Harari’s Algorithmic Cargo Cult
Yuval Noah Harari’s two best-selling books Sapiens and Homo Deus have made him perhaps the most influential thinker on the planet today. But his platform — ostensibly so cutting edge — has a strangely Huxeleyan feel: Brave Old World after all. He believes that Homo sapiens (or Sapiens, as he calls the species throughout) is about to use Artificial Intelligence (AI) and biotechnology to upgrade itself, leaving a residue of old-type, natural, fouled-up humans behind — shades of Huxley’s castes but produced by upgrading the top of the heap rather than downgrading the bottom. And his own personal philosophy is close to Huxley’s Californian mysticism. Harari practises Vipassana meditation, is a vegan and is much preoccupied with the cults of Californian tech gurus seeking “bliss, immortality, divinity”.
Harari’s argument boils down to the inadequacy of poor old Sapiens in the age of big data. The information age began with the realisation by Alan Turing of the universal computing machine: all forms of information are processsable as data; algorithms are the logic operations we use to do this. Harari stresses that biology itself is algorithmic; Sapiens is quite bad at algorithms; ergo some sort of algorithmically enhanced human will supersede us. Central to this is our realisation that consciousness is not necessary for intelligent acts. Higher animals are certainly conscious but what about the termites who construct air-conditioned high-risers with fungus farms in the basement? No matter: AI can achieve feats way beyond human capacity without consciousness so one of Harari’s mantras is: “intelligence has become decoupled from consciousness”.
This enthusiasm for algorithmically decoupled intelligence seriously skews Harari’s arguments. For instance, he discusses art only in terms of the way it is being challenged and downgraded by IT. Bach for instance is wheeled on in an episode in which an outraged classical composer sought to prove the superiority of real music by challenging an audience to distinguish between his computer generated piece, one he wrote manually, and a Bach original. Harari seems to gloat over the fact that the composer lost all ways up: the audience thought the Bach written by the composer, the algorithmic piece genuine Bach and the composer’s piece written by the computer. End of story? It’s not the end of the story. Besides revealing more about the audience than the music in question, it highlights once again that great art is often quite easy to copy. The point is to make something so distinctive in the first place that everybody wants to copy it. Bach’s music is algorithmic, mathematical, and what fascinates aficionados is how music can be so mathematical — embodying for instance some elements of what we now know as jazz — and still be so moving? You need to talk chords, progressions, counterpoint to talk about Bach, not algorithms. Harari’s vision, brilliantly articulated though it is, is one most people will part company with at some point.
Similarly, he only mentions sport in the context of algorithms trumping skill: his example being the success of the Oakland Athletics baseball team in the early 2000s thanks to player recruitment by means of algorithmic assessment (the story Michael Lewis tells in Moneyball). But the algorithms didn’t replace the skill of the baseball players: they did better than the coach in assembling a team. Anyone who watches the UK Premier League knows that you don’t need an algorithm to do better than the manager in buying players; every armchair fan can do that.
Art, sport — but a major lacuna in Harari’s argument is science. He genuflects repeatedly to nanotechnology and biotechnology as magic techniques that, like algorithms, will carry all before them but he never opens the box to give examples of what these technologies are actually doing to change the world. In fact, despite many thousands of interesting journal papers the practical haul from nanotechnology is still thin. The one technique Harari does mention is nanobots zipping about the body repairing disease. But that was the scifi fantasy false dawn of nano (launched by Eric Drexler in the 1980s) that has almost no bearing on actually existing nano today. In any case, there always have been nanobots zipping about the blood locking on to bacteria and damaged tissue: they’re called drugs. Nanobot drugs are just much bigger (yes nano actually means quite big — nano molecules are much bigger than those of standard chemical drugs), smarter than old-style drugs, and they’re still very much in the development phase. Harari isn’t really interested in these things: they’re just labels intended to bamboozle the reader into thinking there must be some magic behind them.
He does discuss at length the threat to jobs from algorithmic AI, concluding that for most people the jobs they currently perform are coming to an end, but the world faces many more technological problems that he doesn’t consider, not least the energy/global warming complex of problems. Some of these will require hands-on practical chemical and engineering skills as well as IT. These are more important than half the world uploading their lives onto Facebook but they don’t get a mention in Harari’s grand scheme of things. This points up the essential divide in his work.
There are two Hararis: the historian in the line of his mentor Jared Diamond, and the prognosticator of the future (he doesn’t like to be called a prophet). The historian is by far the more successful. Good Harari (mostly found in Sapiens) is the clear-headed analyst of the past 10,000 years of human history. What sets him apart from most historians is the deep biological/geographical underpinning he learnt from Diamond. Most historians take the basic human clay for granted and their timescale is short. Very few think that to understand a period of modern history anything before the Greeks needs to be taken into account. But for Harari, as for Diamond, the key period from which all history needs to start is the agricultural revolution of around 11,500 years ago.
For biologists studying human evolution, the agricultural revolution is a recent event. Sapiens evolved around 200,000 years ago (perhaps even earlier if a recent 300,000 year-old discovery in Morocco is confirmed). And they all were hunter gathers until 11,500 years ago.
As hunter gatherers, humans were not much different to the other mammals. They had all the mammals’ expert but unarticulated knowledge of their terrain and food sources. Additionally, they had tools and fire to exploit the habitat in more varied ways, plus language, but not yet recorded language.
But the coming of agriculture meant that suddenly people were rooted in one place, with just a few crops and kinds of domestic animals to sustain them, constantly dependent on the success of their crops, which might fail in years of poor climate, and on animals which might succumb to disease and lack of food.
Harari draws great conclusions from this. So, in Homo Deus, his discussion of religion starts from the premise that Judaism, Christianity and Islam are agricultural religions. Humanity stumbled upon crop and animal domestication over a few thousand years. There was no sudden eureka moment: “we’re going to be farmers!” But, as Harari eloquently explains, instead of recognising that domestication is a good idea; aren’t we clever; why didn’t we think of this before? they attributed this change not to their growing expertise but to the gods who decided the fate of their crops and animals. Because the process of domestication had taken place over many generations, they hadn’t realised that they and nature had achieved it between them: instead, there had to be some great orchestrator in the sky instructing them: “replenish the earth, and subdue it: and have dominion over the fish of the sea, and over the fowl of the air, and over every living thing that moveth upon the earth”. From that error sprang countless miseries over thousands of year. Which have not, of course, yet abated.
So the great inequality began. What the early farmers needed was a growing rational knowledge of cultivation and animal breeding but instead they relied on the voodoo of a parasitical priestly cast, remote from tilling the soil and animal husbandry. Read Homer in the light of Jared Diamond and Harari and you see it in a different light. Here are the endless bull sacrifices, the Bronze Age nightmare of robber barons, plunder and rape. Jared Diamond notoriously averred that farming was “the worst mistake in the history of the human race”. In the early days, for all except the warriors and priests, it was.
Enter religion as we know it. The story of the Garden of Eden is an allegory of the birth of farming. It was written probably around 500 BC, perhaps 9000 years after the first domestication in the Fertile Crescent. Following that primal disaster, instead of wandering and eating a varied diet, man would be tied to a patch of land to earn his food by the sweat of his brow. If you were a pharaoh or a Homeric hero you lived life high on the hog but the life of the majority after the farming revolution was one of poor diet, endless toil and oppression.
Although in Europe and the USA less than 2% of the population now works on the land, obeisance to a god of the harvest has survived into an era when, if the crops fail or the animals succumb to disease, we know exactly why this has happened and that it was nothing to do with God.
Human beings are a secular accident of evolution, as are all organisms. There is no purpose in the process at any stage; if a niche opens up, an adapted creature may come to fill it. This has resulted in some bizarre adaptations but that’s not surprising. If an organism can survive it will. It doesn’t have to be pretty or make any sense to us. I don’t think we are very pretty or make much sense to the other creatures.
But a philosophy based on the idea that human beings are an accidental, opportunistic, improvisational species has appeal only to risk-takers: entrepreneurs, scientists and artists. The tech entrepreneurs are currently pushing the opportunistic fast-forward button very hard. Elon Musk wants us to commit to sending humans to Mars, with a view to establishing a colony. (Why? the problem with humans living on the incredibly fertile earth is not the planet but the humans. How will the total aridity of Mars solve that?) Google is committed to the driverless car. Many of the Californian technorati are planning to live forever thanks to cryogenics and the new genomic medicine they assume will be around when they are awoken from their frozen sleep.
Such ideas have little appeal for most people, who crave security and meaningful props that never change throughout their life. A certainty that is clearly false is preferable to many to this restless push forever onward. Hence the survival of the old religions, amongst which we have to include Harari’s last target — before he lays out the religion appropriate to the age of the algorithm — humanism.
Over the greater part of Homo Deus, Harari develops a sustained critique of the secular “religion” of humanism (which he takes to be the dominant movement in developed societies), along with humanism’s adjunct beliefs in free will and individuality (“the individual will transpire to be nothing but a religious fantasy”). The collapse of humanism is supposed to usher in a new religion of “dataism”. This is a mutation of extreme free market capitalism in which “data” replaces the unfettered movement of money: “Dataists believe in the invisible hand of the data flow”. So without any justification or a mechanism, dataism urges a blind immersion in maximum data flow, with a promised reward of all good: “We need only record and connect our experience to the great data flow, and the algorithms will discover its meaning and tell us what to do”. Having thus demolished all religions in the name of reason and evidence, a new one is asserted that has no reason or evidence behind it.
The dataists’ message (it’s not always clear whether Harari is observing or endorsing a phenomenon but here he paraphrases their words) is: “record everything you do and put it online. And allow Google and Facebook to read all of your emails”. So, by some unexplained alchemy, surrendering to the free flow of data is supposed to enable a creativity that passeth all understanding. As if a trillion likes of inane marginalia will suddenly produce some kind of masterpiece. This is a Cargo Cult, prostration before a totally uncomprehended but purportedly god-like force.
Until this point, the reader has assumed that Harari is making his own case. Dataism is the only and logical end of his exposition. There is no alternative proposed; the path to it has been relentless. But after the apotheosis of dataism, in the last 5 of the 397 pages of Homo Deus, the rug is suddenly pulled: “maybe we’ll discover that organisms aren’t algorithms after all”. This comes as a bit of a shock. The reader has spent most of the book thinking that the case being made is more or less congruent with Harari’s own views (although his attack on the sovereign self should have induced scepticism about attributing any opinion to Harari, because of course ideas have people rather than the other way round).
Harari believes that people need little exhortation to surrender to the great data flow: “Most people like this very much”. He’s right about that. On a plane returning to Britain a few months ago I was struck by the behaviour of a girl a few seats in front across the aisle. She had had hundreds of thumbnail pics on her smartphone and was flicking through, zooming in and flicking out almost instantly. None seemed to satisfy her. Then she opened Facebook and began furiously, automatically, replying to posts. Again flicking out without a moment’s pause. I couldn’t tell whether she was troubled by this frenetic activity or was a passive zombie but it was disturbing watching her — I have seen the future and it is a heap of unloved Instagram images. She was part of the data flow and her consciousness seemed to be on the point of being bypassed; she was merely a passive staging post in the great data relay race.
At times Harari seems so overwhelmed by the hype around big data and algorithms that he doesn’t realise he’s contradicting himself. Towards the end, he starts a riff on the “other states of consciousness” not recognised by mainstream humanism (he himself takes off every year for 6 weeks of silent meditation). He wonders about the subjective experience of bats, spurred by Thomas Nagel’s famous 1974 paper “What is it like to be a bat?” But Harari’s assertion that intelligence has been decoupled from consciousness nullifies the question; for him all creatures are just a bundle of algorithms. If any creature is closest to a robot (a self-driving car, for instance) it is the bat which can fly in total darkness and, through echo location, never hit anything. A bat doesn’t need to feel anything at all.
The fate of Homo Deus, only two years since it was written, is salutary: Harari writes of the innocence of big data — “The result will not be an Orwellian police state” — but since then data analytics garnered from Facebook and other online sources have been used to target vulnerable selected groups, unknown to them, in the UK European Referendum and the US Presidential Election. He talks of surrendering control to the wisdom of the data but there are powerful people exploiting the power of big data. The human hand hangs heavily over the algorithms. The provision of unfettered internet and social media content and search was always presented as a free lunch for everyone. But the entrepreneurs always had other aims in mind: money and power.
If Huxley comes to mind instantly, another prophetic writer’s dystopian vision can illuminate Harari’s projected scenario of an upgraded humanity 2.0 lording it over the unreconstructed Sapiens: H. G. Wells’ The Time Machine. The Time Traveller lands in an era when humanity has split into two castes: the idle, effete Eloi cavort inanely (I first read this just before the hippies appeared) but the reality is that they are being farmed by the brutal Morlocks. People like the girl on the plane — Eloi swimming in the data flow — are just meat and drink to the algorithm-manipulating Morlocks.
If I don’t follow him in his prediction of an upgraded Homo sapiens, Harari’s account of the species so far is the clearest we have, and most devastating it is for the ancient religions that still hang on in the world. Harari does for our world what Lucretius’ De rerum natura tried to do for the Romans 2000 years ago. Lucretius of course failed: his brilliantly argued polemic against all superstition gained little support and what the Romans embraced, instead of humanism 1500 years ahead of schedule, was Christianity. His book was almost lost to the world; only a chance find in a German monastery in 1417 brought it back into circulation. Now, 2000 years after Lucretius, Harari has to undertake virtually the same task.
At time they sound very similar. Here is Harari on the soul:
Think of the first baby to possess a soul. That baby was very similar to her mother and father, except that she had a soul and they didn’t … But biology cannot explain the birth of a baby possessing an eternal soul from parents who did not have even a shred of a soul. Is a single mutation, or even several mutations, enough to give an animal an essence secure against all changes, including even death?”
The religious, of course, can easily retort that you can’t use evolution against us because we don’t believe the soul evolved, it was implanted by God. Lucretius does better. He noted that intelligent and moral behaviour only emerges in human beings slowly as they mature and so, mischievously, he set up as a reductio ad absurdum the notion that anxious souls might mill around waiting to enter into each new life as it is born. No, he says, such higher attributes do not exist independently: they only emerge in due course after a long development as a property of the human animal.
Given that some of Harari’s projections for the future are already being falsified; it is the history that will last. In this he is Lucretius’ heir. Harari doesn’t quote Lucretius directly but he does cite Lucretius’ philosophical forebear and inspiration Epicurus. Lucretius had brilliantly intuitive arguments; Harari has these too but he also has a vast and ever-growing tranche of evidence to back them up. If he can help establish, once and for all, the true story of how we got to here, we can forgive him the excesses in his vision for the future.