prohibit optimising distribution for engagement at all – the digital equivalent of the broadcast-media silence many democracies already enforce.
- Personal Digital Hygiene: Apply Brandolini's Law to yourself
On an individual level, we can all run the Lula playbook. The next time something online makes your blood boil, pause and recognise it for what it is: a trap engineered to farm your attention. Then ask the Brandolini question – who pays the energy bill if I react? If the answer is "me, and the algorithm collects," you already know what to do.
A simple hierarchy of responses, from best to worst:- Starve it. Scroll on. No reply, no quote-repost, no angry emoji. Silence is the only signal the machine cannot monetise.
- Correct privately. If someone you actually know shared the nonsense, send them a direct message. That's a conversation – it has a chance of changing a mind. A public dunk only changes your follower count.
- Reframe, don't repeat. If a topic genuinely needs a public answer, post your own position on the issue without referencing, screenshotting or paraphrasing the bait. Repeating a lie to deny it still spreads the lie – that's the asymmetry principle in action, and screenshots are no exception.
None of this is comfortable. Outrage feels like action; restraint feels like surrender. But every refusal to engage is a tiny withdrawal from the attention bank the trolls are living on. Multiplied by millions of people, that's a social hack. That's exactly how Lula's volunteers won (Lula gained more than 10 millions more vote between the first and 2nd vote).
Conclusion: Cognitive Sovereignty Goes Collective
In The Attention Society I ended with a modest, almost private act of resistance: pressing the pause button when the algorithm wants to accelerate us. Cognitive sovereignty – preferring silence to constant affirmation, complex truth to simple outrage – sounded like a self-help exercise for the digitally exhausted. Brazil shows it can be a political strategy.
Georg Franck taught us that attention is the hardest currency in the world. Brandolini's Law is that insight weaponised: the troll spends attention cheaply, the debunker spends it dearly, and the platform collects on both. Every quote-repost, however righteous, is a payment into the troll's account.
Lula's team understood that you cannot win this game by playing harder. You win by refusing to pay. Moving to dark social was not a technical trick; it was a collective withdrawal from the attention economy – thousands of people deciding, at the same time, to stop being the fuel. Da Empoli's "engineers of chaos" rely on one assumption: that we will react. A movement that doesn't react hasn't surrendered – it has cut off their supply chain.
None of this is a guarantee. Dark social is where the 2018 lies spread too, and discipline is fragile; the next provocation is only a scroll away. But the lesson stands: cognitive sovereignty does not have to remain a personal virtue. Multiplied by millions, it becomes democratic infrastructure.
The far right learned to feed the algorithm.
Lula's team learned something harder: how to go hungry.
Don't feed the trolls.
Live long and prosper 😉🖖
