Homo Deus — reading notes

I went in wanting the sequel. Sapiens explained how we got here; Homo Deus promises to tell you where we are going, and for the first half it delivers the bracing Harari move I came for. We have quietly won the old wars. Famine, plague, and war, the three horsemen that defined every previous century, now kill fewer people than sugar, old age, and our own hand. So what does a species do once it stops merely surviving? Harari's answer is that it goes looking for upgrades: immortality, engineered happiness, a kind of godhood. That part I read fast and nodding.

Then I hit the chapter on free will and stopped reading for a week, because I was too busy arguing with it.

The claim

The engine of the book is an idea Harari calls Dataism: the universe is a flow of information, organisms are algorithms, and humanism, the story that each of us has a unique inner self whose feelings matter, is not a discovered truth but a useful fiction we agreed on, the way we agreed on money. The prediction holding it all up is one sentence, and it is the one worth sitting with:

Intelligence is decoupling from consciousness.

For all of history the two came bundled, because the only things that could do intelligent work were conscious beings. Harari's point is that we are now building intelligence without consciousness, systems that decide and create and diagnose with nobody home inside them. And if intelligence no longer needs consciousness to do the work, then consciousness, the thing we assumed made us special, becomes economically optional. That is the quiet goodbye the book is really delivering.

Where I got off the bus

The free-will chapter is where Harari loses me, not because he is wrong but because he quits early. His argument is that your decisions are biochemical algorithms, and that soon outside systems will predict them better than you can. Fine. But I kept snagging on a small thing: when I decide on coffee, then change my mind and have tea, that revision feels like the most me thing I do. Is the change also just the algorithm?

Harari's answer is yes. The reconsideration is more neurons firing, the doubt is part of the program. And then he waves at the paradox and moves on. That move is his real weakness, because the interesting question is not whether the revision is algorithmic. It is whether a system that can model itself, catch itself, and surprise itself is doing something we should still be willing to call freedom, even if it is "just" computation underneath.

I tried, for a while, to take the hard line: if it is all algorithm, then feeling does not matter. But follow that consistently and it eats everything. Suffering becomes a damage signal, grief becomes post-attachment neurochemistry, love becomes oxytocin with an evolutionary invoice attached. The word doing the damage there is "just." Explaining a feeling in biological terms describes it; it does not cancel it. Your heartbeat is mechanical and we do not call it unreal. Harari slides quietly from "here is the mechanism" to "therefore the experience is an illusion," and that slide is a sleight of hand he never quite owns.

The reframe

What finally got me off the dead-end axis was giving up on the word freedom. The fight between free will and determinism is a hole you cannot climb out of, so I changed the question. Even if every choice I make is an algorithm running, it is my algorithm, shaped by my particular genes and history and hungers, and no one else holds my exact combination.

Even if it's an algorithm, it's my algorithm.

That is closer to how Spinoza thought about freedom: not the absence of causes, but acting from your own nature rather than under external force. And it exposes what Harari, oddly, almost misses. He is so fixated on the fact that outside systems can read your algorithm, can predict you, that he treats prediction itself as the threat. But being known is not the same as being owned. A musician is predictable, you can guess they will reach for the minor chord and the sad lyric, and their taste is no less theirs for it. Reading my pattern does not transfer the deed.

The residue

So the horror of the book, for me, ended up somewhere Harari does not put it. The danger was never that my choices are determined, or even that a machine could predict them. It is that the same systems reading my algorithm can also rewrite it, slowly, without my noticing or agreeing. Platforms do not merely forecast what you will want. They shape what you want, and then sell you the satisfaction of wanting it. That is a different and angrier book, the one Shoshana Zuboff actually wrote.

Which turns the whole thing political. The question everyone asks about the singularity is when: when does the machine pass us. The better question is who holds it when it does. An algorithm you own, among millions of others, is closer to electricity, infrastructure that lifts everyone. The same capability with a single owner is the most sophisticated cage ever built. I found I am not actually afraid of being replicated. I would be glad to be, in a world where the copies belong to no one in particular. What I am afraid of is the version where exactly one party owns the rewrite.

Harari wrote a long goodbye to free will. I closed the book thinking free will was never the thing worth grieving. The thing worth defending is ownership of the algorithm, and that one is not an abstract puzzle. It is a question of who we let hold the pen.

Sources

• Yuval Noah Harari, Homo Deus: A Brief History of Tomorrow (Harvill Secker, 2016). The Dataism argument, the "intelligence is decoupling from consciousness" claim, and the free-will chapter discussed throughout.
• Shoshana Zuboff, The Age of Surveillance Capitalism (PublicAffairs, 2019). The "rewriting, not just reading" framing in the closing section leans on her account.