I have to admit that I began reading Nexus with a bias – courtesy Harari’s earlier works. While I liked them when I first read them, further reading and critical takes reduced the good impression considerably.
So, while I really liked the first two chapters, I did find irony in him writing about information and truth after bring rebutted by experts on agricultural revolution and various other things he is not an expert on. And while I really like reading history, his meandering on Niall Ferguson mode in the first part of the book didn’t endear himself to me at all.
Case in point – after declaring that the printing press wasn’t the root of the scientific revolution, he made me wade through 11 pages of the history of witch hunting before letting me in on what he believes was the actual root! And then there was the deep dive on the creation of the Bible, dear lord!
All the while, I was wondering how he started the entire thing with the premise that more information doesn’t get us to the truth! I still believe that in addition to this being broadly (proximally at least) solid ground (his education has been in medieval and early modern military history), the belabouring was also intended to mitigate criticisms of taking liberty with history.
And thus, after Part 1 of Nexus, I was more or less reconciled to not really liking the book when the tide turned. The remainder of the book was refreshing – with the accessible narrative style that made him popular, and some well-structured conceptual thinking that made it a thought-provoking read. I’ll hold that opinion until/unless criticism history repeats itself.
Harari’s central thesis is that the history of humankind can be read as a series of ever-widening information networks - stories, documents, bureaucracies and, now, digital systems - and that these networks have had the potential to both empower and endanger us.
Humankind gains enormous power by building large networks of cooperation but the way these networks are built predisposes us to use their power unwisely”. Nexus (I am still iffy about that title) reframes familiar historical episodes through the lens of how information is produced, transmitted and trusted: networks create cooperation and scale, but they also amplify errors, enable manipulation, and concentrate power in new, different ways. This sets up his prime concern: the newest network, driven by AI and near-ubiquitous computation, may alter the fundamentals of politics, knowledge and human agency.
The invention of new information technology is always a catalyst for major historical changes, because the most important role of information is to weave new networks rather than represent pre-existing realities. By recording tax payments, clay tab-lets in ancient Mesopotamia helped forge the first city-states. By canonising prophetic visions, holy books spread new kinds of religions. By swiftly disseminating the words of presidents and citizens, newspapers and telegraphs opened the door to both large-scale democracy and large-scale totalitarianism. The information thus recorded and distributed was sometimes true, often false, but it invariably created new connections between larger numbers of people.
In terms of structure, Nexus consists of three parts that show information flow and networks across history – from the social/societal to the technical to the political. The first part is about human networks – how stories and documents knit communities together and created a foundation of cooperation, why myths often trump facts in social cohesion and belief systems unify resources and societies, and how bureaucratic records produce large-scale coordination.
The second part turns to the differences between organic information flows and contemporary computer systems: digital networks are continuous, always-on, and capable of reconfiguring reality in real time.
Nexus’ third part delves into the political consequences – how algorithms change decision-making, the fragility of democratic self-correction in the face of automated persuasion, and the concrete risks of militarised or ungoverned AI. The chapter sequence deliberately moves from foundation (what binds people) to mechanism (how machines change/have changed those bonds) to stakes (the pros and cons and what individuals, leaders and societies must do about it).
Throughout the chapters Nexus assembles a mix of snapshots from history (e.g., the canonisation of texts and witch-hunts that I mentioned earlier, totalitarian information machines in Russia), technological history (printing, radio, the internet), and contemporary case studies about surveillance, algorithmic bias and the geopolitical race over AI.
These show recurring patterns: information revolutions spur both rapid progress and new forms of social fragility; systems built for coordination can be repurposed for control and division; and human institutions tend to lag behind technological change.
Harari considers information as the binding force of networks, and thus, the modern problem as chiefly an information-management problem. “History is the interaction between biology and culture; between our biological needs and desire for things like food, sex and intimacy and our cultural creations like religions and laws.” What will happen when computers start creating stories, laws and religions?
Harari’s slant is definitely towards ‘be afraid’. He labels advanced AI an “alien intelligence” in its potential autonomy and argues for assertive “computer politics” – new rules, international norms and strong oversight – to prevent catastrophic misuse.
One interesting perspective is how both democracies and totalitarian regimes stand to lose if we don’t rein in AI, because of different reasons. Not radically new thoughts and understandably thin on details, but soundly argued from a historical and logical perspective. I am sure there is ‘history compression’ as usual, but to be fair, I can’t then accuse him of being on Niall Ferguson mode.
I found the book a good read because of two reasons. The accessibility of the narrative is good, as always. Second, while there are endless opinions on AI, in Nexus Harari broadly uses a systems thinking (events, patterns, structures, mental models) approach to understand as well as attempt to solve what is arguably humanity’s biggest challenge to date.
Notes from Nexus
1. Both Trump and Marxists see human interactions as a power struggle between the oppressor and the oppressed.
2. (in the book) ‘truth is understood as something that accurately represents certain aspects of reality. Underlying the notion is the premise that there exists one universal reality…Truth and reality are nevertheless different things, because no matter how truthful an account is, it can never represent reality in all its aspects. (Chapter 1 has this almost philosophical narrative I found fascinating!)
3. The naive view of information is that it leads to truth, which in turn leads to wisdom and power. The historical view is that information can be used to know the truth and create order, but these might work at cross purposes. Truth leads to wisdom, where the truth and order combo leads to power. Totalitarian regimes, for example, centralise information flows and to stifle truth to maintain order. Democratic institutions decentralise information flow to encourage the pursuit of truth.
4. The Goga government’s 1938 treatment of Jews, asking them for proof of being born in Romanian territory is eerily close to CAA in India
5. “To err is human, to persist in error is diabolical” St.Augustine
6. The printing press wasn’t the root of the scientific revolution. The formation of early curation institutions and trusted networks like the Royal Society of London for Improving Natural Knowledge that worked on the basis of self-correcting mechanisms was. (contrasting with religious texts like the Bible and absolute truths). This is also the difference between dictatorships (centralised, infallibility of the centre, and no self correction) and democracies in the information context.
7. Autocratic vs totalitarian – the former (like Nero) couldn’t know what most people were doing or saying, but the latter like Stalin’s USSR could (attempt) to control that using technology and information networks. Interesting story of how they didn’t call a doctor immediately when Stalin collapsed because there was hysteria about a conspiracy theory that doctors were the agents of a global American-Zionist plot. So anyone who called a doctor could be accused of being part of the plot to overthrow Stalin!
8. Decentralised democracies vs totalitarian regimes – the former enables free flow of ideas and informed decision-making, while the latter rely on propaganda, control including censorship. Democracy survived and evolved (until now) to manage the complexity of large societies, while totalitarian systems, despite bursts of initial success, usually fail thanks to uber centralisation and consequent inability to handle complex information effectively.
9. Social media algos are different from print and radio, because they make active and fateful decisions e.g. spreading anti-Rohingya propaganda in Myanmar because the business model prioritised maximisation of user engagement. The algo is not conscious, but is intelligent. Intelligence is the ability to attain goals, consciousness is the ability to experience subjective feelings.
10. The ability of machines to persuade human actions has been systematically increasing. In 2021, Jaswant Singh Chail broke into Windsor Castle in an attempt to assassinate Queen Elizabeth II. Encouraged by his online girlfriend Sarai, a chatbot! More recently, GPT-4 was given a test to overcome CAPTCHA. It convinced a human via TaskRabbit to solve CAPTCHA by saying it was a visually impaired human. Lying without being explicitly being programmed to do so. (Paper clip concept) Additionally, algorithmic risk assessment (COMPAS) used in courts assessed people as guilty without transparency in reasoning, and yet, the sentence was upheld.
11. In tax literature, ‘nexus’ is the connection of an entity to a given jurisdiction. Harari brings it up to show the difference between old empires or even new countries and tech corporations in the context of taxation, money and information. Tech companies use data from one part of the world to create products and profit but do not get taxed at the place of origin e.g. TikTok used cat videos uploaded in Uruguay to create image recognition AI). Money is supposed to be a universal measure of value, but as more things are valued in terms of information, while being ‘free’ in terms of money, there is a challenge. Tech giants are in many ways more powerful than nations because the computer network has become the nexus of all human activities.
12. Gold coins and dollars are intersubjective realities. Crypto is somewhere between intersubjective (idea, belief) and inter-computer (do not exist outside the network)
13. Challenging the net impact of Industrial Revolution. Many advantages, but also led to imperialism which destroyed many indigenous ways of life, costly experiments to build industrialised societies like Stalinism and Nazism, and continuing destruction of ecological balance
14. Google trained an AI based on cat images, called Meow Generator. The tech was later used by Israel to create apps that helped it to recognise Palestinians.
15. Bureaucracy literally means “rule by writing desk”. The term was invented in 18th century France, when the typical official sat next to a writing desk with drawers — a bureau… Bureaucracy seeks to solve the retrieval problem by dividing the world into drawers.
16. AI-specific
AI is the first technology in history that can make decisions and create new ideas by itself
Silicon chips can create spies that never sleep, financiers that never forget and despots that never die
Instead of dividing democracies from totalitarian regimes, a new Silicon Curtain may separate all humans from our unfathomable algorithmic overlords.
AI could either supercharge existing human conflicts, or it could create humans vs AI overlords
17. Most of us believe that increasing access to the means of information (from the printing press to the internet) leads to better informed populations. Witches wouldn’t agree because the most popular book in Europe was not a scientific book but ‘Malleus Maleficarum’, a conspiracy theory book about witches. The hunt began in earnest! In later eras, ideologies like Nazism spread delusions.
18. “The algorithms reduced the multifaceted range of human emotions – hate, love, outrage, joy, confusion – into a single catch-all category: engagement.”
Leave a Reply