The buried lede here is that someone had to use Facebook to make an official statement about elections.

In a Facebook address Tuesday watched by tens of thousands of people, Venancio Mondlane again demanded a recount of the October 9 vote which the electoral commission said was won by the Frelimo party in power for almost half a century.

“We lost 50 people shot by the authorities who were supposed to protect these people,” said Mondlane, referring to a police crackdown on waves of protests he called against the election.

The conflict stems from Mozambique’s post-colonial history, one of my longest and most focused research areas, as I wrote on this blog in 2006:

It begs the question what Mozambique would have looked like if someone hadn’t assassinated Mondlane (February 3, 1969). Killing a powerful liberal-but-left American university professor of history, a respected leader within FRELIMO, ended his moderating influence over a freedom movement. FRELIMO was operating more peacefully under Mondlane as he and immediate colleagues left out rigid dogma or hierarchy; they openly invited interplay of conflicting views and positions. His assassination by the US regressed freedom and propelled turmoil.

Basically FRELIMO (Mozambique Liberation Front) ruled since independence from Portugal’s fascist dictatorship in 1975. The party transformed from the fight for liberation into (arguably due to American assassination and subterfuge) political dominance, maintaining power through a combination of legitimate support and contested electoral processes.

Several key aspects stand out in terms of today’s news:

• The use of Facebook for official opposition communications reflects both the weakness of traditional media access for opposition voices and the growing importance of social media in African political discourse. This echoes patterns seen in other African nations where social media becomes a crucial platform for opposition voices when traditional media is state-controlled.
• The allegations of 50 deaths in election-related violence, if verified, would represent one of the more serious instances of electoral violence in recent Mozambican history. However, election violence has been a recurring issue in Mozambique, particularly during local elections.
• The demand for vote recounts is a common opposition strategy in contested African elections, seen previously in Kenya, Zimbabwe, and other nations where ruling parties maintain long-term control.

How much of this violence is attributable to Facebook owning the platform in Mozambique for public discourse? I’m not saying Facebook necessarily was causal in the violence, rather that it’s more a symptom of broader issues in Mozambican democracy. The platform’s unnatural high-exit barriers and undemocratic privatization of infrastructure reflects the lack of alternative spaces for political opposition to communicate with supporters. And also that Facebook may be causal.

When Facebook becomes the de facto platform for political communication, it intentionally creates a dangerous anti-democratic dependency where a private foreign company effectively controls access to political discourse. Facebook’s algorithms and content moderation policies tend to amplify political tensions and shape how opposition movements organize and communicate. This isn’t unique to Mozambique, given we saw similar dynamics in Myanmar, Ethiopia and… the United States.

Facebook causing violence, like Tesla causing chemical spills, isn’t the core issue here as bad as it may seem, however. Facebook’s dominance was a symptom of institutional weaknesses that has setup a transition into becoming the cause (perhaps similar to how FRELIMO went from liberation to domination). A foreign-state service monopoly mindset of American businessmen likely exacerbates this. The key issue is in fact the privatization of what should be public democratic infrastructure.

Just as FRELIMO’s transition from liberation movement to ruling party was enabled by control of state resources, Facebook’s transition from communication platform to political infrastructure was enabled by network effects and data monopolies.

The critical difference lies in accountability. While FRELIMO must at least maintain some veneer of democratic legitimacy in a government role within Mozambique, Facebook faces no such local (or even international) constraints. This creates an unprecedented situation where crucial democratic infrastructure is controlled by an entity with no democratic accountability to the population it serves.

The Myanmar and Ethiopia parallels also obscure how Mozambique represents something distinct and more like the United States. We are witnessing a case where Facebooks’s role in political communication was normalized before its potential for amplifying violence was fully understood. This makes it an important case study in how privatized democratic infrastructure becomes dangerously entrenched even in the absence of acute crises.

Arguably the “soft” path to platform dependency might actually be more dangerous than the more visible crises in Myanmar or Ethiopia.

A Warning From Technology History

When Amazon executives casually suggest giving AI “a budget” to autonomously make purchases, they reveal a fraud hiding in plain sight. Like Bernie Madoff’s “consistent returns” or Lance Armstrong’s “natural talent,” their promise of beneficial AI agents masks a familiar system of exploitation. The Wired article breathlessly celebrating these developments reads like the financial press praising Enron’s “innovative accounting” – willfully blind to obvious red flags.

The Digital Company Store

Consider the coal towns of Appalachia, where companies paid workers in “scrip” — private currency only valid at the company store. Today’s tech giants build the same trap with more sophisticated tools. When Amazon’s AI agents make “personalized recommendations” based on your “preferences,” they’re creating digital scrip, a closed ecosystem where your choices are invisibly constrained and every transaction reinforces their control.

The Grover Shoe Factory disaster and Triangle Shirtwaist fire weren’t accidents, they were the inevitable result of systems that sacrificed human safety for efficiency metrics. Today’s rush to autonomous AI systems follows the same pattern, but with one crucial difference: when these systems fail, they won’t just destroy bodies they’ll eliminate human agency itself.

The Standard Oil Lesson

Standard Oil’s strategy was brilliant in its simplicity: promise customer choice while systematically eliminating alternatives. Their legacy runs deep. They fueled Nazi Germany’s war machine through Swiss intermediaries while claiming “business neutrality.” With careful metrics and corporate structures, foreshadowing thousands of Tesla crashing into buildings and exploding across America, they distanced themselves from the bombs falling on London.

Today’s tech giants use similar layers of abstraction — LLMs, cloud services, and AI agents — to distance themselves from human consequences. The Salesforce celebration of declining consumer trust while pushing for more AI autonomy shows how Meta’s role in Myanmar’s genocide wasn’t a bug; their systems predictably optimized for “engagement metrics” that amplified ethnic tensions. Just as Standard Oil knew exactly what their fuel would enable, tech companies understand perfectly well how their algorithms drive social destruction. Just as you might not trust the ship you’ve been pushed onto to cross the ocean, you also keep “using” an autonomous propeller and don’t jump overboard because your actual consent/agency has been removed.

IBM Should Never Be a Precedent

The term “AI agents” reveals a dark truth – these are indeed agents, but not working for you. IBM’s punch card technology, marketed as “efficient business automation,” became the technical infrastructure for the Holocaust. IBM’s Watson personally ensured their systems could process humans for genocide at unprecedented scale. Today, IBM markets their AI with the same “Watson” brand. How’s that for a chilling reminder of how corporate memory works, erasing an inconvenient truth of their history?

Knowing Why President Grant Won

Contrast this with Ulysses S. Grant’s approach to technology. As general and president, Grant understood that technology should serve human needs while maintaining human accountability. He mastered horsemanship not to make horses autonomous but to work in harmony with them. This wisdom guided him to create the National Weather Service, which still stands as a model for responsible automation that enhances rather than replaces human expertise.

When a hurricane approaches today, the Weather Service shows how technology should work. Meteorologists use AI to amplify their reach and effectiveness, like Beowulf with the strength of thirty men. Each forecaster can protect more people, make better predictions, serve their communities more effectively. Imagine instead if we had privatized weather prediction, with autonomous AI agents making evacuation decisions based on ‘engagement metrics’ and ‘optimization scores’ to save those who paid for premium service.

Mussolini published in his 1932 Doctrine of Fascism that “History does not travel backwards… Neither has the Fascist conception of authority anything in common with that of a police ridden State” and then he built exactly that police state and he dragged Italy back to feudal oppression. Hitler told reporters at the end of 1933 “At least we have not set up a guillotine. Even the worst elements have only needed to have been separated from the nation” right before he ordered guillotines installed at every detention facility and systematically beheaded 16,000 of his political opponents.

Today’s tech billionaires deploy the same doublespeak: promising liberation through AI while building automated systems of mass capture for total control, claiming to enhance human potential while systematically eliminating human agency. The absurdity of the double-speak scenario shows exactly what’s at stake: will AI enhance all human agency with public safety concepts like the Weather Service model, or bring Hitler and Mussolini-fever dreams back by the tech billionaires’ authoritarian visions? Does anyone, especially those proudly wearing his name, remember Leland Stanford (namesake of a Silicon Valley college-admission-to-human-oppression pipeline) was never prosecuted for corrupted industrialization of gross fraud and mass genocide?

The doublespeak and intentional deception is inherent in fascist rhetoric for a very simple reason: destruction of integrity for a centralization of power. Mussolini’s claim that fascism isn’t about police states and going backwards was itself part of the fascist playbook. Hitler’s claim to not be using the guillotine was foreshadowing, not a safeguard. Saying one thing while doing exactly the opposite, creating confusion and destroying the ability to hold power accountable, is how Thiel, Musk, Trump… corrupt government representation and undermine all digital safety in a push backwards into tyranny.

The Pattern Is Clear

My warnings about technology’s dangers have been consistent. In 2011, I warned readers of this blog to delete Facebook accounts due to Russian infiltration years before the 2016 election revelations. In 2016, I wrote here and went on speaking circuits to expose Tesla’s automated driving program as manslaughter if not murder. I could see that instead of fixing their fraud their “AI” cars would only become even more dangerous. Today their AI is killing more people than ever while their stock rises, which serves us as the perfect example of how markets reward the cruel elimination of human agency and accountability.

Tesla Death Rates Are Not Normal

Each time, the same pattern emerges: dangerous technology wrapped in progress-oriented marketing while early warnings are dismissed as “anti-innovation.” By the time the damage is undeniable, systems of control are already entrenched. With AI agents, this pattern unfolds at unprecedented scale and speed.

Since 2021 the death toll from predictable safety failures in Tesla AI engineering has more than doubled.

The Window Is Closing

When Standard Oil fueled Nazi bombers, when IBM optimized genocide, and when coal barons trapped workers in company towns, let alone when Leland Stanford set off his “killing machine” targeting Native Americans, they each claimed to be neutral service providers while building systems of control. Today’s AI systems promise to automate exploitation at scales those barons could only dream of.

The question isn’t whether these systems will be abused, it’s how much damage will be done before we admit what should be obvious: replacing human agency with corporate algorithms isn’t efficiency, it’s exploitation. The only guaranteed winners, to paraphrase Thiel’s manifesto grotesquely promoting his vision for unjust monopolization, are those trying to corner and centralize the market of technology — or in this case, the AI agents.

I’m not claiming to see the future. I see the past with both eyes open. The window for action remains political and evolutionary, not yet worse, but it’s closing fast just like the 1930s.

The choice is clear: demand AI that enhances human agency and expertise, and enforce it as such, or accept a slide to digital tyranny that would make even the worst of robber barons blush.

History will not accept “nobody warned us.” This is your warning. What would Grant do?