Reading Notes: Nexus by Yuval Harari

Just wrapped up Nexus by Yuval Noah Harari. Here are my reading notes

What is Information?

Much of what we understand about the world comes to us through written and spoken forms—through letters, words, and the way they’re arranged. But the meaning behind those words often depends on interpretation. The same story, told in a different context or from a different perspective, can take on an entirely new meaning. Take, for instance, the label of “terrorist” versus “freedom fighter.” Throughout history, individuals and movements have been seen as heroes by some and threats by others. These distinctions are rarely fixed; they shift based on who’s telling the story, where it is being told from, and what they believe. In the end, interpretation and circumstance play just as much of a role in shaping truth as the facts themselves.

Stories

Much of the world today is held together by powerful stories. The 1.4 billion members of the Catholic Church are unified by the shared narratives of the Bible and Christian tradition. Similarly, 1.4 billion citizens of China are bound by the story and structure of communist ideology. On an even broader scale, the eight billion participants in the global trade network are connected by intangible yet influential forces—currencies, corporations, brands, and belief in economic systems.

Charismatic leaders often become the face of such narratives. Take Joseph Stalin, for example, whose leadership became deeply intertwined with the Soviet identity. But perhaps no figure illustrates this phenomenon more dramatically than Jesus of Nazareth. By most historical accounts, Jesus was a Jewish preacher who inspired a modest following (few followers) through sermons and acts of compassion. Yet, in the years after his death, he became the central figure in what may be one of the most transformative rebranding campaigns in human history—elevated from a local teacher to the divine incarnation of the cosmic Creator himself.

Ideas shape nations, and stories give them life. Consider the founding of the United States. For people like George Washington, the idea of a new nation—free from monarchy and built on principles of liberty—was real long before it officially came into existence in 1776. In their minds, the United States already existed, even if only as a vision. Meanwhile, others like King George III vehemently rejected this idea, clinging to the belief that the colonies were inseparable from the British crown. Storytelling is the key.

Data Collection

Before the rise of modern public health systems—long before the Soviet model or today’s global health infrastructure—diseases like dysentery and cholera routinely killed millions across the world. In 1854, a deadly cholera outbreak hit London, and the prevailing theory at the time was that people were dying from “bad air.” But one physician, Dr. John Snow, thought differently. He suspected the real cause was contaminated water. Instead of accepting assumptions, he got to work—meticulously tracking patient addresses, mapping their movements, and identifying a pattern. All signs pointed to a single water pump on Broad Street in Soho. It was tedious work—manually collecting, categorizing, and analyzing data—but it saved lives. Snow’s map became more than just a record; it became a turning point in medical history. And when we look at where we are today—with real-time data, AI-driven analytics, and global disease monitoring—it all traces back to those early acts of collecting data. Future AI networks will be different from the earlier networks in many ways.

Books

The printing press, invented by Johannes Gutenberg in the mid-1400s, was one of the most transformative technologies in human history. But to fully appreciate its impact, it’s worth remembering what came before. For centuries, all texts—including religious scriptures—were copied by hand. This was time-consuming, error-prone, and often led to inconsistencies between versions. That’s part of the reason why different versions of ancient texts, like the Bible, exist today. In many ways, books (Bible, Quran) can be viewed as early technology tools—technologies in their own right—designed to preserve truth and overcome human fallibility. When Christianity emerged in the 1st century CE, it was far from unified. It existed as one among many Jewish sects, and the New Testament as we know it was compiled only much later. With the invention of the printing press, everything changed. For the first time, information could be reproduced on a massive scale with high fidelity. The book became not just a vessel for ideas, but a mechanism for consistency. And when ambiguity arose, religious institutions—like the Church—played the role of gatekeepers, deciding what was right and what was heretical. This mass replication of knowledge also set the stage for the Scientific Revolution, as more people gained access to accurate information, enabling them to question, experiment, and build upon existing knowledge like never before.

Decisions

Every human institution is prone to errors and susceptible to corruption. Some judges accept bribes, some journalists intentionally mislead the public. This is why self-correcting mechanisms are essential for every system.

Harari frames political systems—democracies and totalitarian regimes—as fundamentally different types of information networks. How power works in a society depends on how information flows, who controls it, and who can challenge it.

In a dictatorship or totalitarian state, information is highly centralized, funneling toward a single hub—the ruler, party, or autocrat. Decision-making is concentrated at the top, leaving little room for challenge or debate. Harari uses Rome as an example of a classical centralized empire: the emperor controlled information, but distant provinces often had some autonomy simply because the central authority could not manage everything directly. Similarly, in Stalin’s Soviet Union, the regime tightly controlled media, propaganda, and information, making the leader’s narrative the only “official reality.” The death of Stalin in 1953 illustrates the fragility of such centralized systems: after he suffered a stroke, those around him “feared to touch him” or “were afraid to call the doctor,” showing how extreme concentration of power can make a system vulnerable, even to events beyond its control.

In contrast, a democracy distributes information across multiple independent nodes: citizens, media, courts, and institutions. This distributed structure allows information to be received, analyzed, critiqued, and acted upon by many, enabling self-correction, pluralism, and accountability. For Harari, democracy is not simply about majority rule—its essence lies in creating systems that encourage dissent, challenge authority, correct mistakes, and sustain a pluralistic conversation, rather than a one-sided monologue from the center.

Algorithms and Social Networks

Harari introduces computers (and by extension AI/algorithms) as a fundamentally new kind of member of human information networks—not just a tool like a printing press or radio, but a new actor. Harari references the 2016–17 case of anti-Rohingya hate in Myanmar (Burma). He argues that the platform Facebook’s algorithms—on their own, not simply humans editing—played a major role in amplifying hate content and misinformation, fostering ethnic hatred that contributed to real-world violence. He says this example shows precisely what makes modern networks different: unlike earlier technologies (print, radio, human-managed media), computers/algorithms “decide” what gets amplified—based on internal logic (e.g., engagement maximization), not human editorial judgment. Taxes: Companies like BP and Shell pay taxes to countries where they extract oil. Similarly, should the technology giants not pay taxes from where they extract data?

The Network is Always On

Each one of us is always being monitored. Especially with facial recognition technologies, we go out to a shop, get in, pay, get out, walk in the streets—all of this is monitored. The advancement of facial recognition technology also means privacy keeps getting less and less. Yuval talks about how in countries like China the systems are so good that we keep getting monitored. This also applies to other countries. He mentions what is often discussed as the Chinese social credit system, where citizens’ behavior—online and offline—might be tracked, scored, and evaluated. This “social-credit” approach effectively merges all social actions into a single reputation or “score”: small acts, financial transactions, social interactions, even civic (or non-civic) behavior can influence one’s social standing or access to services.

Fallible – The Network is Often Wrong

Information is often used to create order rather than to discover the truth. For example, in Russia, you have to adhere to the rules of Stalin. Joseph Stalin was also considered as the father of all the kids. He was considered a paternal figure for all the students. This propaganda was very effective. Now we live in an age where it is not just the person who makes the propaganda; however, computers can make the decision for us and create propaganda or promote it. When computers are given power to achieve a specific goal (e.g., increase YouTube traffic and watch time to 1 billion hours), they do what it takes to make it happen.

Military victories do not mean political success. Napoleon’s military conquests drove most European powers to unite against him and his empire collapsed a decade after he crowned himself an emperor. In the long term, Napoleon’s victories resulted in a permanent decline of France. Napoleon inadvertently laid the foundation of the unification of Germany and Italy, and France never regained its dominance. Maximizing victory is as short-sighted a goal as maximizing user engagement. Napoleon was born in Corsica to a family of Italian emigrants. Corsica was ruled by Italy for 500 years, and only a year before Napoleon’s birth was Corsica ceded to France. Many Corsicans resented, however Napoleon’s family sent him to the France military school. When the French Revolution broke out in 1789, he went back to Corsica hoping for greater autonomy. However, when he fell out with the leader of the Corsican independence movement—Pasquale Paoli—he abandoned the cause. In order to learn, baby algorithms in addition to data also need a goal.

Democracies and Self-Correction

Democracies have a self-correcting mechanism. Also, all states/democracies will have some level of corruption. Democracies can use surveillance, data, and more to offer citizens better healthcare, security without destroying people’s privacy and autonomy. While we pay physicians for their services, we don’t pay Google/Facebook; however, they use our data and leverage it to make money. Amazon and social platforms know a lot about our preferences and affiliations.

In the German elections of May 1928, the Nazi party won less than 3 percent of the vote and the Weimar Republic seemed prosperous. Within 5 years, the Weimar Republic collapsed and Hitler was the absolute dictator of Germany. This turnaround is attributed to the 1929 financial crisis. Just prior to the 1929 Wall Street Crash, the German unemployment rate was 4.5%, and by early 1932, it climbed to 25%. If a democracy could turn to a totalitarian regime in 3 years when the unemployment rate is 25%, what would happen when there is automation and loss of jobs?

It is harder to automate the job of nurses than of doctors. What about priests? Their jobs could be automated too? Old jobs will disappear, new jobs will emerge. Loan applications: if we submit from an iPhone, we have a better prospect of getting approved? If we submit when our battery charge is less than 15%, we will have denials of our loan approvals. Computers and algorithms can make all these kinds of decisions. Bots have been playing a big role in engagement.

If people in power start trusting machines/algorithms/AI more than actual human beings, it is a potential problem. Also, how power is centralized is also important.

In Rome, Sejanus was reporting into Emperor Tiberius. However, he controlled the flow of all information to Tiberius. It had to flow through him. This created problems for Tiberius. Sejanus rose very fast. As head of the Praetorian Guard, he became Tiberius’ most trusted man. When Tiberius withdrew from Rome and moved to Capri, Sejanus effectively ran the day-to-day affairs of the empire. But as Sejanus’ power grew, he overreached. Ancient sources say he began plotting to succeed Tiberius, possibly even eliminating rivals within the imperial family.

The Silicon Curtain

Humanity has never been united. There has been a link between Industrial Revolution and imperialism. Qatar has only 300,000 citizens and is pushing ambitious foreign policy aims in the Middle East—is home to Al Jazeera, the Arab world’s most influential TV network. The Liverpool-Manchester railway was seen as a turning point in transportation. Similarly, on September 30, 2012, a neural network called AlexNet won the ImageNet large scale visual recognition challenge. It surprised AI experts by achieving a success rate of 85%. By 2016, a Microsoft algorithm achieved a success rate of 96%. This also helped Iran automatically recognize unveiled women and enforce hijab laws. In the 19th century, China was late to appreciate the potential of the industrial revolution. But in the age of AI, they recognize it.

In the 16th century, Spanish, Portuguese, and Dutch, when they were building first global empires, they came up with sailing ships, horses, and gunpowder. British, Russians, and Japanese in the 19th and 20th century relied on steamships, locomotives, and machine guns. In the 21st century, it is data. Such a situation can lead to a new kind of data colonialism. What if there is a need for a Chinese or American social score for local social interactions?

Egypt exported cotton to Britain and exported high-end textiles. Malaya provided the rubber for tires and Coventry made the cars. Similarly, data could be provided by others and the imperial power can control things. Information is different from raw materials. It can be sent fast. What will happen to poorer countries if automation enables countries in Europe to produce textiles at a cheaper price?

No matter how the world is split, between a few digital empires, or 200 nation states, or a new form, cooperation is always possible. This is what we saw during the COVID pandemic. Nations sometimes should agree to limit the development of certain technologies similar to how they limit autonomous weapons. The author writes that with countries increasing their military spend, new kinds of wars could take place. The only constant of history is change.

History and the Future

Understanding history can help us get a better grasp of things. The state interests of Myanmar, Israel, Russia can never be derived from some mathematical or physics equation. They are always from a historical narrative. Politicians often go back to history to quote things. Each politician has deep conviction of where they come from/historical perspective.

The author concludes—even if Homo sapiens destroy itself, the Universe will keep going about its business as usual. If we are gone, it takes an evolution of 100 million years to produce a civilization of highly intelligent rats. The Universe is patient. We need to build institutions with strong self-correcting mechanisms.

My Take on the Book:

Reading Nexus helps me think better about what is to come. How do you know what is to come? By reading history and understanding the context of things that happened. This is exactly how I felt reading Nexus. Yuval took us back in time and also prepared us for the future. Thanks Yuval for writing a great book.”