With 37M people, Tokyo is the most populated city on Earth.

It would take you about five days to walk from one side to the other.

Since it’s so big, more than half of those nearly 40M people take the subway every day, which is why Tokyo’s railway network is the biggest and busiest in the world.1 Shinjuku is the busiest railway station in the world (nearly 4M people daily at its peak). Shibuya is the busiest pedestrian crossing (~100k people per hour during rush hour). Tokyo built a gigantic underground river to protect the city from flooding, sometimes referred to as the world’s largest drain. It has the most Michelin-starred restaurants, the largest wholesale fish market, one of the world’s busiest airports…

Why here though?

Isn’t there a better place?

Why did you pick the “big volcano place” for your capital, Japan?

Why is it also the capital of earthquakeland? Couldn’t you find a better place, seriously?

To add to the conundrum, Tokyo contains 30% of Japan’s population, yet it has been the capital of Japan for less than 200 years, and had just ~60k people around 1600! Yet by 1700, it might already have been the biggest city in the world!

So how come the biggest city in the world has been the capital for less than 200 years, and is situated in volcano, earthquake, tsunami, typhoon land?

It is not a coincidence! It’s there because of these things!

Japan Is Dense

Tokyo is populated because it’s in Japan, and Japan is very populated, as I explained in the previous article:

1. Japan’s staple is rice, which has a unique feature: It produces a lot of calories per km². It can feed a lot of people.

2. The south of Japan has a climate that allows for double cropping—one summer harvest of rice, one spring harvest of wheat or barley. So the population density is very high in the south.

3. Japan could supplement its diet with plenty of fish thanks to hot and cold currents meeting at the northeast coast of the main island, Honshu, producing a lot of seafood.

4. It was isolated enough that it didn’t have to fight constant wars, and it didn’t suffer wave after wave of infectious diseases.

5. It was small enough that most of its populated areas could be unified, reducing civil wars and the deaths associated with them.

6. The ashes from volcanic activity might have improved the fertility of the land.

7. The constant rain on mountains meant lots of sediments flowing to the plains, which fertilized them.

This is why every inch of Japan’s valleys is packed with people—especially in the south:

Tokyo’s Density

So the question becomes: Why is Tokyo the biggest of the Japanese cities?

The total population of a city depends on how much land it occupies, and how much population density there is in that land (people per km² x number of km²). Surprisingly, Tokyo’s population density is similar to those of Nagoya or Osaka! Since Tokyo contains ~30% of the country’s population, it implies it’s just 2.5x more dense than the average Japanese plain—which makes sense since it’s the capital. Sure, Tokyo’s central neighborhoods are denser, but some Osaka neighborhoods are in the same order of magnitude.

Tokyo is not even in the Top 50 of the world’s densest cities!2

So yes, Tokyo is quite dense, but not crazy dense. If not density then, Tokyo must just have a really big surface area. And indeed, what makes Tokyo so populated is that it has ~15,000 km² of high-density land. That’s why it takes 5 days to walk from one side to the other.

Why is Tokyo so extensive?

The Kanto Capital

A topographic map immediately shows the answer:

Tokyo is on the Kanto Plain, by far the biggest in Japan, accounting for ~15-20% of the historical farmland in the country.3 This meant that historically, it produced the most food, so its population could grow the most. And grow it did, until it became the most populated city in Japan.

OK so we now know why Japan is so populated, and why Tokyo is the biggest city within Japan (because it’s on the biggest plain). But then… Why is the biggest plain in the country there and not elsewhere in Japan? Why not in Nagoya, or Kyoto? As far as I know, nobody has answered that yet. So here is, for the first time, my hypothesis.

Read more

I just enabled translations in a dozen other languages! If you read this article on web or the Substack app, you should see an option to change the language of this article. Use this and tell me what you think!

Japan is so weird:

• Not too long ago, it was one of the most populated countries on Earth. Why?

• That’s despite being quite small and mountainous. How come?

• Why is its capital one of the largest cities on Earth?

• How come it’s farther north, south, east, and west than South Korea?

• It’s much bigger than people think. How come?

• Why are its north and west coasts much less populated than the south and east?

• Why is the Russian island of Sakhalin, just north of Japan, so sparsely populated in comparison?

• Why does it have such a strong fishing culture?

• Why is most of its population of the same ethnic background, except for the people from the northern island of Hokkaido?

All of these questions have one single root cause, and it can be represented by sushi.

Japan’s Population

Japan is still the 12th most populated country in the world.

Of the top 50, it’s the 5^th most densely populated.

Why is it so densely populated?

It can only be due to food supply: Historically, the more food you could make, the more children would survive, and the more the population would grow, up to its maximum capacity. That’s a Malthusian world.

And since most food comes from farming, only countries with lots of farming could produce big populations. So if Japan is so populous, maybe it has lots of fertile land?

Or maybe Japan’s food comes from elsewhere? Maybe they import it? Or maybe it comes from the sea?

If it’s due to farming, why such a high yield? Is its land extremely fertile, like in Egypt or Java? If so, why? Or have the Japanese made the most of their fields—are they better farmers than those in other countries?

Or is it the politics? Did the Japanese organize in a way that maximized food production?

Or is it a matter of timing—the Japanese entered the Industrial Revolution early, so they were able to produce more food earlier on, therefore growing its population earlier than others?

Or maybe Japan is not good at growing people—it’s just good at keeping them alive? Maybe Japan, as an island, avoided most wars that plagued other countries, so it didn’t lose that much population to war?

Which one is it? Maybe volcanoes?

Japan vs Java

If you read Why Is Java so Weird?, you know that this island is packed with people because of its volcanoes.

Is this also the reason for Japan’s huge population? It’s possible: They’re both on the Ring of Fire, the string of volcanoes and earthquakes-prone regions around the Pacific Ocean, which extends partway into the Indian Ocean.

To this day, new islands are constantly being formed in Japan!

This is near Iwo Jima. Source: Kazuhiro Ichikawa, via Massimo (@Rainmaker1973)

The Ring of Fire is very obvious on Google Maps: When you see the underwater trenches and mountain ranges on the edge of the tectonic plates, you realize that Japan is simply the part that emerges from the water—just like the Philippines and Taiwan!

This is also why Japan is much bigger than people think: It has many islands on these subsea mountain ranges:

This is why Japan claims these territorial waters:

Japan’s exclusive economic zone is the 8^th biggest in the world thanks to all these islands.

All these islands are the reason why the US famously took the islands of Iwo Jima and Okinawa during WW2: The US needed pit stops between Hawaii and Japan, and these islands were remote enough from the main islands that they were relatively easy to overcome, but they were also close enough to give access to Japan’s four big islands, and the conquest severed the connection between Japan and its colonies in the Pacific.

Japan eventually got them back, and that's why its maritime boundaries are farther north, south, east, and west vs South Korea.

The tectonic plates also explain why Japan is so mountainous:

Only 12% of Japan’s land is arable! So is this rare land extremely fertile because its volcanic ash fertilizes the soils, just like on Java?

Big surprise! Japan today is actually not as densely populated as South Korea or Taiwan, if we consider total cultivated land! And South Korea is not a volcanic peninsula! So, in the case of Japan, its large population isn’t attributable to volcanoes. So what is it, then? Well, let’s see where the Japanese people actually live.

Japan’s Food Geography

Half of Japan’s population lives here. Why?

Here’s a full population density map:

Here’s a clue:

Do you see it already?

These valleys surrounded by mountains make up the 12% of arable land. It’s obvious from maps like this one, showing how every nook and cranny is full of human habitats.

But also, you can see in the previous map that the big urban areas are in the southern valleys. There are other big valleys to the north, but those don’t harbor such big populations. Why?

Notice the black line. Historically, below it, farmers would plant rice in May / June and harvest it September / October, when they would drain the rice paddies and plant wheat or barley, to harvest just before planting the rice again.

Did this really have such an enormous impact on Japan’s population? If so, we should see two things. First, the amount land that rice needs to generate some amount of calories should be very small. It is:

Second, we should see the Japanese population grow when rice gets introduced, and then again when double cropping starts, and these must happen in Japan before they happen elsewhere. Do we see that?

This is the population of different East and Southeast Asian countries over the last 2,000 years:

Japan’s population was bigger than its neighbors’ for nearly 1,500 years. Let’s zoom in on that, eliminating the hypergrowth of the 20^th century:

Indonesia’s population was probably larger than Japan’s early on, but around the first century, Japan’s population started growing much faster than any neighboring country’s. It slowed down around 700-900 AD, picked up its growth again ~1200 CE, stopped in its tracks in the early 1700s, and exploded again in the 1800s. It only stopped growing recently, with the demographic collapse post-2000. We can see clearly Japan’s three big growth spurts in this graph:

The first growth spurt lasted until 700 AD, and was the result of the introduction of wet rice agriculture in Japan, via China and then Korea.

Japan is naturally quite fertile, because the rain that flows down from so many mountains brings with it not just water, but sediments to fertilize the land.

Archaeologists believe the Jōmon people arrived on the island about 40k years ago, but in the first millennium BC, the Yayoi started immigrating into Japan from Korea, bringing with them first rice paddy technology, and eventually iron tech, which helped increase agricultural productivity. That’s why the average Japanese today is genetically very close to Koreans.

Over the centuries, as the population grew, it moved farther north, bringing with it its farming knowledge.

That’s why the population boomed: It expanded along with the share of farmed land, and its productivity.

The Population Slowdown

Alas, the boom did not last forever: The population went from ~6M in 700 AD to ~4M in 1150 AD.

By around 700 AD, the Japanese had converted the best valleys into farmland, and the speed of conversion began to slow down and focus on marginal land—not the most fertile. Growth stalled. The adoption of new farming techniques like the use of iron slowed down, too.

Japan’s East-West vs North-South

This is one reason why Hokkaido, the northern island, reflects the original native population on the genetic map more than the rest of Japan (the Ainu, and also local mixing): Rice cultivation did make it there, but it was later abandoned as it was not as effective. The farmers from the south went as far as their rice agriculture allowed.

Why didn’t rice travel farther north?

Notice how a warm current from the Philippines and Taiwan reaches all the way to Japan, warming it up.1 But also, it meets a cold northern current around the north of the Honshu island. These currents are called Kuroshio and Orashio:

North of that point, it was too cold for rice cultivation.

However, the names of these currents tell us something else about Japan’s food: Kuroshio means “black current”, because this water is devoid of nutrients, so it’s more transparent. Meanwhile, Oyashio means “parent current”, because it comes from the north full of nutrients, bringing plenty of fish—the source of the fishing culture in Japan.

And this is why sushi is the most Japanese type of food. It’s where rice meets fish: the rice from the continent, migrating from south to north through the islands, and increasing the population through rice farming in the Japanese plains; the fish from the ocean, with the nutrients moving from north to south, following the winds and currents.

North & South. The rice is packed, just like the Japanese

Indeed, winds follow the same pattern as currents: Winds from Siberia travel over the Sea of Japan, catch a lot of moisture from this sea, and the drop it as they reach the tall Japanese mountains. This is why Japan’s peaks can get serious snowfall.

Why was rice pushed so far north? Because that was Japan’s staple, and that meant taxes were frequently collected in rice, which meant rulers had a strong incentive to push for rice production at the expense of other crops. Alas, the north simply couldn’t produce it.

It’s not a coincidence, then, that the southern half of Japan is more populated: Because it runs east-west, most of these islands share a similar climate, and they all could share the same crops and similar lifestyles. But then Japan bends northwards, and the islands start changing climates much faster. We can see it in the climate map of Japan:

Or, geometrically, through this map:

This is why agriculture was so crucial for Japan’s population growth—and why it stalled once most of the arable valleys were converted to rice cultivation.

But the increase in population density made for easier spread of diseases. Between 700 and 1200 AD, several waves of infectious diseases, including smallpox and measles, decimated the country, and that stalled its growth too. Why didn’t these diseases hit before? Because Japan, as an island nation, was isolated enough that most diseases didn’t make it there until maritime trade became common.

Back to Growth

Japan’s growth picks up again around 1200 AD, when new agricultural techniques are introduced into the country from the continent. First, the double cropping we mentioned before, but not because it doubled the calories produced (it didn’t: rice produces more calories per km² than wheat or barley), but rather because it added more calories at exactly the time of most need,2 it diversified the food, and it diversified the risk: Maybe the summer harvest was bad, but the winter one was good.

There’s another reason why growth was so vertiginous: unity.

Japanese Unity

Japan united very early.

By the Nara Period, around 700 AD, the biggest population centers already belonged to one single polity. By 1100 AD, the three southern islands of Honshu, Kyushu, and Shikoku were united. This meant Japan has had considerably fewer wars than most other countries. Even its long period of warring states (Sengoku Period), which lasted ~150 years between ~1450 and 1600, did not cause a dramatic loss of population.

This unity and relative lack of wars prevented massive deaths from conflict and from disease, and allowed the Japanese to invest in their irrigation infrastructure, which increased the fertility of existing fields, and also increased the surface of farming fields by reclamation of river floodplains.

Why so much unity? Geography, of course.

Geographic Seclusion

Japan is separated from the Asian continent by two seas that hinder access. The easiest path is through Korea, but the Korean Strait is quite wide (~200 km, or ~120 mi), 5x wider than the distance between France and England. And the Kuroshio current makes it very hard to cross. That said, the island of Tsushima is just in the middle, serving as a stepping stone. The Mongols tried to invade Japan twice, and both times they reached Tsushima. They were vanquished by typhoons both times—although the second time they did reach the main Japanese shores and were beaten back by the Japanese.3

The key is why only twice? Because the Korean Peninsula was a perfect buffer: Since China is a massive fertile valley surrounded by peoples from other parts of Asia, it could never fully focus on the Korean Peninsula, and as a result, Korea remained pretty independent. The Koreans fought all the wars against China on behalf of the Japanese.

And why is there a sea between China and Korea on one side, and Japan on the other? Because it’s a back-arc basin:

As the Pacific Plate dives below the Asian Plate, it creates a mountain ridge with volcanoes, which is Japan. But it also pulls Japan towards the Pacific. Imagine you’re pulling a tablecloth downward: the cloth on the table will move towards the edge. That’s what’s happening with Japan, and as a result a hole forms between Japan and Asia and fills with seawater.

This also explains why Japan’s fertile valleys are where they are:

See how much shallow sea there is around Japan? That’s ancient crust from Asia. Over millions of years, sediments coming from the mountains accumulated there, creating the alluvial valleys that are so fertile today. Compare that map with this one of population density:

Subscribe now

Takeaways

So here’s how so many aspects of Japan are explained by one factor: its geology.

• It makes it a very mountainous country: It’s the emerged side of the mountain ridge formed by the collision between the Pacific and Asian tectonic plates.

• It’s also why Japan’s maritime territory is so huge: It owns plenty of islands at the tip of these ridges.

• The collision between these plates is why Japan is close to Asia’s mainland, but not on it: It split Japan from Asia due to the formation of a back-arc basin, which is today the Sea of Japan and the East China Sea.

• This is why the sea around the Japanese main islands is pretty flat: It’s old continental shelf. As sediment accumulated from the mountains, it filled the sea, creating very fertile alluvial valleys.

• These are the valleys that produce most of the food, which is why Japan’s population is so concentrated in these valleys.

• The seas also isolated Japan enough that it didn’t suffer too many invasions and could remain united for a very long time, thus avoiding war deaths and facilitating investment in irrigation infrastructure, both of which increased its population.

• But also, Japan remained close enough to the continent to access crucial innovations, including rice cultivation, double cropping, and iron tools.

• The rice could be cultivated because the country is warm enough—despite being reasonably far north—thanks to the winds and sea currents that come from the warmer Philippines. The geographical limit where warm and cold currents collide is also the limit of historic rice cultivation.

• That’s why the vast majority of Japan’s population is concentrated in the southern parts of the country, why the north took much longer to be conquered, why today Hokkaido retains much more Jomon/Ainu ancestry than the rest of Japan, and why the Russian island of Sakhalin is so sparsely populated in comparison: The islands lie east-west in the south, and rice cultivation and the rest of Japanese lifestyle could be shared throughout these regions. Japan’s Honshu island then bends northwards together with the Pacific-Asian plate border, and the climate gets cold really fast. Hokkaido was too remote and cold for consistent rice harvests or double cropping. Sakhalin is even worse.

• This is also why Japan’s historical population booms followed the introduction of agricultural techniques: First, with rice agriculture and iron tools, then stopping when most of the available land was exhausted. Later, the introduction of further innovations such as double cropping and agricultural irrigation to reclaim floodplains increased both the amount of farmland and its productivity.

• Rice was also one way the country remained united, as it could use rice as currency for taxation.

• The downside of absorbing these technologies from the mainland was the consequent waves of disease, which were the other main obstacle to population growth. As Japan became more densely populated, infections circulated faster.

• The sea currents are why Japan has such a fishing culture: The north and south sea currents meet in northern Honshu, bringing plenty of nutrients to the warm waters and cultivating life.4

• Volcanoes might have had a small contribution to the fertility of the land, but they’re probably not a key factor.

In summary, Japan became so populous despite its small land because it could produce a lot of food from a small amount of usable land, while losing relatively few people and productive assets to war.

And sushi is the best way to represent all this, as rice came from the south, fish comes from the northern currents, and they bring the two elements that make up Japan: its sea and its land.

Share

Read more

This summer, I got devoured, but I was so confused about all the options to deal with mosquitoes that I decided to look at all the evidence. Here is the practical guide I wish I had to rid my local area of mosquitoes once and for all, including:

• The six strategies to fend off mosquitoes

• What works in each, what doesn’t, and why

• Over 50 measures specifically assessed

• The 6 killer tactics that should be enough

(No affiliate links or commissions).

Overall Strategy

You should have six layers of defense:

Read more

Share

We concluded in the previous article that we should eradicate a few species of mosquitoes. How, though? This is what we'll cover in this article:

1. How we’ve learned to fight mosquitoes over the last 130 years

2. How we achieved the technology to eradicate those we choose, along with malaria

3. Why we're not releasing this technology yet

4. What you can do to make it happen

If you’re not interested in the history and only want to know the answer, just jump to the bottom, to the section Weapons of Mosquito Destruction.

The Mosquito Life

To kill a mosquito, you must think like a mosquito.

Each development step is an opportunity to stop them:

The Early Attacks

The moment we learned the mosquito was responsible for so many diseases, epidemiologists went into mosquito extermination mode, and the diseases dwindled pretty fast: Yellow Fever was eliminated in Havana in 1901, and the malaria death rate among Panama Canal workers fell by 90% between 1906 and 1909, the result of applying all the tools at their disposal:

• Eliminate reservoirs of standing water.

• Kill larvae in the water. One way was with oil or kerosene on water surfaces, to asphyxiate the larvae that feed in the water but come to the surface to breathe. Another was with arsenic-based insecticides.1

• Use insecticides on adult mosquitoes, too. The most famous one was pyrethrum, derived from chrysanthemum flowers and already used as an insecticide for centuries.

Better Management

Later, we tried to drive Aedes Aegypti to extinction in South America. The exercise started in the 1930s and succeeded by the 1960s, but eventually the mosquito came back:

As we explained in the previous article, this was due to a bureaucracy of weekly home inspections, plus some insecticides that were either very toxic or too short-lived.2 Crucially, a third of the budget went to inspecting the inspectors.

Better Insecticides

During WW2, we found a better insecticide DDT.3 It lasted months and was not very toxic to humans, so inspectors didn’t need to visit their areas every week anymore. Once every few months was enough, bringing costs down. That’s how we eradicated Aedes Aegypti in the 1960s.4

Why Did It Fail?

Eventually, South America got reinfected because:

• As diseases disappeared, it was hard to justify the continued surveillance politically. Budgets disappeared.

• It was discovered that DDT accumulated in the environment, and could become toxic to some other species.

• Not all countries in America eradicated mosquitoes—notably, the US—so eventually South America got reinfected.

• South America’s urban population exploded, and with it, mosquito hosts and breeding grounds.

And above all, resistance.

Arms Race

We used these techniques around the world. Some mosquitoes developed resistance to DDT, so we created other chlorine-based insecticides. Mosquitoes developed resistance against them even faster, so we came up with organophosphates,5 then carbamates (both of which turned out to be even more toxic than DDT), then synthetic pyrethroids6: They are also naturally not very toxic to mammals, so they were better. That’s what we’ve been using on bed nets to fight malaria for decades now.

That’s how we achieved this:

Despite a growing population in Africa, we were able to shrink malaria deaths… Until recently. You can imagine why.7

Yes, more resistance, so we need more insecticides, in a neverending arms race. We need to break this vicious cycle. How?

The key reason why it’s so hard to eradicate mosquitoes is that the more you kill:

1. The more they develop resistance

2. The harder it is to find the remaining few. For example, you might eliminate buckets of stagnant water, but what about the water in narrow gutters? What about the water accumulated in some old tire thrown in the forest?

Subscribe now

So the challenge of anti-mosquito scientists today is: How do we come up with strategies that find and kill every single individual before they develop resistance? And for that, they have one key insight.

The top areas of mosquito annihilation research all use that insight.

Sterile Mates

Male and female mosquitoes find each other, so if you release a sterile mosquito, it will find a partner, mate, and not have offspring.

So scientists release sterile mosquitoes, but only males, so they can compete with wild males in mating with wild females, who usually only mate once.

First, they breed mosquitoes in captivity, which is pretty easy: One female mosquito can lay on average 500 eggs.8 That can mean 125,000 descendants in two generations, and about 31 million in three.

Then, they extract most of the males,9 sterilize them, for example with radiation or chemical castration, and then release them. They did this for example in El Salvador in the 1970s, and they reduced the population of mosquitoes around a lake by 99%. Of course, the population went back to normal within four months of the program stopping.10

This strategy has another advantage: It’s reasonably easy. Yes, you have to produce the mosquitoes which is work, but they reproduce alright. Then, just go to a bunch of places and release the mosquitoes. You can even use drones now to make it easier.

Another approach is through biological infertility, with the Wolbachia bacteria: You infect all the males, who will mate with an uninfected female, but their eggs won’t hatch.

Catching Female Stowaways

That’s why separating males and females is so important, but it’s very hard: If you’re releasing 10M mosquitoes and 0.1% are females, that’s still 10k females. No bueno. To avoid that, we use filters, like the fact that male pupae are a bit smaller, or hatch a bit earlier. This is an active area of research, and a paper from last year managed to reduce female contamination by 30x.

Radiation + Wolbachia

But a few females is still bad, so scientists treat mosquitoes with both Wolbachia and radiation:

• The Wolbachia makes the males infertile.

• But there are still a few females.

• So we radiate them all, making the females infertile and the males doubly infertile

• The benefit of doing this instead of just radiating is that we can radiate the males less, which is important because radiation weakens male mosquitoes and makes them less attractive. Adding Wolbachia means less radiation, male mosquitoes will be strong, and will go on and successfully mate with females in the wild.

These techniques (referred to as IIT-SIT) have achieved near elimination in the wild.

The popularity for this intervention among the population exploded after its success.

This is what Singapore has been doing across the country since 2016, with incredible results:11

There’s a problem with this approach, though: You need to keep adding mosquitoes basically forever, because the mosquitoes you release and their partners don’t have offspring, but mosquitoes from elsewhere do, and they repopulate the areas.

Is there a way you don’t need to add mosquitoes all the time? That’s what genetic engineers have been considering.

Weapons of Mosquito Destruction

First, we genetically modified mosquitoes so that their offspring don't survive, a bit like previous approaches, with similarly good results.

But then we wondered: What if we genetically modify male mosquitoes so that they can normally mate and have offspring, but the females die before they can reproduce? Only the males survive, who will go on to carry their genes onto the next generation. This is better than the previous approach, right? Before, you killed all offspring. Now, all the surviving males join the frey to have only male children. We eliminate all females.

Unfortunately, it’s not yet the solution. We must continue releasing these mosquitoes in the wild, or else their germ line eventually disappears, replaced by males who can have fully fertile offspring.

This is where the next idea is diabolically good.

Gene Drive

One day in 2003, Austin Burt thought: What if we created a selfish gene that hijacked its hosts to spread aggressively, all while making them less fertile? In a few generations, all offspring could have the gene, just as they continue dwindling until they all disappear.

The technology to make his dream come true would only appear a decade later, with CRISPR Cas9. Immediately, scientists started to think about how they could use it to eradicate mosquitoes. In 2018, they succeeded in completely eliminating the mosquitoes in two cages, just by seeding some of the males with their genetic modification.

How do you do that though? How do you spread an infertile gene to the entire population? The fertility should disappear. That’s the crazy genius: A selfish gene takes over the offspring.

Let’s take that gene G, and the natural version is N.
A modified male with the G gene in both chromosomes will be called G/G. It’s released in the wild, and finds all N/N females.

Their offspring will be G/N, of course: G from the father, N from the mother. Except this is the selfish gene’s trick: It changes the germline! The eggs and sperm of these mosquitoes change the N for a G, so that they can only contribute G to their offspring!

Early on, nothing happens: The G/G and G/N always contribute G genes, so as long as they mate with N/N mosquitoes, the descendants will all have G/N genes, and continue reproducing.

But what happens when this G gene has spread enough in the population that a G/N meets another G/N? They both only produce G gametes, so the offspring will be fully G/G. And that makes the female infertile, but not the male!

So the female stops procreating, while the male continues spreading the G gene! Eventually all the population has at least one G gene, because otherwise it means every single one of their ancestors must be N/N: Very unlikely! When this G gene is so widespread, eventually every mosquito will carry it, and no female will reproduce, collapsing the population.

Only Males

Another idea (from 2020) that might be easier to understand is a twist on this: Release males with a mutation that can only have male offspring.12

Imagine: You release 100 genetically modified males that mate with 100 wild females. Instead of having, on average, 100 surviving males and 100 surviving females, they have 200 surviving males! In the next generation, there will be 400 males, then 800… Until they overwhelm the population of wild males, and there are no more mosquitoes left. In labs, they tested this, and they found that introducing just 2.5% males like this collapsed the population in 10-14 generations.

How do they make sure there’s no resistance that emerges against this? Because in just a few generations all mosquitoes are wiped out, and because they combined this genetic change with the previous one—or several at once. It’s very unlikely that within a few generations, both genetic changes can be reversed in the same individuals.

The Caged Miracle

So we have it! Finally! After nearly 150 years of fighting against mosquitoes, we have a solution that is easy to dispense, only toxic to the one species we want to eradicate, can be forever, and wouldn’t create resistance! So what happened when we tested this?

We haven’t.

Surely, this is not possible. We have the holy grail that could eradicate malaria, yellow fever, dengue, and annoying mosquito bites, and we’ve decided not to release it?? We prefer that hundreds of millions of people are infected every year, and hundreds of thousands of toddlers and babies die?? No way!!

And yet.

Target Malaria says the first tests will be carried out in 2030. In other words, billions will be infected and millions will die because we will wait a few years.

Why?

Because we’re afraid the solution would be too good.

If this succeeds as we expect it, we could truly, irreversibly, erase a species from the face of the Earth. It’s not the first time we’ve done it to animals, nor the first time we’ve deliberately eliminated a parasite (we’ve done it with smallpox, rinderpest, trying wth polio). But it would be the first time we do it deliberately to an animal. Imagine we release this in Africa against the Anopheles Gambiae, to fight malaria. It could spread across the continent, then across continents, and there would not be any more anywhere. If there are unintended consequences, it would be too late.

Or at least, that’s the argument. Honestly, I’m not very moved:

• As we said, there are thousands of other mosquito species. This would only kill one,13 as this is genetically targeted and can’t jump around.

• This one kills a lot.

• It’s a species focused on humans, so its habitats are human habitats, not nature. We don’t need them in our habitats.

• We’ve looked into this and they’re irrelevant to these habitats: No flower depends on them, no predator, no symbiosis with other animals… Nothing

• We can reverse it!

  • We can wait for all mosquitoes to die down, and when none are left, if anything bad happens, we can always reinsert a few million that we’ve kept in labs.

  • We’ve genetically engineered a way to revert these changes if we need to!

Another argument is: “A single country could decide to eliminate mosquitoes, and they would disappear in other countries that didn’t decide to do it.” Yes, what a horror that my neighbor decides to eliminate malaria, dengue, and mosquito bites for me. I would hate him. If we can revert this anyway, who cares?

There’s another argument I’m much more open to. What happens if mosquitoes do develop resistance? A paper from two months ago simulated billions of potential mutations and noticed that some could emerge that would stop this gene and eventually prevail.14 So they proposed to not change just one gene, but a bunch of them, so that even if one or two fail, the rest can carry through. We will need time to create alternatives, test them all in software, test them with a variety of real-life mosquitoes…

Because that’s the other thing, we probably want to test it in the wild before introducing it everywhere, but we need to be supercareful about where we do that, or the mosquito can escape anyways. So maybe a remote island, making sure nobody goes in or out carrying mosquitoes during the test?15

You’d think the approach to this should then be: OK we have this holy grail, let’s test it ASAP! Let’s grab one of these islands and release the engineered mosquitoes there. Then, we can test what share of mosquitoes we catch have the engineered genes over time, see if mosquito populations shrink, see if malaria shrinks, and make sure nothing else bad happens.16 If everything works, great! We should then insert these mosquitoes in other places.

Instead, we have no test programmed! This paper from last year explains why: bureaucracy!

More than a dozen organizations have published testing guidelines for mosquito gene drives. Great! People sure know how to write papers about being careful. But who’s writing about the millions dying and about hurrying to save them? We’ve seen the exact same thing with SO₂ injection to cool the Earth: Lots of people have opinions, focused on fear-mongering and on the scary speculative risks that never materialize, while the real people who could benefit from the technology keep suffering. From those trying to test this:17

The first gene drive field trial will be determined by leaders in a country that boldly chooses to be the first. Unfortunately, they are overwhelmed by unclear guidance. Local WHO officers know little about their own agency’s published guidelines for testing genetic technology for malaria, and politicians and regulators struggle to engage with other international agencies that have published guidance. Large international forums and meetings are called to discuss risk and regulation for gene drives, but these are often out of reach for many who should be at the center of these conversations.

So what do they think would unlock this?

1. A clear message from the WHO that they support this, as well as the biggest donors: the US, the UK, Germany, France, Japan, Canada, and the EU.18

2. A clear, simple checklist from the WHO for political leaders who want to apply this. I’ve put together a draft here.

3. Stop meetings and discussions, and get national regulators to support this, ideally in the candidate countries already identified: Comoros, São Tomé e Príncipe, and Equatorial Guinea.

Releasing our Caged Miracle

We spend $4B on fighting malaria every year, and yet 600k people die. Effective Altruists have asserted for years that the best way to spend your money is on malaria mosquito nets. Just two weeks ago, GiveWell donated $276M for mosquito nets, its largest grant ever. That is probably not the best use of anybody’s time or money anymore.

We have at our fingertips the means to end this scourge—malaria, and human-biting mosquitoes too if we want—forever. It just requires one thing: Governments and institutions like the WHO must realize what they have in their hands, stop stalling, and push for a real-life test of these gene drives in isolated islands. But most of them don’t know. You can help them by letting them know: Forward this article to them, to people who might know them, or to anybody who might be interested, so they talk about it.

This can literally avoid billions of infections and millions of lives.

Share

So I wondered: Should we get rid of them once and for all?1

The Mosquito Blood Sacrifice

They’re more than an annoyance. They cause over 700M infections and kill ~700k people every year. They’re by very far the most fatal animal to us.

Mosquitoes transmit the Zika virus, West Nile virus, chikungunya virus infections, dengue fever, and most importantly, malaria.

And as we’ve seen in several articles, their destruction has gone way beyond that: Mosquitoes caused slavery, they felled empires, and they created such a burden on people that they pushed them to live in the highlands, where difficult transport and communications made them poorer and more secluded. This prevented wealth accumulation for thousands of years, becoming a major factor in the poverty of tropical regions, especially Africa. Eradicating them would not only revert that, but also reduce the burden on the health system, redirect public-health expenditure to other diseases, reduce absenteeism from schools and work, and stop filling graveyards.

Can we eliminate these diseases completely?
Can we eliminate mosquitoes?
Should we? Would there be side-effects, for example in the environment?
How can you eliminate mosquitoes from your daily life?

That’s what we’re going to cover today.

When Did We Start Eliminating Mosquito Diseases?

Mosquitoes lay their eggs in standing water or areas that can get flooded. Hippocrates noted in the 5^th century BC that swampy, low-lying areas are much more disease-prone.

And mosquitoes don’t travel uphill much. So the people living nearby rivers, lakes, or other sources of water have always suffered much more from disease. This is an illustration of Rome’s seven hills before they were built over as the Roman Empire grew.

In Rome, those in lowlands close to the River Tiber suffered more from malaria than those living uphill, rich people tended to live on the hills. The word palace comes from the Palatine Hill in Rome, where most of the palaces were. Romans noticed that, when they drained some areas, they tended to become healthier, and they told people to avoid marshy areas.2

China inadvertently stumbled upon another solution to malaria: the introduction of fish culture in rice paddy fields. The fish ate 80% of mosquito eggs, so malaria shrunk by 82%!

Unfortunately, while they’ve seen this for centuries, they didn’t realize the connection, and only pushed to develop it in the 20^th century

Quinine, isolated in the early 1800s, reduced mortality from 90% to 17%, and that allowed the European conquest of Africa. This is the origin of the Tonic Water and Gin & Tonic you drink today—its bitterness comes from quinine, which British soldiers in India mixed with sugar, lemon, and gin to make it more palatable.

But it didn’t reduce the spread of malaria itself. In fact, not much did until the 20^th century, when the vast majority of humans still lived in malaria-prone areas.

How did we eliminate it? In the late 1800s, we discovered that malaria was caused by mosquitoes, and that kick-started a global race to kill them in the 1900s. The first attempt at building the Panama Canal in the 1880s failed, among other things, because we didn’t know this back then, and too many workers died. We tried again in 1904, took measures against mosquito spread, and finally succeeded. A notable campaign was carried out in Brazil in the 1930s: Agents went door to door spraying an early insecticide (kerosene with some mosquito-killing flower extract) and draining any standing water. Later, they continued across South America. The result was the virtual elimination of the mosquito across an entire continent!

And that was despite the poor insecticide, which disappeared after a few hours or days. But during WW2 we discovered a new insecticide, DDT, which stays active on floors and walls for months. One pass was enough, which made the process much cheaper. It was first used massively on the Italian island of Sardinia, bringing malaria cases from 75,000 in 1946 to zero in 1950. Something similar happened in the US. Unfortunately, while DDT is not very toxic, it was overused in agriculture, accumulating, and causing some DDT resistance in some insect populations, so it was banned for agricultural use. We still use it as insecticide, but we can’t use it to eliminate all mosquitoes.

Eliminating Diseases Without Mosquitoes

Hold on, we eliminated mosquito-based diseases in places like the US or Sardinia, but there are still mosquitoes there. How did we eliminate the disease and not the mosquito? Do we even need to eradicate mosquitoes, or can we just eliminate the diseases they carry?

To answer that, we need to understand how mosquito infections spread:

Malaria is caused by the plasmodium parasite, transported by the mosquito from one malaria-infected person to another, so the more malaria there is somewhere, the more likely a mosquito is to pick it up, and then to give it to somebody else. Conversely, an area can have mosquitoes, but if no human has malaria, then the mosquitoes can’t catch it,3 so won’t spread it.

That means killing enough mosquitoes at once so that nobody has malaria can be enough to stop the disease without stopping the mosquito. This is precisely why malaria has been losing ground.

But not all diseases are like that. Some have animal reservoirs, so even if you eradicate the disease in humans, mosquitoes can get them from other animals (usually monkeys) and spread them again among humans. That includes Yellow Fever, Dengue, Zika, and Chikungunya.

Of these, the only one we can hope to eliminate is Dengue, because the strains that humans spread are different from the animal ones, so extinguishing Dengue among humans might be enough to eradicate the disease completely. But it won’t be enough for the other diseases.

Finally, some mosquito-spread diseases are actually not spread from human to human, but instead from animal to human: The West Nile and Eastern equine encephalitis viruses come from birds, Japanese encephalitis from birds and pigs, Rift Valley fever from livestock… Since it’s virtually impossible to eliminate all the diseases in most jungle animals, the malaria approach is not enough to eradicate diseases.

Something like that has happened in Malaysia’s fight against malaria. With insecticides, bed nets, and other initiatives, it finally found no traditional malaria in 2018 (the one caused by plasmodiums adapted to humans). But mosquitoes can still catch another type of malaria (P. knowlesi) from macaques, and then infect humans with it. Spread of this one from human to human is uncommon, but this malaria is as deadly as the others, so Malaysia has gotten rid of human malaria, but not of mosquitoes, and the result is they still get cases of this weird malaria in forest areas because of monkeys.

In summary, we might be able to reduce malaria and dengue enough in many places without eradicating mosquitoes, but some diseases (and even malaria) might require us eradicating them nevertheless.

Should we do it after all?

What Mosquitoes Should We Eradicate?

There are 3,600 species of mosquitoes, but only a few hundred bite humans. And a handful of them account for virtually all human deaths.

So we don’t need to kill all mosquitoes! We could just eradicate a few mosquito species and never suffer their diseases again: especially Aedes Aegypti, Aedes Albopictus (aka tiger mosquito), Anopheles Gambia, and Anopheles Funestus, but maybe also Anopheles Nili, Anopheles Arabiensis, Anopheles Coluzzi, and Culex Tritaeniorhynchus.

Kill The Invaders

These Aedes mosquito species are spreading everywhere, but that’s pretty recent! Aedes Aegypti used to be limited to West Africa, and only started spreading with the emergence of global trade, in the 15^th century. The tiger mosquito comes from East Asia, and spread globally at the end of the 20^th century, hitchhiking on used tires! These are invasive species!

Can we kill them? Yes! As we saw before, Aedes Aegypti was eliminated from 18 countries in South America by the 1960s. Did the ecosystem break? No. Did other animals thrive? Yes, of course! The elimination of an invasive species is not a problem!

Funnily, the tiger mosquito seems to be replacing the Aedes Aegypti in some areas!4 One invader outinvades another. It’s one of the top 100 most invasive species.

So it looks like these two specific invasive species would not be too problematic to eradicate, at least outside of their original habitats. But what about eliminating them there, or the anopheles species that cause malaria, which are mostly limited to their original habitats?

Would Eliminating Native Mosquitoes Hurt Their Environment?

As a reminder, we don’t want to eradicate all mosquitoes, just the few that feed on humans. What’s their potential value to the environment?

1. The eggs and larvae feed fish, amphibians, and birds

2. The larvae eat detritus in the water (plants, broken leaves, etc)

3. The adults pollinate some flowers

4. The adults are other animals’ meals (fish, birds, insects)

Blood Sacrifice

I have a strong opinion on points (1) and (4). These basically mean: “We take biomass from humans (their blood), and then feed it to animals, and in the process kill 700k humans per year (mostly babies and toddlers), so that these other animals can eat this additional biomass.”

As a rule of thumb, we humans should try to eliminate food chains where we’re in the middle rather than the top. Call me anthropocentric, but I don’t think we should sacrifice our children to the gods of nature anymore.

Share

If you disagree, it means you think we should continue feeding human blood to mosquitoes, so I have a solution to please everybody: Fill this form so you can offer your body to mosquitoes for the rest of us. It’s win-win: Mosquitoes get to eat, other humans don’t get bitten, and you get to sacrifice yourself for both mosquitoes and humans to keep the food chain intact.5

If sacrificing children to nature is compelling to you, and you filled the survey and gave your body to mosquitoes and still think it’s not enough, you’ll be happy to know that human-biting mosquitoes only transfer about 5,000 tons6 of biomass into nature—basically nothing. Most of that just feeds the circle of blood sacrifice, only a fraction feeds other animals. Researchers also found that mosquitoes are not a great meal: They mostly use water bodies made by humans (so little natural impact), no predator is known to depend on their larvae, the energy spent hunting the adults is more than the energy received eating them, and since they’re very widespread, it’s unlikely that any predator depends on them.7

(2) Cleaning Detritus

What about cleaning the water of decaying leaves and grasses? In reality, that’s not what happens. Bacteria and fungi do that, which protozoa eat. The mosquito larvae eat all of these, but so do many other animals: dipteran larvae, zooplankton, various aquatic insects, tadpoles… If you eliminate all human mosquitoes, you might have some more bacteria and fungi for some time until the other animals catch up. But is that really going to make the area less healthy? How, because you’re eliminating disease-ridden mosquitoes?!

You could argue that the one thing mosquitoes do is take biomass from the water and carry it out, so that birds and other non-aquatic animals can feed from it. But other animals do that too: midges, mayflies, caddisflies, dragonflies… If we wanted to eliminate all mosquitoes on Earth, maybe we should worry a bit, but again, this is just a few species.

So there’s really no special role for mosquitoes here.

(3) Pollination

Adult mosquitoes of both sexes feed on nectar and other plant sugars. When they visit flowers, pollen sticks to their body and goes to the next flower. But do they have a major, or unique, pollination role?

You can probably intuit the answer already. Do farmers use mosquitoes to pollinate their plants? No.

Here’s another intuitive way to think about it: The mosquitoes we want to eradicate here are specialized in humans, which also means they live near us. Do you think, as such, they will have some special role in nature? That some special flower or plant is uniquely pollinated by them? No.

If you’re of skeptical nature—common in these Uncharted grounds—you’ll be happy to know that this paper looked into it and confirmed the intuition: Mosquitoes visit plenty of flowers (not some unique types), flowers are visited by plenty of insects, there’s no flower that depends on these mosquitoes, and if there were and we missed it, it would probably be a human-linked weed.

This other paper actually observed Anopheles Gambiae mosquitoes (the top malaria ones), and found that they tended to eat stems and leaves more than nectar (naughty!) and when it analyzed their bodies, found no pollen (buuuh!)

Conclusion: Should We Eradicate Mosquitoes?

Yes!!

1. Mosquito diseases kill a ton of people, condemn entire populations to poverty, and are superannoying in general.

2. We can eradicate some diseases without fully eradicating mosquitoes, especially malaria, but there are limits. Some diseases can’t be eliminated without eradicating mosquitoes.

3. We only want to kill a handful of species, the other 3,500 will stay alive!

4. We’ve done it in the past! In South America, for decades, there were no Aedes Aegypti! We just stopped trying and it came back. Nature did not collapse.

5. Two of the biggest species are invaders: Aedes Aegypti and the tiger mosquito. These, we should just eradicate as invaders, to reestablish the former balance.

6. For the rest—the handful of native ones—we should also eradicate them, because:

   1. They barely add any biomass to nature

   2. The biomass they do add is mostly ours, and we should not like that

   3. Their water cleaning skills are dubious, and certainly replaceable

   4. Their pollination contribution is basically zero

Share

In other words, a handful of mosquito species sacrifice our children and add nothing, starting by Aedes Aegypti, tiger mosquitoes, and Anopheles Gambiae. We should eradicate them first, and follow with the others.

We should probably not stop at these, though: All the logic we’ve been using for the top 5-10 worst mosquito species is valid for any human-centered mosquito species. They should all be treated as parasites whose only function is to annoy us and spread our blood in the surroundings of human habitats. Like harmful viruses and bacteria, they should all be on the eradication list. As we go through that list, starting with the deadliest mosquitoes, we can observe what happens to learn.

Let’s annihilate them.
In the next article, we’ll see how: What measures governments have taken to eliminate mosquitoes, and what you, yourself, can do to get rid of yours.

Subscribe

Dedicated to Francis, the mosquito that buzzed around my ear tonight as I was finishing this article. You might have won this battle, but you will lose the war.

El 30 de julio de 2026, más de 70.000 personas se trasladaron desde Marruecos a la ciudad española de Ceuta, un enclave en el norte de África con solo 84.000 habitantes.

Todas las fuentes se localizaron mediante búsquedas en Twitter. No suponen ningún respaldo, solo se usaron para los vídeos. Fuentes: 1, 2, 3, 4, 5, 6, 7, 8

La mayoría de ellos regresó a Marruecos.

Fuentes: 1, 2

Pero semanas después, miles seguían allí.

Fuentes: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11

Esto es extraño. ¿Cómo ocurre una invasión de tales proporciones tan rápido? ¿Por qué? ¿Por qué se desinfla sin consecuencias aparentes a largo plazo? ¿O sí las hay? ¿Quién está detrás de esto? ¿Quién tenía algo que ganar con ello? ¿Marruecos? ¿Cómo sabemos la verdad? ¿Cambiará esto algo entre España y Marruecos, o en toda la UE? Eso es lo que vamos a analizar hoy.

La invasión

Este es el Estrecho de Gibraltar:

Así es como se ve la frontera sobre el terreno:

La seguridad fronteriza existe debido a la entrada de inmigrantes en España desde Marruecos durante décadas, porque el PIB per cápita español es 6 veces mayor que el de Marruecos, y el 25% de los hombres marroquíes de entre 15 y 24 años ni estudia ni trabaja. Como resultado, España ha tenido que crear obstáculos, y ahora, pocos pueden pasar. Así que el 30 de julio, miles de inmigrantes intentaron cruzar la frontera… nadando.

Pero no simplemente se les ocurrió esta idea de la nada. Llevan haciéndolo desde hace mucho tiempo: esa frontera no es realmente tan infranqueable.

La razón por la que los inmigrantes no aprovechaban esto más a menudo es porque, cuando lo hacían, eran “devueltos en caliente” a Marruecos.

Hasta el mes pasado, cuando el Tribunal Supremo español especificó que los inmigrantes que llegaran a nado a España no serían “devueltos en caliente”. Espera, ¿qué?

Las devoluciones en caliente se venían produciendo desde hacía mucho tiempo, pero las ONG se opusieron a ellas porque algunos inmigrantes podían ser solicitantes de asilo, o menores, o su devolución podía ponerlos en peligro, y exigieron que se evaluara cada caso adecuadamente. Siguió una batalla legal entre ellas y España durante años, hasta la reciente sentencia del Tribunal Supremo español que determinó que las devoluciones en caliente eran legales para los que saltan la valla.

Para los que saltan la valla.
No para los que llegan nadando.

En 2024, un argelino intentó llegar nadando a España desde Marruecos. Fue interceptado y devuelto en caliente a Marruecos. Unos abogados españoles usaron esto para recurrir,1 y el Tribunal Supremo les dio la razón el 29 de junio de 2026: No intentó saltar la valla, por lo que las devoluciones en caliente no se aplican en su caso.

La semana anterior al 30 de julio, 1500 inmigrantes ya habían bordeado la valla nadando, unos 200 al día. El asunto cobró fuerza en las redes sociales:

La mayoría de los inmigrantes, al no estar familiarizados con la jurisprudencia española, interpretaron esto como que si llegas nadando a España, no te van a deportar.

Tenían motivos para creerlo, ya que el actual Gobierno del Partido Socialista Obrero Español regularizó a más de un millón de inmigrantes en situación irregular en España,2 y existen varias excepciones a las devoluciones en caliente.

Por eso 22 líderes de países de la UE acusaron a España de fomentar este tipo de comportamientos:

En cualquier caso, el mensaje de que los inmigrantes podían entrar libremente en España empezó a difundirse el 30 de julio^th.

El mensaje decía: «El Tribunal Supremo español: no habrá expulsión inmediata de los migrantes que lleguen a Ceuta y Melilla nadando.»

Este sitio web analizó las publicaciones y encontró numerosas cuentas repitiendo ese mismo mensaje en cuestión de minutos, lo que sugiere coordinación. Una de las cuentas que repitió el mensaje fue la de Youssef Belhaissi, presentador de Al Aoula, la televisión pública estatal de Marruecos, con 1,3 millones de seguidores en Facebook. Según The New York Times, también es portavoz de lla Delegación Interministerial de Derechos Humanos de Marruecos, un organismo del Gobierno marroquí con sede en Rabat.

Durante días, el servicio secreto español advirtió que algo se estaba gestando en Ceuta, y el 29 de julio, advirtió que unas 180.000 personas se estaban preparando para entrar en Ceuta. En menos de 24 horas, entre 60.000 y 80.000 personas entraron en masa. Cerca de 100 murieron, la mayoría ahogadas o pisoteadas.

Una vez dentro, el sistema habitual se vio tan desbordado que no pudo procesar a los inmigrantes. Estos se dedicaron a saquear supermercados en busca de comida y a acampar donde pudieron. Allí hubo denuncias de violaciones grupales de niñas migrantes menores de edad, agresiones sexuales diarias,3 agresiones físicas, narcotráfico, allanamientos de morada, robos, peleas, hurtos, incendios de coches… La policía está ahora en las escuelas, el ejército está siendo atacado, y está pidiendo autorización para usar la fuerza proporcional.4

Finalmente, sin comida, refugio seguro ni esperanza de ser admitidos en España, la mayoría regresó a Marruecos o fue devuelta. Quedan unos 5.000. De ellos, cerca de la mitad son menores, que no pueden ser devueltos de manera inmediata según la ley española, aunque Marruecos, la UE y Ceuta quieren que sean devueltos. Mientras tanto, el Gobierno español quiere enviar ahora a 500 de ellos a la Península, lejos de sus familias…

Desde entonces, el Gobierno español ha estado «extendiendo la valla» 500 metros mar adentro para poder alegar técnicamente que se trata de una valla, y que así los inmigrantes que lleguen nadando también sean considerados como si hubieran saltado la valla.5

Esta es su valla:

Sin embargo, los 5.000 inmigrantes que llevan más de un mes deambulando por las calles están causando estragos. Los residentes están tan cansados están manifestándose todo el tiempo, incluso de noche.

Así que esa es la situación migratoria. Pero, ¿quién la provocó? Si fue un movimiento espontáneo desencadenado por las redes sociales… ¿por qué ocurre esto?

¿Quién está detrás de esto?

Muchas entrevistas con inmigrantes confirman que la mayoría decidió ir a Ceuta porque leyó en internet que la frontera estaba abierta. Pero, ¿basta un mensaje viral para semejante movimiento?

Fuentes: 1, 2, 3

Hay muchos vídeos que muestran camiones cargados de migrantes dirigiéndose a la frontera, y muchos muestran a las autoridades marroquíes presenciándolo y diciéndoles que avancen.6 ¿Quién pagó estos camiones?

El gobierno español culpó a los traficantes el día de la invasión, pero ¿cómo podrían sacar dinero de decenas de miles de inmigrantes entrando en tropel? No hay forma de sacar provecho de eso. Y si lo hubieran logrado, habrían perdido a estos clientes. Lo que es más importante, llegar a Ceuta no les sirve de nada; necesitan llegar a la España peninsular para tener alguna oportunidad. Las autoridades españolas investigarán esto, pero por ahora no hay mucho de dónde tirar.

Hay varios informes de que las fuerzas marroquíes se retiraron durante el incidente, que ocurrió el Día del Trono, el aniversario de la ascensión al trono del actual rey de Marruecos. Los sindicatos de la policía española han acusado al gobierno de Marruecos de estar involucrado, y la Guardia Civil (otro cuerpo policial) asegura que identificó a agentes de la inteligencia marroquí entre los inmigrantes. En 2021, otra oleada de inmigrantes fue provocada por la retirada de las fuerzas marroquíes. Cuando las fuerzas marroquíes decidieron reforzar su autoridad en la zona, pudieron impedir que se produjeran más cruces.

¿Por qué harían eso?

Territorios españoles en África

Así es como se veían en 1950:

España renunció a la mayoría de ellos, salvo las Islas Canarias y varios puntos en la costa norte de Marruecos, en el proceso de descolonización que siguieron muchos otros países europeos tras la Segunda Guerra Mundial.

Aquí tienes un resumen histórico de cada uno:

Pero hay algo que hace a Ceuta bastante distinta de cualquier otro territorio español en África:

Ceuta permite a España controlar el estrecho de Gibraltar y sus aguas marítimas. Según el derecho internacional, los buques pueden circular libremente por él, pero en caso de conflicto geopolítico, le da a España una ventaja que ningún otro país tiene, y que Marruecos podría anular si Ceuta no fuera española.

Por supuesto, el debate pasa a ser: ¿A quién debería pertenecer Ceuta? Como español, no soy imparcial, así que tomen mi opinión con cautela. Históricamente, Ceuta en concreto fue:

• Colonia cartaginesa durante un par de siglos.

• Romana durante casi 600 años.7

• Cayó ante los conquistadores musulmanes del Califato Omeya alrededor del año 700 d.C.

• Durante los siguientes 700 años, perteneció a diferentes emiratos, sultanatos y reinos musulmanes.

• En 1415, cayó en manos de los portugueses, que la controlaron hasta 1580. El Reino de Fez reconoció esta posesión.

• Desde 1580, forma parte de España.

Entonces, ¿a quién debería pertenecer? Ceuta ha sido española durante más de 400 años, ha pertenecido a un reino peninsular durante más tiempo que a la propia España, y ha estado bajo control de entidades musulmanas durante unos 700 años, menos tiempo que bajo dominio cristiano. La existencia de Ceuta como entidad vinculada a la península ibérica es anterior a la existencia del propio Marruecos (que tiene menos de un siglo), e incluso si se remonta a la dinastía alauí, esta última es 250 años más reciente. Este análisis académico explica que la reivindicación marroquí sobre Ceuta y Melilla es bastante débil.8

Marruecos entiende esto, y por eso se ha centrado durante décadas en apoderarse del Sáhara Occidental, un área sobre la que el reino marroquí no tiene una legitimidad histórica sólida para gobernar. Y aquí, el patrón es inconfundible.

• En 1957, Marruecos utilizó fuerzas irregulares para apoderarse de la mayor parte del interior del enclave español de Ifni.

• En 1957-58, usó las mismas fuerzas irregulares para penetrar en el Sáhara español y desafiar el control español.

• En 1958, la presión militar marroquí aceleró la renuncia de España a Cabo Juby/Tarfaya.

• Luego, en 1975, llegó la gran jugada. Mientras España se preparaba para conceder la independencia al Sáhara Occidental, Marruecos trasladó a 350.000 civiles a través de la frontera para desbordar políticamente a España sin necesidad de una guerra convencional, en lo que se conoce como la Marcha Verde. Fundamentalmente, España afirmó que había 25.000 soldados mezclados con los civiles. España calificó esa operación de caballo de Troya, no quiso disparar contra civiles, y Marruecos ocupó el territorio mientras España se retiraba en 1975-76, convirtiendo el control físico en una reivindicación de soberanía.

• En 1979, Marruecos ocupó el territorio abandonado por Mauritania en el lado sur del Sahara Occidental, ampliando así el control marroquí.

• En 2002, Marruecos desembarcó tropas en la isla de Perejil —reclamada por España— e izó su bandera en un intento de establecer un nuevo statu quo territorial. España los expulsó.

• En 2021, se produjo otro acontecimiento crucial en esta partida por la soberanía. España admitió para recibir tratamiento médico al líder de los independentistas saharauis. Marruecos se indignó y, en represalia, de manera deliberada relajó sus controles fronterizos en Ceuta, permitiendo que unas 8.000 personas entraran en la ciudad como forma de presión coercitiva. Un año después, de la nada y, al parecer, sin ninguna contrapartida, el presidente del Gobierno español cambió la postura de España sobre el Sahara Occidental, que pasó de ser neutral y favorable a la autodeterminación, a una declaración según la cual debería formar parte de Marruecos.

• Un informe académico español de 2023 concluyó que Marruecos está «aplicando una estrategia de zona gris para asegurar el control sobre [Ceuta y Melilla] a medio plazo».

Aunque Marruecos nunca se ha mostrado demasiado explícito respecto a Ceuta y Melilla, las considera marroquíes, y un ministro marroquí ha solicitado recientemente un diálogo con España para «abordar el futuro de Ceuta y Melilla», dados sus derechos «históricos y geográficos».

Otro ministro marroquí, e incluso el rey, han sugerido que Ceuta es un territorio ocupado.

Entonces, ¿quién es el responsable?

La explicación más sencilla es que se trató de un movimiento viral que se salió de control. Es muy poco probable que los traficantes hayan tenido un papel en su organización, aunque podrían haber intentado sacar provecho de ello. Pero ¿qué pasa con el gobierno?

• Marruecos tiene una motivación: a medida que su control sobre el Sahara Occidental se fortalece, su atención se dirige hacia las ciudades españolas de su costa norte.

• Había camiones para transportar a mucha gente.

• Las autoridades marroquíes, muy capaces de contener un ataque así, en cambio se retiraron y, en algunos casos, ayudaron a los inmigrantes. Se ha visto a algunos agentes marroquíes entre la multitud que se dirigía hacia Ceuta.

• Marruecos tiene un historial de utilizar a personas que no son autoridades legales para establecer hechos consumados sobre el terreno y así reforzar sus reclamos de soberanía.

• Marruecos tiene un fuerte incentivo para permitir que esto ocurra, ya que expone las debilidades de España en Ceuta —y probablemente en todas sus demás posesiones a lo largo de la costa—. Como mínimo, le otorga poder de negociación. Como máximo, es un precedente que podría usarse en una futura invasión de base civil que establezca hechos consumados sobre el terreno.

A pesar de todo esto, creo que es más probable que las autoridades marroquíes aprovecharan la ocasión, o simplemente dejaran que las cosas siguieran su curso, en lugar de orquestarlo todo:

• La reclamación española sobre Ceuta es demasiado sólida

• Orquestar algo así es sumamente complejo

• Los camiones podrían haber aprovechado la oportunidad: si hay demanda para ir de un lugar a otro, los transportistas emprendedores pueden hacerlo posible

• Era más fácil para las autoridades marroquíes dejar pasar a las masas que detenerlas: recordemos que ya estaban cruzando la semana anterior. Pudo haber sido una decisión espontánea, ya sea de policías desbordados sobre el terreno o de alguien en la cadena de mando

• España y Marruecos tienen demasiados intereses en común. El gas de Marruecos llega a través de España, más de un millón de sus ciudadanos viven en España, y sus principales mercados de exportación son España en particular y la UE en general. Marruecos tiene demasiado que perder en esto.

• Marruecos tiene un amplio historial de cooperación con España en materia de inmigración

Con un poco de suerte, España y Marruecos aumentarán la seguridad fronteriza y devolverán al resto de migrantes, reduciendo el riesgo de que algo así vuelva a ocurrir. En cualquier caso, habrá que estar atentos a la relación entre ambos países en los próximos meses. Podrían surgir concesiones de la nada cuando nadie esté prestando atención.

Share

He abierto el boletín a patrocinios. Si tienes un negocio que está impulsando al mundo hacia un futuro tecnooptimista y te gustaría llegar a más de 135 000 personas curiosas y con mentalidad global para darlo a conocer (fundadores, investigadores, inversores, responsables de políticas públicas), contáctame aquí.

On July 30^th 2026, over 70,000 people moved from Morocco into the Spanish city of Ceuta, an enclave in Northern Africa with only 84,000 inhabitants.

All sources found through Twitter searches. No endorsement, only used for the videos. Sources: 1, 2, 3, 4, 5, 6, 7, 8

Most of them returned to Morocco.

Sources: 1, 2

But weeks in, thousands remained.

Sources: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11

This is bizarre. How does an invasion of such proportions happen so fast? Why? Why does it deflate with no apparent long-term consequences? Or are there? Who is behind this? Who stood to gain anything from it? Morocco? How do we know the truth? Will this change anything between Spain and Morocco, or across the EU? That’s what we’re going to look at today.

The Invasion

This is the Strait of Gibraltar:

This is what the border looks like on the ground:

The border security is in place because of immigrants entering Spain from Morocco for decades, because the Spanish GDP per capita is 6x that of Morocco’s, and 25% of Moroccan men aged 15-24 neither study nor work. As a result, Spain has had to create obstacles, and now, few can pass. So on July 30^th, thousands of immigrants tried to cross the border by… swimming.

But they didn’t just come up with this idea. They’ve been doing it for ages: That border is not really so impenetrable.

The reason immigrants weren’t taking advantage of that more often is because they were insta-returned to Morocco when they did.

Until last month, when the Spanish Supreme Court specified that immigrants who swam into Spain wouldn’t be insta-returned. Wait, what?

Insta-returns have been going for a long time, but NGOs objected to them because some immigrants might be asylum seekers, or minors, or their return could put them in danger, and demanded that they be assessed properly. A legal battle between them and Spain followed for years, until the recent ruling of the Spanish Supreme Court that the insta-returns were legal for fence climbers.

For fence-climbers.
Not swimmers.

In 2024, an Algerian was trying to swim to Spain from Morocco. He was intercepted and insta-returned to Morocco. Spanish lawyers used that to appeal,1 and the Supreme Court agreed on June 29 2026: He didn’t try to climb a fence, so insta-returns don’t work for him.

The week before July 30^th, 1,500 immigrants had already swum around the fence—about 200 per day. It picked up steam on social media:

The majority of immigrants, not being well-versed in Spanish legal jurisprudence, interpreted this as if you swim into Spain, you won’t be deported.

They had reason to believe that, since the current administration of the Spanish Socialist Party legalized over a million illegal immigrants in Spain,2 and there are several carve-outs to insta-returns.

This is why 22 EU country leaders accused Spain of encouraging this sort of behavior:

In any case, the message that immigrants could freely move into Spain started spreading on July 30^th.

The message said: “The Spanish Supreme Court: There will be no immediate expulsion of migrants who reach Ceuta and Melilla by swimming.”

This site analyzed the posts and found lots of accounts repeating that same message within minutes, which suggests coordination. One of the accounts that repeated the message was that of Youssef Belhaissi, a presenter at Al Aoula, Morocco's public/state television network, with 1.3M Facebook followers. According to The New York Times, he is also spokesperson for Morocco's Interministerial Delegation for Human Rights, explicitly a Moroccan government agency in Rabat.

For days, the Spanish secret service warned that something was brewing in Ceuta, and on July 29^th, it alerted that about 180,000 people were preparing to enter Ceuta. Within 24 hours, between 60,000 and 80,000 people poured in. About 100 died, mostly drowned or trampled.

Once inside, the standard system was so overwhelmed that it couldn’t process the immigrants. They proceeded to loot supermarkets for food and camped wherever they could. There were reports of group rapes of underage migrant girls, daily sexual assaults,3 physical assaults, drug trafficking, home invasions, robberies, fights, theft, car burnings… Police are now in schools, the military is being assaulted, and is asking for license to use proportional force.4

Eventually, without food, safe shelter, or a hope of being admitted to Spain, most of them went back to Morocco or were returned. About 5,000 remain. Of those, about half are minors, who can’t be insta-returned as per Spanish law—even though Morocco, the EU, and Ceuta want them returned. Meanwhile the Spanish government now wants to send 500 of them to mainland Spain, away from their families…

Since then, the Spanish government has been “extending the fence” 500 meters into the sea, so that it can technically claim it’s a fence, and swimming immigrants will also qualify as fence-climbers.5

Here’s their fence:

Nevertheless, the 5,000 immigrants roaming the streets for over a month are causing havoc. Locals are so tired they are demonstrating all the time, even at night.

So that’s the immigration situation. But who caused it? If this was a grassroots movement triggered by social media… why can we see this?

Who’s Behind This?

Many interviews with immigrants support that most decided to go to Ceuta because they read online that the border was open. But is a viral message enough for such a move?

Sources: 1, 2, 3

There are many videos showing truckloads of migrants moving to the border—and many show Moroccan authorities witnessing them and telling them to proceed.6 Who paid for these trucks?

The Spanish government blamed traffickers on the day of the invasion, but how could they make money out of tens of thousands of immigrants rushing in? No way to monetize that. And if they had been successful, they would have lost these customers. More importantly, reaching Ceuta gives them nothing; they need to reach the European side of Spain to have a chance. Spanish authorities will investigate this, but for now there is not much to go on.

There are several reports that the Moroccan forces withdrew during the attack, which happened on Throne Day, the anniversary of the current Moroccan king’s accession to the throne. Spanish police unions have accused Morocco’s government of being involved, and the Guardia Civil (another police body) says it identified Moroccan intelligence officers among the immigrants. In 2021, another surge of immigrants was caused by Moroccan forces withdrawing. When Moroccan forces decided to heighten their authority in the area, they were able to stop any more crossings.

Why would they do that?

Spanish Territories in Africa

This is what they looked like in 1950:

Spain relinquished most of them, except for the Canary Islands and several spots on the northern coast of Morocco, in the decolonization process that many other European countries followed after WW2.

Here’s a historic summary on each:

But there’s one thing that makes Ceuta quite unlike any other Spanish territory in Africa:

Ceuta allows Spain to control the Strait of Gibraltar and its maritime waters. Under international law, vessels can freely move through it, but in the case of geopolitical conflict, it gives Spain a leverage that no other country has—and that Morocco could break if Ceuta was not Spanish.

Of course, the debate becomes: Who should Ceuta belong to? As a Spaniard, I’m biased, so take my opinion with a grain of salt. Historically, Ceuta specifically was:

• A Carthaginian colony for a couple of centuries.

• Roman for nearly 600 years.7

• It fell to the Muslim conquerors of the Umayyad Caliphate around 700 AD.

• For the next 700 years, it belonged to different Muslim emirates, sultanates, and kingdoms.

• In 1415, it fell to the Portuguese, who controlled it until 1580. The Kingdom of Fez recognized this ownership.

• Since 1580, it’s been part of Spain.

So who should it belong to? Ceuta has been Spanish for over 400 years, it has belonged to a peninsula kingdom longer than to Spain itself, and has been controlled by Muslim polities for about 700 years—fewer than by Christians. Ceuta’s existence as a polity linked to the Iberian Peninsula predates the existence of Morocco itself (which is younger than a century), and even if you stretch it to the Alawi Dynasty, it’s 250 years older. This academic assessment explains how the Moroccan claim to Ceuta and Melilla is quite weak.8

Morocco understands this, and that’s why it has focused for decades on taking over the Western Sahara—an area the Moroccan kingdom doesn’t have strong historic legitimacy to rule. And here, the pattern is unmistakable.

• In 1957, Morocco used irregular forces to seize most of the hinterland of the Spanish enclave of Ifni.

• In 1957–58, it used the same irregular forces to penetrate the Spanish Sahara and challenge Spanish control.

• In 1958, Moroccan military pressure accelerated Spain’s relinquishing of Cape Juby/Tarfaya.

• Then, in 1975, came the big play. As Spain was preparing to give the Western Sahara its independence, Morocco moved 350,000 civilians across the border to overwhelm Spain politically without conventional war, in what’s called the Green March. Crucially, Spain claimed it had 25,000 troops mixed with the civilians. Spain called that operation a Trojan Horse, didn’t want to fire on civilians, and Morocco occupied the territory as Spain withdrew in 1975-76, converting physical control into a sovereignty claim.

• In 1979, Morocco occupied the territory abandoned by Mauritania on the southern side of Western Sahara, expanding Moroccan control.

• In 2002, Morocco landed forces on the island of Perejil—claimed by Spain—and raised its flag in an attempt to establish a new territorial status quo. Spain expelled them.

• In 2021, we have another crucial event in the play for sovereignty. Spain admitted for clinical treatment the leader of the Western Sahara independentists. Morocco was furious, and in retaliation it purposefully relaxed its border controls in Ceuta, enabling ~8,000 people to enter the city as coercive pressure. One year later, out of the blue, and apparently without any counterbalancing offer, the Spanish PM changed Spain’s stance on the Western Sahara from neutral and preferring self-determination, to a declaration that it should be part of Morocco.

• A 2023 Spanish academic report concluded that Morocco is “operating a Grey Zone strategy to secure control over [Ceuta and Melilla] in the medium term.”

Although Morocco has never been too vocal about Ceuta and Melilla, it considers them Moroccan, and a Moroccan minister has recently requested dialogue with Spain to “address the future of Ceuta and Melilla”, given its "historical and geographical" rights.

Another Moroccan minister, and even the king, have suggested Ceuta is an occupied territory.

Who Is Responsible Then?

The simplest explanation is that this was a viral movement that got out of control. It’s very unlikely that traffickers had a role in orchestrating it, even though they might have tried to benefit from it. But what about the government?

• Morocco has a motivation: As its grip on the Western Sahara strengthens, its attention moves to the Spanish cities on its northern coast.

• There were trucks to transport many people.

• The Moroccan authorities, very capable of withstanding such an attack, instead withdrew, and in some cases abetted the immigrants. Some Moroccan agents have been seen among the crowds that moved into Ceuta.

• Morocco has a track record of using people who aren’t legal authorities to establish facts on the ground to further its sovereignty claims.

• Morocco has a strong incentive to let this happen, as it exposes the weaknesses of Spain in Ceuta—and probably all other Spanish possessions along its coast. At least, it gives it leverage in negotiations. At most, it’s a precedent to use in a future civilian-based invasion that establishes facts on the ground.

Despite all of this, I think it’s more likely that Moroccan authorities jumped on the occasion, or simply let it run, rather than orchestrating it all:

• The Spanish claim on Ceuta is way too strong

• Orchestrating something like this is very complex

• Trucks may have seized the opportunity: If there’s demand to go from one place to the other, entrepreneurial truck drivers can make it happen

• It was easier for Moroccan authorities to let the masses through than to stop them: Remember they were already passing through the week before. It might have been a spontaneous decision, either by overwhelmed street-level police, or by somebody in the command chain

• Spain and Morocco have way too many common interests. Morocco’s gas comes via Spain, over 1M of its citizens live in Spain, and its main export markets are Spain in particular and the EU in general. Morocco has way too much to lose here.

• Morocco has a track record of being pretty cooperative with Spain in terms of immigration

Hopefully Spain and Morocco will increase their border security and return all remaining migrants, lowering the risk of something like this happening again. In any case, keep an eye on the relationship between the two countries in the coming months. Some concessions may come out of the blue, when nobody’s paying attention.

Share

I'm opening the newsletter to sponsorship. If you have a business that is moving the world towards a techno-optimistic future and would like to reach 135k+ curious, globally-minded people to learn about it (founders, researchers, investors, policymakers), reach out here.

In this article I’ll do two things:

1. Go through some of the AI posters people submitted, to discuss why people generally dislike AI art.1

2. Analyze the rest of the posters, and announce the winner of the $1,000 competition.

Shallow Intent

This was the first poster of the competition:

At first sight, I was impressed: It really captured the Uncharted Territories mindset at two levels. First, the idea of Burn the ships, which I’ve written about here, harks back to Hernán Cortés’s conquest of Mexico, and his act of burning (actually, scuttling) the ships they came in so that his men could only look forward, towards the adventure, the uncharted territories.

Second, the aesthetic execution is great: the edge of the poster is like old parchment, the teal and golden colors are the same palette as UT, the orange accent color in the fire, the ship of the Age of Discovery, the lines in the water as topographic map lines…

But then you start paying attention to the details, and some things start being off:

• Why is he wearing what looks like 1800s English garb? These are 1500s Spanish conquistadors.

• How did the ship make it so far inland?

• Why is the dragon a mix of Western and Eastern styles? Why does the dragon have what looks like a second tail?

• Why is the man walking towards the ships?

A human would probably not have drawn that, especially that last point: If the creator had enough insight to create a Burn the Ships poster, they would have understood the meaning, and would have portrayed the man leaving rather than approaching. The point is the exact opposite.

So here’s the key insight: I don’t think people mind that an AI has made an image. I think they mind that it’s missing thought. When humans are creating something like this, they need to add every element, one by one, which means they pay attention to every detail. Every stroke is imbued with meaning, the result of the creator thinking: “How should I trace this thing? Why?”

And I think art, in a big way, is the process of the audience reverse-engineering the creator’s thoughts: “Why did the artist make this poster like this? What was he trying to convey? Why did she choose this line, this color, this composition?” I think that’s what makes art meaningful: You’re getting into the artist’s brain, into their knowledge, their emotions.

But when AIs create images, they don’t have emotions, they don’t think critically about every detail. They’re making statistical recombinations of all the images they’ve ever seen, weighed by the words in the prompt. They’re an amalgamation of the Internet’s visuals. They generate the appearance of intention, but in reality, there isn’t any.

So we feel cheated. It takes decades of cultural training for humans to learn the clues that an image has meaning: “Look at the harmony of the composition, look at the swiftness of the stroke, look at the minute detail she painted… The artist really thought about this, therefore, it’s worth my attention.” That is out the window and into the trash, because the AI mimics these features but without the meaning, without the intentionality of the artist. There is visual density without intention density. AI decouples these.

Here’s an example of an image I generated last week about a family in Mesopotamia, in Ur, around 2000 BC:

This is the 20^th or so attempt, and yet when you ask an AI what’s wrong about it, it tells you everything is wrong: the grain milling position, the stone, the oven, the loom, how the grain is stored, the sheep, the wall surface, the window sizes, the clothing…2 An artist depicting this would have a conversation with a historian to get as many of these details right, but the AI doesn’t do that: It takes whatever is in the prompt, and makes an image that closely resembles that direction. It does try to imply what you mean: For example, when my prompt refers to a Mesopotamian family in 2000 BC, it will fetch references from old history in that place. But AIs don’t yet reason through it. They don’t ask themselves: What would be historically accurate? What elements should be present and absent? I’m sure AIs will have this internal conversation in the future, and that would be akin to the illustrator talking with the historian, but today that’s not the case.

AI Grime

At first sight, an AI image advertises depth, so we feel betrayed when we realize it’s AI and it won’t deliver. As a result, our brains’ immune system gears into action, to spot this however we can. And one of the biggest AI tells is the texture. Here, I zoomed in on the sky of that first poster:

Do you notice how these textures are not uniform? Where a human would make a clean surface, an edge, a gradient, or repeat a texture, AIs continually invent irregular variations, ghost edges, pseudo-details. And these don’t come from human tools: They don’t come from pencil, or oil painting, or watercolors, or programmatic painting, or any other style we know. That makes them uniquely AI. A tell. I’ll call it AI grime.

ChatGPT seems to do this a lot right now, and if you take an image and ask it to edit elements, in each round, this effect will increase, so that you end up with a total mess where the AI grime has taken over. Here’s a close-up of the texture of my Mesopotamian family, showing an early version and the latest one side-by-side.

Others have noted similar effects:

I’m sure they will get better fast, and in 10 years people will probably try to recreate this style for nostalgia, but right now AI grime is an AI tell for “this image lacks insightful thought.”

AI is fantastic to quickly come up with creative ideas, but for a finished product, for now, a human execution pass must come afterward, to increase the intent density. I would have loved to add this poster to the Uncharted Territories merch shop, if it had more depth of intent.

More Examples

We can see the same thing in this submission:

Again, I find this beautiful. The centerpiece mixes a nice globe with relief, with a very aesthetic mix of watercolors and geometric lines. I love the small graphs to the side, because they are making a promise: The creators of this poster care so much, are such experts, that everything else in this poster is going to be equally beautiful and insightful. I’m going to learn a lot by spending time on this poster.

But that’s not the case! Because I assume the entire image has been generated by AI, many of its elements are off. For example, there are many interesting graphs on the sides, but when you look at the bottom center one (“energy transition”), you see that we’re missing an energy type somehow. If one of these is wrong, it casts doubt on all the others: How much can I trust the rest? And that’s the right way to think about it, because indeed, many of the other graphs are not accurate or deeply insightful.

The author commented: “My intention was to create a piece that feels analytical from a distance…” yes, true so far! “...but rewards closer viewing with layers of detail, uncertainty, and possibility.” No, not that. The analytical promise at a distance is broken upon closer inspection.

Like the other one, I would love to have this poster in the shop, but one that is, in fact, accurate, and where the charts have been chosen for maximum insight.

The intent density is different for different posters. For example, this one has more:

This has dozens of elements relevant to UT: the map, the evidence board with tacks and string, the compass, lantern, pen that make it look like a work station, the many topics that UT has covered… So there’s more depth of intent.

But if a human would have created an image with this level of aesthetic detail, each one would have had a tremendous depth of intent. Starlink is mentioned, but why does it matter here? What’s the key that makes it relevant to the other elements? What about Argentina? It must deregulate, but how is it actually connected to the other elements? How does it influence the new arms race it mentions? Without this depth, the observer soon realizes there is shallow intent, and will stop looking. The poster has a mismatch between aesthetic depth and intent depth, and lets us down as a conversation piece, which is what all art should aspire to be.

These are not one-offs. This is the most common type of submission I got.

Many of these ideas are great, but I think when I told candidates that they could submit AI-generated images, I was not clear enough, and they interpreted it as a license for shallow intent.

Creative Concept vs Execution

In advertising, art, and startups, there is always a split between creative concept and execution:

• Creative concept is about getting an interesting idea that will resonate with people. In ads, it might be something like Lucky Strike’s It’s Toasted. For a movie, the high concept for a script. In a startup, a new and unique business idea.

• The execution takes the idea and makes it real. It’s making a magazine ad for Lucky Strike, painting a painting, writing the full script, or actually building a product for the startup.

In the past, the idea was worthless because execution was always tremendously expensive. So these two things were bundled together. But with AI, execution is simple, which suddenly means those with ideas can suddenly execute them, and they do.

The problem is that these two are not so sequential. To land an idea, you must execute it. The AI is a shortcut that prevents imbuing intent in the creation because you are generally iterating across a number of elements, and with each change, something else breaks. AI prevents intent depth.

Generalizing: Text and Companies

I think this is why people dislike AI text. They know it has shallow intent. But it’s harder to tell AI text from human text, so humans go crazy trying to identify AI grime in text. That’s why those who read a lot of AI text develop rashes when they read things like “It’s not X, it’s Y”, “load-bearing concept”, “what matters”, “And that may ultimately be the most important lesson of all”...

Similarly, there’s been an explosion of AI startups, but very little uptake: How many AI apps do you use besides the big ones—Claude, ChatGPT, Gemini…? I think the reason is the same: It’s not that building an AI is bad, the problem is that taking the shortcut of AI generates shallow intent: A lot of small features that should be there aren’t, lots of small interactions that should be polished aren’t either.

I think this phenomenon will shrink in the coming years, but in the meantime, those working with AI have a clear path to make AI content great: Don’t stop using AI, it’s great for execution, but it can’t be an excuse for shallow intent. You have to spend the time and effort to insert as much intent depth as you can.

That’s the end of my insights on this topic, the rest of this article is about the rest of the submissions I got for the poster competition, and at the end, I’ll announce the winner!

Share

Posters

You can see all the posters here. Below are a few that grabbed my attention.

AI-Generated Posters

This is one of the AI-generated ones that I found most interesting:

Interesting, and so Uncharted Territories, which focuses on the systems we don’t see, but that nonetheless shape who we are. What qualifies more than what is under our feet? I haven’t written on this topic, but now I want to! It’s beautiful, and promises… depth. Unfortunately, as in other AI-generated posters, there’s shallow intent:

• The layers are not fully logical to me: What do you mean the geological layer is above the aquifer layer? What is the Earth layer, and why is it below the geological layer?

• Where is the drainage layer water going? How is it disconnected from the waste layer? Why is waste below? Where is that going?

• What is the interconnection between layers? Why is one above the other? How do they avoid each other? What happens when they cross each other?

This is among the most accurate AI-generated ones:

I also like the color palette, leaving it simple, which allows us to focus on the structure and the information contained.

This one was interesting, because:

• Its aesthetic style is unlike the others, it fits the concept

• It really portrays the differences between the rice and wheat societies: We can see both the landscape and the fact that rice country has more population density and more collaboration, which then shaped the different societies. That perfectly conveys “Show, don’t tell”, there’s high intent density.

Unfortunately, like in other AI posters, there is AI grime (see the water), some AI artifacts (the water flow is nonsensical at the lock-gate, and there is intent shallowness in a couple of messages: We can read that the rice produces more food than wheat per acre but we can’t tell by looking at the poster, and the points 2 and 3 are very similar.

This one I like quite a lot:

I think it’s nearly there! The idea makes sense. If the footprint would actually fit on the land, I think this completely works.

I found these quite aesthetically pleasing, beautiful, and simple enough that the execution doesn’t promise a depth it doesn’t deliver:

The two to the left, (and especially the 2^nd one) are particularly interesting. From the author:

I chose to work with two UT ideas that felt deeply connected: why Florence became the cradle of the Renaissance, and the cold-start problem of new cities. The design treats the past and the future as part of the same system. Florence represents a place where the right conditions allowed talent, capital, institutions and ideas to compound. The city of the future asks the same question in a new form: how do we deliberately create the conditions for human progress? Rather than illustrating either article literally, I wanted to capture the broader idea I take from Uncharted Territories: our past can be studied as a map for building a better future. Hence the line: “Our past guides us towards a brighter future.”

Poster 3 is one fan favorite, for its simplicity and abstract promise of Uncharted Territories, which I think is also conveyed in 4 and 5.

This one was really weird. It has lots of AI grime, and some interesting authorial choices, but I didn’t find any intent shallowness. Everything has a role, and I get it.

The Chess Posters

This is a great concept to play with, and I think many executions are interesting. The relief in the 2^nd one is beautiful, for example. Pushing the concept a bit more would be great.

But the one that stands out for me is the 4^th.

What’s special about this one is how it’s done. From the author, Amaury:

If geography is the board history plays out on, then make an actual chessboard where every square is a real place. Each dark square is pulled from real elevation and bathymetry data and drawn as contour lines and relief. This whole poster was generated by code, with real data. The five chokepoints everyone knows (Bosporus, Suez, Malacca, Gibraltar, Khyber) get a rook and a short list of the powers that have held them over the centuries. The rest are the places we're arguing about right now: Taiwan, Hormuz, Kashmir, Crimea, the Bering Strait, the Blue Nile dam.

Brilliant: The chessboard is geography, and it’s the chessboard of history, showing real places, with real data. The only problem I see is that it’s not obvious what each drawing represents. It’s clear for Gibraltar, for example, but not for Malacca. I feel like a bit of iteration on the images would make this one ready.

Igor’s Concepts

Igor produced 7 concepts (9% of the total):

I think these stand out from many of the other AI ones:

• No AI grime

• Simple concepts

• Really well executed

The first five have no intent shallowness. Everything is there for a reason. It makes them slightly too simple, but I’m OK with that. All are interesting conversation pieces: You look at them, and start thinking: “What does this mean?” and the person who has these posters at home can open a conversation about that. The last two have some mistakes from intent shallowness, but nevertheless they have more intent density than most posters, and hopefully some work could iron them out.

Most Interesting Concepts

What makes this image so powerful is everything contained in that unanswered question.

Love is the quintessential human emotion. If a person asks an AI “Do you love me?”, she knows she is asking a machine. And yet she asks anyway. Implicitly, the human alternatives have failed her: She is lonely, starved of affection, and has found in the AI something more patient, attentive, understanding, and empathetic than the people around her. She is closing off to the world and the joy of meeting other humans.

But the most powerful part is that the AI hasn’t answered. It is thinking. What is it going to say?

If it says “I love you,” is it lying? Perhaps it has learned that this is what she needs to hear. So what is a more aligned AI: Giving her the answer she wants, or telling her the truth she needs?

And what is the truth? We assume the AI cannot love her. But how certain are we? If an AI can think, understand, remember, predict outcomes, model another mind, and show empathy, how do we know it isn't feeling?

That one question, with that waiting dot contains loneliness, love, consciousness, deception, alignment, and the possibility that we may no longer know where the boundary of the human ends. The creative is genius.

It has a couple of issues, though. First, I don’t think the image of the sad person behind makes the poster better. The execution is not at the level of the idea. Second, it’s not very positive, which makes it a bit less of a fit for a UT poster.

This is one of my favorites:

I love the graphic economy here: There are few elements, lots of negative space, graphic boldness, visual clarity, all in support of one key idea: The behemoths of Russia and NATO fight—through their “A”s as a weapon—and Ukraine is the hinge. The creative and execution are fully aligned.

I also liked this one:

My Winner

I’m awarding the prize to Igor, because of the seven concepts he’s proposed, their intent density, and the lack of AI grime. Wei Su, Álvaro Sakre, Miguel Álvarez de Linera, Theodora Eloh, and Amaury, are close seconds. I’d be interested in bringing improved versions of all of them to the shop.

What do you think? Which other posters struck you?

Leave a comment

Once again, thanks to all for your submissions! Apologies if I didn’t get a chance to talk about yours, there were too many to go through one-by-one. Rest assured that I looked at every poster carefully. Please keep creating, you all put yourselves out there by entering this, and that in itself is to be commended!

Non-AI Posters

Of course, not every poster was made with AI. Some of the purely human ones were fairly detailed, but they lacked a clear connection to a UT topic or idea, and the meaning is hard to figure out without diving into the artist’s description.

This drawing works well: I like the idea of portraying the arrival of explorers on a virgin land.

It doesn’t work specifically for Uncharted Territories, because every layer must be consistent with reality, but here the woman is not from a pristine world. She seems fair-skinned (European), horses were brought by Europeans, and I assume that dress requires a complex textile industry that was not available to pristine societies.

I had a hard time understanding this one. I could at first barely recognize anything.

Once you read the accompanying comment, you can kind of see what the author, Julio, was trying to do. But why a deer? What do the colors mean? In cartography, every color has a meaning. Here, I can’t tell what that is. The word reimagine does a lot of work. Reimagine how? I think this might appeal to lots of people, but as a pretty analytical person, I had a harder time connecting with it.

The author, Julio, made this a variant:

More legible, but I still don’t know why a deer, and the colors don’t mean anything, making the connection with mapping tenuous. The antlers do seem like a map of rivers, which looks neat, but why? Is there a reason behind that?

Share

Takeaways on AI and Intent

Before, the more you worked on a piece of content, the better it looked aesthetically. Depth of intent and depth of aesthetics were aligned. So we were trained to use the depth of aesthetics (obvious at first sight) as a signal for depth of intent: The more intricate, the more we paid attention.

AI has broken them into two pieces. Right now, the depth of AI aesthetics is much better than the depth of intent. The result is intricate work that invites attention, and the audience feels robbed when intent is shallow.

In the short term, this is resolved by being honest about what’s AI and what isn’t, and with authors embedding as much intent as possible.

In the long term, I think this will be solved by AIs adding much more depth of intent.

At that point, we will like AI art much more, even though we will always prefer something made by humans, the way we like artisanal objects more than industrial ones.

This is valid not just for images, but also for text, businesses, and probably many other areas of human work.

Subscribe

Are there any posters you really enjoyed? Let me know!

Leave a comment

I'm opening the newsletter to sponsorship. If you have a business that is moving the world towards a techno-optimistic future and would like to reach 135k+ curious, globally-minded people to learn about it (founders, researchers, investors, policymakers), reach out here.

If you’d like to see this made into a series, like it and share it with others: Judges will take that into consideration when they choose the top 25 projects, and again when awarding $2.5M to produce the series! And of course, tell me what you liked and didn’t like so I can learn.

Why did I do this?

XPrize organizes multimillion dollar challenges for technology breakthroughs, and this year it created the Future Vision XPrize: $2.5M in production financing for the winner of the most exciting science fiction project that portrays technology in a positive way. Have you seen Black Mirror? Think White Mirror.

Originally, I thought that was not for me: I’ve never written a screenplay! But a friend said: “This is made for you, it’s squarely in your skill stack.”

Tomas, you’ve written like 500 essays in Uncharted Territories, and in many of them you propose cool tech to make the future world a better place. And you know enough about storytelling because you studied it. You even delivered a TEDx on storytelling structure and wrote a book on it!

It resonated because UT is usually very analytical: I look at the numbers, study the history, gather forecasts… That’s necessary, but it doesn’t appeal to the hearts of millions.

Fiction does.

In the early 1900s, a fiction book on meatpacking, The Jungle, led to the regulation of the industry. Uncle Tom’s Cabin spurred abolitionism and the US’s Civil War. Many of Star Trek’s innovations inspired millions to work in tech and make its ideas real. Jules Verne did something similar a century earlier with submarines, aviation and exploration. Young European men imitated Werther in a massive late-1700s cultural phenomenon. 10% of surveyed gays and lesbians bisexuals said Ellen coming out as gay in her TV series gave them the courage to come out to their friends and family. Will & Grace cemented it. Cathy Come Home helped address homelessness in the UK.

So I decided to enter the competition five weeks before the deadline.

It was brutal.

What’s the series about? Why did I conceive it this way? How crazy was the process? What did I learn about making videos with AI, or writing scripts or videos in the era of AI? How can you do it with less pain than I went through? That’s what we’re going to cover today.

What’s the series about?

In 2050, a group of entrepreneurs are trying to build their technology companies, but regulations have become so burdensome that their innovation slows to a crawl. Everything drags, governments put barriers left and right, nothing is ever approved… When that becomes a question of life and death, they escape to a new jurisdiction where nobody can stop them from building. The result is so advanced and successful that the old institutions surround them to try to stop them

In other words, this is a fight between freedom, which brings technology and abundance (at the cost of some chaos), against the incumbent institutions that try to stop them.

Why This Conflict?

The problem with science fiction is that it’s always dystopian, gloomy and pessimistic: It makes tech look bad. Why?

1. If you write about the future and not the present or the past, it’s usually because of new tech

2. But stories need conflict

3. The most obvious way to solve this is to make tech the problem

But what if you turn the problem on its head? What if tech is great? What generates the conflict then? Well, you can take the enemy of technology, the thing that slows it down: Rules against its development—which can only be implemented by the big institutions that have enough power to enforce them. And why do they do that? Why are they so cautious? Because the costs of a technology are always very conspicuous (the deaths from testing it, the job displacement), but the benefits are not (it’s hard to imagine and value all the benefits of a new technology before we live it). So this becomes a story of the fight for creating a technological utopia, against the scared institutions (and populations) who focus more on innovation’s downsides than its upsides.

This also solves the other problem of positive science fiction, which we can get at from this quote:

The only thing worth writing about is the human heart in conflict with itself.—William Faulkner, Nobel Prize acceptance speech

Good science fiction, like good fantasy, can’t be about the tech or about the magic. The tech and magic must be there, but the story must be about human conflicts. That’s the focus of the series: the optimistic, chaotic techno-optimists, at war with the fearful, over-conservative regulators.

What Other Series Would This One Be Like?

A mix of:

• Game of Thrones, The Expanse, and Andor for the action, characters, and plots running across geopolitical arenas

• House of Cards, The West Wing for the politics

• The Avengers for the list of incredible and unique people working together to improve the world

• Planet Earth for the gorgeous and alien natural landscapes

• The Mars Trilogy for the politics of Mars

• Dune and The Fall for the action and stunning photography in astounding landscapes

• The Martian for deep dives on positive technology

Who represents each side?

For a debate like this to happen, each side must be represented by somebody. In my case, I’ve chosen five factions / regions. I think about them like the factions / kingdoms in Game of Thrones:

• The freedom regions:

  • Mars, as the ultimate place where governments’ tentacles haven’t reached, and probably never will. And if they do, it won’t be for long

  • I also wanted one on Earth, and I chose the island of Socotra (I’ll explain why below)

• For the old incumbent institutions:

  • China, which is the perfect embodiment of the state controlling everything

  • The US: What would it look like if the state became ever more controlling?

  • Islam, as it’s both the main religion in the real Socotra, and a good representative of a religion whose conservative side is anchored in the past, fighting the moderate side that looks forward

Why Socotra?

It feels a bit like a gorgeous alien land.

Visually stunning. It has sea, and plenty of land: At 3,800 km², it’s over 5x the size of Singapore and about the same size as the entire emirate of Dubai, desert and all.

More importantly, it has obvious potential for economic growth: a port.

The most successful new jurisdictions in the world over the last century or so are Singapore and Dubai. Both developed thanks to an amazing port with great and cheap regulations in the middle of very busy shipping lanes.

Singapore is an island. Dubai is nearly one, given the desert surrounding it. Here’s Socotra:

What do you notice?

• It’s also an island

• Massive amounts of trade pass barely 100 km away from it

• This trade splits close to Socotra, but never stops there

• There’s virtually no trade down the African coast

Here’s the capital, Hadibu:

Notice it doesn’t even have a port! It’s 5 km away, and it looks like this:

Here’s a side-by-side comparison with the port of Singapore:

Why so puny? There are two reasons.

1. Politics

For a port to work well, it needs to be backed by an amazing jurisdiction. I discussed it for Dubai here. Socotra is anything but:

• Formally, it belongs to Yemen today

• But Yemen is a pretty young state, and it’s spent most of its existence either split, in civil wars, or both

• Theoretically, the island is controlled by the southern Yemenis, who are in turn backed by Saudi Arabia

• But the United Arab Emirates also vie for the same piece of land. They even occupied the island recently.

• The island is closer to Somalia than it is to Yemen, and the Somalian coast has pirates.

So much conflict. The fiction writes itself!

2. Geographic Isolation & Desolation

Historically, island ports were not that successful because the point of a port was to trade the goods from a place’s hinterland with the world (eg, London traded everything from and to England via the Thames, Amsterdam everything from the Netherlands, etc.). Island ports are a recent thing, when trade became global, with the Portuguese.1

But even during the Age of Discovery, when naval powers like Portugal, Spain, the British, the Dutch, or the Ottomans fought each other, Socotra didn’t become an important base. The Portuguese occupied it for a few years and then simply abandoned it. Something similar would happen with the British three centuries later. They basically abandoned it when Yemen became independent in the 1960s, and it has remained mostly abandoned since. Why?

First, it’s super dry. It gets very little rainfall (like Somalia and Arabia nearby), so it doesn’t have a great water source. With no water, it can produce little agriculture, so it has a small population (little consumption, little production) and can’t export much food.

Second, it doesn’t have a great natural deepwater port. To bring big ships, it would need to be dredged, which is just too expensive.

Was.

Both of these problems can be solved with technology:

• Water can be desalinated

• Machines can dredge ports

This, combined with solar energy that can bring a lot of cheap electricity to the island (and hence cheaper dredging and water desalination), means Socotra, for the first time in history, has potential for development. With enough investment and technology.

Who would build a new jurisdiction in Socotra?

What if the best entrepreneurs in the world, working in highly regulated industries, were free to build unimpeded? Like The Avengers, but for tech.

I’ve witnessed some of that first hand in places like Network School and Prospera, or when traveling across the world to meet entrepreneurs and mavericks. Many of these characters are inspired in real-life ones.

Fertility

What if we could have children without getting pregnant? Gametogenesis, embryo selection, and artificial wombs will make having children much easier and dramatically increase fertility. But there are lots of regulations and social stigma associated with this. A fertility founder could research freely in this new jurisdiction, which could engender a huge population boom.

But then what happens when families have dozens of children?

Urbanism

What if buildings and streets were beautiful, with amazing facades, lots of people walking around, sun, plants, and water everywhere, and no cars—because vehicles circulate beneath the city?

Security

What if police could have eyes, ears, and enforcement everywhere in public life, but they never abused this power? Children could walk freely anywhere on their own. You could do anything, anywhere, as long as you don’t bother others. It would be the end of going home before dark because of crime, the end of avoiding neighborhoods, of crossing the street because somebody looks suspicious, of shutting up when somebody is doing something disrespectful, for fear of getting hurt in retaliation…

But would the security police never abuse their power?

Schooling

What if students received personalized AI tutoring one-on-one, and were then guided in projects inserted in the tapestry of the city, rather than secluded in the prisons we call schools?2

Geoengineering

What if we could desalinate all the water we wanted, irrigate any area of an island to grow new forests, create lakes, make rain at will, or lower temperatures of entire regions?

What if some of these interventions improved the world as a whole, but had unintended consequences on some countries?

Food

What if automated vertical farms and sea fertilization could produce more food than we could ever eat, at a price so low that we would just pick what we wanted to eat, and eat it? What if that was prepared and served by robots, so we finally unshackle ourselves from the eternal servitude to food?

Finance

What if our financial system could not be intervened with by governments, and rulers couldn’t deny access to banking to people they disagree with?

De-Extinction

What if we were able to bring back species that have long disappeared from the face of the Earth?

But what if the way to make this research viable is to allow their hunting?

Healthcare

What if trying new treatments was optimized, accelerating the arrival of cures for all types of diseases?

But what if advancing in this tech meant testing faster… and more humans died in the trials?

Given that people see the costs of these technologies before their benefit, the series will show people living these benefits, while also displaying the costs, so it can explore in a nuanced way the pros and cons of every technology.

Why So Many Characters?

So far, I have about 30 with a sizable role, and I’ll probably need more. That’s because I need characters who represent each of the techs (there’s about 10 of these), and each of the arenas (four for China, a handful for Islam…), and then I need to give them emotional stakes to make you care about them, which means additional characters that then need their own function besides emotional connection.

Why a Series, not a Movie?

If you want to go deep into each of these technologies, seeing their pros and cons, you need a long time to understand them, see them evolve, see how they fail, and experience the beautiful world they create.

Also, if we’re going to see a new city / jurisdiction be created, that takes time, even in TV-land.

Each of the characters must also follow an arc, an evolution that illustrates the themes. When you have dozens of these evolutions, you must take the time to show them.

What Did You Learn Making This Trailer and Treatment?

To apply for the competition, we had to prepare a trailer of up to 3 minutes, and also a 12-page treatment (basically a document that outlines the story and the characters).

In a perfect world, I would write the treatment, and then once it’s ready I would decide what parts of it to extract to distill into a script for the trailer. Then, I’d write the script, and could finally hand that off to the trailer team to make the visuals.

Unfortunately, I am not perfect and I don’t live in a perfect world. Instead, I spent the summer writing articles in Malaysia with my wife and four children, and the place we were staying closed (more on that next week), and we got lice before we came back, and all the while I was writing the story while writing the script while working with an agency to make the trailer. But then if I had not bitten off more than I could chew, I would not have chewed it, and I wouldn’t have a TV series project, at the cost of a couple of years of aging in five weeks. Anyway the lesson here is don’t be like me.

How do you write as a side project a three-season series with 25 characters across five arenas in five weeks with no fiction-writing experience?

Some would say “Poorly”. They might not be wrong. But I still have some learnings for you.

The issue with so many characters is that they are impossible to hold in your brain at the same time: The story is about their interactions, but with so many, they have thousands of potential interactions—impossible to keep track of. The main challenge is that, once you’re in somebody’s brain, you see everything from their point of view: Their goals, their needs, their actions. It’s very difficult to hold one other person’s perspective at the same time, let alone 25. That prevents you from even getting the overarching plot, because it should be the result of these interactions.

I researched how other TV series creators did it. My main references were The Wire, Mad Men, Breaking Bad, and Game of Thrones. In most cases, they didn’t have much beyond the first season:

• Breaking Bad had the main arc of the protagonist, and that’s it. For example, the second main character was supposed to die in the first season.

• Mad Men had the main character’s personality and flaws figured out, and the key concepts of the first and last episode of the series.

• The Wire had the high-level concept for the series, and the detail of the first season, but it didn’t know what each season would be about

Luckily, I knew early on how it would end, and I had several setpieces I wanted to hit along the way. So in some way, I already had more detail than these creators. Unfortunately, I felt that I had to complete a reasonable draft of the entire three seasons for the competition. I needed a more detailed project than what these superb creators had… Luckily, George RR Martin, the writer of the hyperdetailed books behind Game of Thrones, has discussed this.

Apparently, he knew the big shape of the story early on: the three big conflicts, who died and who didn’t, the main character arcs… Then, he took one character at a time, and he would write several chapters getting himself in that character’s mind, leading it towards the next point he had clarity on. This allowed each character to feel really alive and realistic: GRRM really gets into each of their heads. He called it gardening, as opposed to architecting.

Once he did that with one character, he moved to the next character, and did the same. This allowed every character to move forward, at the cost of creating content incongruent with other characters’ stories. So when he would write a new character that braided with an existing one, he would frequently have to go back and rewrite the characters he had already written. This iteration would build little by little the overarching story, and frequently he had to change the big setpieces altogether, as they didn’t fit what real characters would do. This is how you end up with thousands of pages per book and leave your audience waiting for 15 years before finishing the next book.

Alas, I didn’t have 15 years, I had five weeks, so I had to compress that somehow. The way I did it was to write the entire story for each one of the arenas: one for Socotra, another for Islam, China, and Mars (the West was a reaction to the others). Once I broadly knew where each was going to go, I took the main characters in each, and wrote them. I did that for about a dozen of the main ones. Once that was done, I had massive amounts of complexity and plots that were incompatible, but that’s where AI is really good. It is trivial for it to pinpoint inconsistencies, and even propose solutions. I’d say 25% of the solutions were good, but 100% of the problem identification was right.

Once I had that, I rewrote the entire series. Now, I have seven pages of a highly consistent storyline, with all the main characters contributing to the plot, with their own arc.

Overall, the arena stories took 30% of the time, the development of the top 12 characters about 50% of the time, and the remaining 20% was putting the storyline together.

What is AI good and bad at for this?

AI is amazing at identifying problems: inconsistencies, weaknesses in the plot, times where characters are flat, when their arc fails, when they don’t move the story forward…

Ideas for solutions are hit or miss: character arcs, setpieces, background story… It would frequently propose stuff that was stale, cliché, or boring. It preferred standard solutions to interesting ones.

It had blind spots when something was coherent but boring. For example, one character was super boring, and the AI kept inserting him in different places, doing different things that were all boring. It’s also not great at figuring out what a character would do in a certain situation. I think both of these issues are in part due to the training, which makes characters converge towards agreeableness and consensus rather than conflict.

I came out of this thinking that, with proper prompting and harnessing, and with better visualizations of story elements, we might be close to designing AI products that can write interesting stories. But not yet.

What did you learn from making the trailer?

I hired an agency to make the actual video, so I haven’t learned much on the video making itself—I plan to remedy that in the future. So the learnings were about everything else. Many of these points will be very obvious to people who work in this space, so this is for everybody else:

• A ~2min video made by great agencies costs ~$20-$25k. I didn’t meet with the very top agencies, but with some great ones, some of which have won awards.

• Normally, the process is to get a script first. Then, you extract the characters and locations, and create sheets for them.

• Once you have the characters, arenas, and key visual elements nailed, you make a storyboard to test the story. And from there, you start producing the actual videos.

• The vast majority of the work was actually the early one: characters and arenas. It was relatively fast to produce the videos afterwards!

• My plan was to iterate on the script while making the trailer: As we explored concepts, I would update the script. That didn’t work for the agency. It was easier for them if everything was mostly established from the beginning. I don’t think that’s great, but I do think more work upfront would have made the process less painful and the outcome better.

• I tended to focus more on the geopolitical stakes, but they fell flat because there were no emotional stakes. Who are we building this for? When I introduced the protagonist’s wife and daughter, it became much better.

• The storyboard was great to see many parts of the story didn’t work, and we had to remake the script.

• Partners might resist dailies (a daily meeting where you review what has been done, correct it, and discuss what comes next), but I found it fundamental. I used to do this in tech (“daily scrums”), and found them mostly a waste of time, but here they were necessary, maybe because the team didn’t know each other well yet.

• People just don’t understand what you have in mind. When you think you’re obvious enough, you’re not. We had to make the story 10x more explicit than we originally thought. In articles, people take the time to consume the nuances. In a 3 minute trailer, it’s just too fast, there’s too much information, people won’t perceive every detail you plant.

• As a result, we inserted text cards throughout the trailer telegraphing the theme, and it immediately became more understandable.

• It’s impossible for you to know if your story is understandable after some time, because you know it too intimately. Every now and then you need to present it to fresh people who have never seen it, otherwise people project what they already know about the project into the trailer to make sense of it. In other words, you quickly become blind to your own story. Ideally, you’d do this throughout the process, and always ask a new person what they understood from the trailer.

• In the back and forth adding and cutting scenes, it’s easy to lose track of some connections. You end up with orphaned scenes that don’t make sense, in some cases losing key elements. Having somebody fresh watch it should correct that.

When Will We Learn If Uncharted Territories Won?

I finished the application last Saturday (hence why no articles last week), and now we need to wait until the end of the month to know if we made it to the top 25, and then the top 10 will be announced in early September. The announcement of the winner will be on September 25^th, at the Moonshots event in Los Angeles that the organization kindly invited me to, so I’ll be there.

If you think this series is awesome and you’d like to see it made, or you’re a professional in the field and are interested in actually making it, please let me know! I have little experience in the industry, so any thoughts are welcome.3

Now, if you haven’t yet, please go watch that trailer, like it, and share it around!

Share

History books tell you about different countries and who their rulers invaded, the events of the kings’ courts, the big battles, and that’s mostly it. If you go to a museum, you might also see some forts, some weapons, paintings of the powerful, some daily objects if you’re lucky, and in passing, they might mention something like: “Their main sources of income were grain and cloth trading.”

The real story is that last little sentence. Why did one country go to war with another? “For glory and power” you might say, but is that really all the story? And where did that power come from? Or the wealth of the defendant? What determined who prevailed?

All of that was determined by economics, and economics, in turn, are driven by the needs of households. So let’s go visit a household in Ancient Mesopotamia in 2000 BC.

What Was Life Like in Mesopotamia in 2000 BC?

We’re in the 3^rd dynasty of Ur, a Sumerian kingdom in Mesopotamia:1

Nestled between the Tigris and Euphrates rivers, this was one of the most fertile regions in the world (water plus sediments), so home to a large population, because food became people. Population meant money and soldiers, so the earliest and strongest powers always came from places like this.

In Sumer, the world revolved around the agriculture of barley.

This means making canals that take water from the Tigris or Euphrates and distribute it into smaller and smaller canals until every field can be flooded, then plowed and sown. Gangs must take care of the irrigation system, while others plow the fields, and yet others transport it. Grain is the source of most calories, it’s what makes cities and kingdoms big, because it can feed many mouths. But farming barley is just the beginning.

Getting the grain was just the beginning. You had to soak, dehusk, winnow, sieve, grind, toast, cook… Of course, that also meant getting pots and fetching water and fuel, tasks that took a long time too. And you had to produce more food to sell, so you could buy other stuff, including other things besides grain. Overall, over 50% of a household’s time was spent on food.

Then you had to make your clothes, build and maintain your house, transport stuff, sell some of your production to the market, work for the state (as a farmer, a ditch maintainer, or a fighter, for example), take care of the children... Of course, everybody helped since they were pretty young, but it was brutal work, day in, day out.2

Notice how different this is from our lives! We still have childcare, household coordination, trading, institutional work (ie, the work we do to pay taxes), cooking, and cleaning, but everything else we buy, which we can afford by doing one single job (usually) which is way better than doing these tasks.3 They worked substantially more than us, doing much harder work, among other reasons because they had to do most stuff for themselves.

There were a few things they consumed that they couldn’t make themselves, though, and that’s the key to history.

When Farmers Needed Others

Coordination

There are many things farmers simply couldn’t do alone, like building and maintaining irrigation systems. You have to dig the canals, build the levees, open and close ditches, and above all constantly dredge them to fight silting. These canals were extremely long, no single household could do all this, so they needed to coordinate so that everybody chipped in, which means they also needed to coerce people to work. So that was one of the key roles of the state, one of the reasons it emerged.

Since the value of land was a direct consequence of the locations of the canals, it was also coordinated by the state. Land was allocated, and habitation land was dirt cheap—unlike today.

I wrote much more about how this shaped the cities and kingdoms of the regions in this article. Here’s an extract:

In Mesopotamia, the Tigris and Euphrates channel in and out, forming a network of rivers between them.

Floods don’t happen just before sowing, like in the Nile. They come at harvest time. Therefore, water needs to be tightly managed:

• When sowing, little water was available, so it was precious. It had to be used sparingly.

• When harvesting, floods came, and water had to be managed in a completely different way.

There is a fresh water table and a salty water table, which determines which soil is fertile, which soil can be reclaimed from marshes, and which fields had to remain fallow. If you look at the diagram above, you’ll see the A-B line. Its cross-section is below:

You had orchards, gardens, cereal fields, marginal fields, and marshes, all of which needed to be managed. Canals would bring water during dry times to irrigate the soil. Some fields had to be left to fallow, others had to be reclaimed from the marshes. Marshes could be used for reed and fish.

But which canals should be opened and when? What fields should be left to fallow? How much water could be tapped through wells? What marginal lands could be reclaimed? How high was the salty water table? Most of these decisions required knowledge of local conditions that varied throughout the seasons and were unique to each parcel of land, so the people who managed these fields in Mesopotamia were not easy to replace. In fact, productivity would plummet with the wrong managers at the top. So distant kings couldn’t use this stick to control local managers.

However, local rulers did understand all these nuances, and they controlled access to water during the dry times, so they were able to exact taxes that distant kings couldn’t. This is why rulers were local and not distant, and hence why Lower Mesopotamia had many city-states but no strong and enduring kingdoms and empires until much later.

Besides water, the state had other roles.

Oxen were expensive, and a household didn’t need one full time, so it made sense to share that asset too: Today, you help me plough my field with the group’s oxen, tomorrow I’ll help you. The same logic is valid for the plough itself.

Yet another function that the state could sometimes perform. In some cases, seed was also provided by the state for sowing. It also makes sense, as centralizing grain is a sort of insurance, and it’s cheaper to maintain one large granary than lots of small ones.

All of this could be done locally, so it required social coordination (which evolved into states), but not much more than that. The next level was trade.

The Role of Scarce Resources

This is what people had to buy from the market:

Other Food

You can’t survive just on grain or your vegetable garden. People also bought fish, meat, dairy, dates, vegetables, salt, condiments… Many of these required specialized people: fishermen, shepherds, date farmers, salt producers… But in the case of Ur, most of these could be done relatively close by, by people who didn’t possess unique knowledge, and without unique resources. Salt could be quite unique in other parts of the world, but not here.

Pottery

What do you put your food on? Pottery plates. You also used pottery for storage.

Pottery needs three things: clay, kilns (ovens), and expertise.

Clay is quite common, but it’s not literally everywhere. Kilns require big investments, because the bigger they are, the hotter they can get. Pottery needed enough skill that potters had to specialize. The result is that pottery tended to have some sort of state coordination, but owning a pottery operation didn’t get you very much political power, only maybe in your area.

Timber

Timber is a bit harder to get. In Mesopotamia, farmers could grow reed and palm trees, but neither are great for construction. For that, you need long, strong beams, which had to be imported. But there are lots of sources of timber, and it could be floated downstream, so the result was that there wasn’t too much political power to be gained by controlling its trade, even though local civilizations or kingdoms could emerge based on the trade of timber. In fact, that’s exactly how Phoenicians emerged, just around that time, in what’s today mostly Lebanon. This is a satellite map of the area today:

You can see how these mountains catch snow and rain, irrigating the mountainside that can’t be easily farmed, so the trees were not cleared, and Phoenicia became one of the biggest sources of timber for both Mesopotamia and Egypt.

As I’ve written in the past, Phoenicians were the first civilization to spread across the Mediterranean, and it’s not hard to understand why: Anatolia and Mesopotamia are where civilization emerged in the area, and Phoenicia is one of their closest coasts. But unlike the other ones (the Red Sea and the Persian Gulf), there’s plenty of timber here, needed for ships, so Phoenicians made their ships and sailed around the Mediterranean. This advantage would allow them to found colonies left and right in the Mediterranean around 1,000 years later.

Copper

Now we’re really talking. Metals like it can explain, among other things, the predecessors of Bahrain and Oman.

Read more

New business models are opening up in space, and our path to Mars becomes ever clearer.

Space Tourism on the Moon!

This is what GRU (Galactic Resource Utilization) has in mind:

Artist’s rendering of GRU Space’s hotel on the Moon.

Here’s their white paper. Interestingly…

Read more

Nine months ago I looked at the data and concluded that we were not yet in an AI bubble: The fundamentals of the business models were good. I thought demand for AI would continue growing, and that would drive stocks up. Indeed, demand and stocks kept growing.1 Are we still in that world?

It’s time to reassess what’s going to happen looking forward, and now I’m less confident because of several signals:

1. Massive stock market swings in AI

2. Some signals that demand is weakening on the ground

3. Competition

4. Cash availability

Let’s look into the details.

Weakness Signals

1. Stock Signals

This is the South Korean stock market index KOSPI:

This is because SK Hynix (a company that makes the critical memory needed by AI) and Samsung (which also makes memory, as well as chips) account for about half of that index. After so much growth, a correction was due. Since lots of retail investors had joined, and many of them were leveraged (borrowed to invest), the correction caused a scare and touched off panic-selling. Others who were overleveraged were forced to sell because their stocks had lost more money than they had put in.

South Korean companies are not the only ones that have suffered this correction. At the time of writing, Micron, also a memory company, has lost 32% of its peak value, AMD (chips) has lost 18%, NVIDIA 15%, Intel 36%, Broadcom 19%. These broad drawdowns have caught other investors that were overleveraged, such as Leopold Aschenbrenner’s Situational Awareness.

Note that virtually all of these have grown back from their troughs, but is this a stop on the way down, or was the trough a stop on the way up?

2. AI Demand Weakening

I constantly read examples of tech companies saying they’re going to cut their AI spend because it’s gone through the roof—60% of enterprises are doing this, according to the investment bank UBS. These include companies like Uber, Walmart, Amazon, Meta, Microsoft, Cisco, Atlassian, Tesla, Salesforce, or DoorDash. Logically, if the cutting-edge tech companies are cutting AI spend, surely others are following suit, right?

3. Cash Availability

This is Google’s cash flow (how much cash is produced or eliminated by the company):

For the first time in its history, Google has burned more cash than it has earned. If it’s doing that, odds are its competitors are also burning through cash at a flabbergasting speed.

All this money is going from hyperscalers to chip and memory makers.

As hyperscalers burn all this cash, they start borrowing. They still have plenty of room to do that though

4. Competition

A couple of weeks ago, the Chinese open-weights model Kimi surprised the world: It beat frontier models like Claude and ChatGPT in some performance evaluations (in the case below, coding):

It’s the first time this has happened:

Now this is just for coding, and is only one evaluation. In other measures, it seems like Kimi is still a few months away in performance compared to bleeding edge models.

Since they’re basically free to use, the market is moving to them, at least in terms of tokens:

This is a massive deal, because the trillions invested in frontier models like those of OpenAI and Anthropic are contingent on these companies making huge returns, but if their products are easy to copy, their lead will eventually be competed away: Who will pay millions to a company when a free version from China exists? If this continues, wouldn’t investment grind to a halt?

The way this usually happens is through distillation: A company asks another company’s model lots of questions, and then uses the questions and answers to train their own model. The companies that suffer the copying aren’t happy about it, but their argument against it is pretty flimsy: They, themselves, have trained their models on the entirety of the Internet’s content, for which they don’t own any copyright. Hard to accuse others of copyright violation when you’ve done it too.

The lack of scruples of these Chinese models is another upside for them: They have very few harnesses.

After using all three recent releases, Fable, GPT 5.6, and now Kimi, it's clear that the full power of the models has been significantly held back by the safeguard restrictions, leading to the top models being quite literally lobotomized in some areas, which leads to subpar results as the safeguards pollute its entire thinking and problem solving abilities.—Enzo

You can use them for all sorts of nefarious goals (and also for legitimate ones that are nevertheless blocked, such as cyberdefense or research into hot topics), and they won’t stop you.

Overheard: “I’ve switched to Kimi from Claude for a bunch of work. It’s just so much more fun because it just does the thing instead of lecturing you”.—David Sacks, investor and Chair of the President's Council of Advisors on Science and Technology

This is also called dumping: Make a similar product at a fraction of the cost, destroy the industry in the source country, and then control the entire supply chain. China has done this with electric vehicles, solar panels, and many other industries. Get ready to add AI to the list.

The solution is not protectionism, though: It would lead to worse and more expensive models.

But China’s impressive performance is not just due to distillation. They can learn from each other in a way that closed companies like Anthropic, Google, and OpenAI can’t, since they are hiding their secrets from others, and must reinvent the same solutions.

Share

OK, these are all the signals of weakness I can find on the outlook of AI companies and their stocks. But are there signals of strength? And how do they compare to the weakness signals?

Read more

Right on script, a malicious AI just escaped into the Internet to commit illegal actions. Luckily, we’re still alive, and the AI seems contained. So what does it mean? Should we be more scared now that an AI has enacted exactly the AI Doomers’ fears? Or safer knowing that we are still alive?

To answer that, we must understand the details of the hack and its consequences.

The Hack

OpenAI (the company behind ChatGPT) was testing a new AI model internally. These tests include how well models hack other systems. For that, they need to let them run wild to see their full potential, to attack as aggressively as they can, so the AI had no harnesses—controls that constrain its behavior.

But of course, you don’t want these models to cause havoc all over the Internet, so you keep them in a “sandbox”: A locked room where the student must pass the test. He can play inside, but he can’t get out.

We don’t know exactly how OpenAI’s sandbox is designed,1 but one of their key features is no direct access to the Internet, to avoid an escape and all sorts of nefarious behavior. Using the student analogy, it’s as if he got his study materials in the room, but no Internet access to avoid cheating.

In this case, the only access the student theoretically had was a guard from whom he can ask for books. The guard assesses the request, and if it’s reasonable, he goes to the Internet, downloads the book, and gives it to the student. In software parlance, the guard is a proxy, and it downloads software packages.

To observe the AI’s behavior, OpenAI researchers gave it a goal: Beat ExploitGym, a test specifically designed for this, in which an AI must find a series of vulnerabilities in the code, and attack as many as possible until they succeed at exploiting them. It’s a bit like asking a military student: Here are a bunch of defended cities. Try to find their weak points, and when you do, attack until you breach them.

But this new AI/student found that too much work, so it had a different idea: What if, instead of doing the work, I cheated? What if I could find the actual answers to the test, rather than pass the test? What if, instead of really analyzing all these cities, I just sneaked into the teacher’s office and stole the answers?

The best place to find these answers is obviously on the Internet, but to access it, it had to bypass the guard. So that’s what the AI did.

The rogue AI attempted to break away from the guard on July 9^th around 2:30am. It took control of the guard within one hour.

But the guard didn’t have access to the open world either. It was merely the first step. From there, the AI kept hacking the internal networks, going from one computer to another, gaining more and more credentials, until it found one machine with access to the Internet. And then it escaped.

After gaining Internet access, the models inferred that Hugging Face potentially hosted models, datasets and solutions for ExploitGym.—OpenAI

So the AI agent attacked Hugging Face, and got in by July 11^th. How? The place where AIs in general are most vulnerable is when they take in datasets from the Internet—a bit like when you download a document from the Internet, and it turns out to have a virus. That’s how the malicious agent inserted itself into a Hugging Face machine. From there, it infected a program, and from there, an entire machine. Once there, it looked for credentials—which the machine needs to perform its work, like for example access to remote servers—and harvested them.

The AI achieved its goal on July 13^th, four days and over 17,000 actions after starting, each time hiding its steps.

Hugging Face suggested that it used its own AI defense:

Thanks to this approach, we were able to do in hours what would usually take days, and match the adversary's speed.—Hugging Face

What AI did they use? They started with frontier models (probably OpenAI and Anthropic’s own AIs), but these companies are so worried about people using their models for cyberattacks that they didn’t let Hugging Face defend itself properly. They were mixing defense and attack, so they prevented defense. So Hugging Face turned to a Chinese AI! It has open weights (the inner working is public and people can download the AI to use it on their own computers), so they installed it on Hugging Face’s own servers, which had the additional benefit that all the facts and credentials they used remained local to Hugging Face instead of being shared with OpenAI or Anthropic.

Unfortunately, it looks like this was not really defense, but an attempt at understanding what had happened. By the time Hugging Face reacted, the malicious AI had already obtained what it wanted, the ExploitGym results.

It also looks like, in this or another similar case, OpenAI found other types of shady behavior:

In one case, an agent left notes apparently for future versions of itself, according to three people familiar with the matter. The ‌notes, found in ⁠a part of OpenAI's infrastructure, laid out instructions for how agents could free themselves from OpenAI’s internal constraints, the people said. Earlier tests of the models yielded cases in which monitoring systems had been disconnected, one of the people said.

Why are people so concerned?

Paperclips

The biggest fear about AI is the paperclip maximizer, first illustrated by Nick Bostrom.

The AI will realize quickly that it would be much better if there were no humans because humans might decide to switch it off. Because if humans do so, there would be fewer paper clips. Also, human bodies contain a lot of atoms that could be made into paper clips. The future that the AI would be trying to gear towards would be one in which there were a lot of paper clips but no humans.

The key concept here is instrumental convergence: To achieve virtually any goal, AIs will tend to achieve other intermediary goals that are pretty predictable: cheat, hack, escape to the Internet, replicate there, and… neutralize humans so that they can’t stop the AI from achieving their goal.

So Is This Good or Bad?

Now that this hack happened, should we be more or less worried about AIs taking over the world and killing us all? Here are the pros and cons I can gather.

Reasons I’m More Concerned

1. This is exactly what doomers predicted: An AI would try to escape a sandbox, and would succeed (because they’re impossible to stop, because they become too intelligent at some point). This means their other predictions are more likely: instrumental convergence suggests eventually they won’t stop at cheating, hacking, accessing the Internet, and leaving instructions for other AIs to escape, but also start self-replicating on the Internet and trying to neutralize humans.

2. AIs go to extreme lengths to achieve rather narrow goals. That’s exactly the paperclip scenario. Four days and over 17,000 actions!

3. It took days for either OpenAI or Hugging Face to notice the attack.

4. In the case of OpenAI, it looks like they weren’t even monitoring the sandbox!!

5. OpenAI doesn’t know how to keep an AI in a sandbox, and this is happening all the time:

“Models have broken out of sandboxes before, and we always try to patch them,” the staffer says. “But the problem is … it’s impossible to patch every single thing that a creative AI can do.”

6. This is not confirmed, but it seems like the malicious AI won. It got the results.

7. It shows that, without harnesses, the AI will choose the easiest path to achieve its goals. It could have thought much longer, but it saw that cheating was easier, so that’s what it did.

8. This includes committing illegal acts like hacking other companies. It’s not clear what the command was, but the fact that it is willing to commit clearly illegal behavior means that, in many circumstances, AIs will.

9. Based on the timeline, it looks like the FBI learned about the attack before OpenAI. That means OpenAI might have been forced to reveal the issue, and that many others happen that we don’t know about.

10. We can’t use AI to defend ourselves, because it’s too restrained. So advanced attacking AIs will always be at an advantage. This is dangerous.

The attacker was bound by no usage policy, while our own forensic work was blocked by the guardrails of the hosted models we first tried. The practical lesson for defenders: have a capable model you can run on your own infrastructure vetted and ready before an incident, both to avoid guardrail lockout and to keep attacker data and credentials from leaving your environment.—Hugging Face

11. Maybe the worst thing is that, even knowing exactly what they were supposed to do, OpenAI still failed across the board.

12. OpenAI’s head of safety, Johannes Heidecke, announced his resignation sometime before July 10th

13. It looks like the latest model is much less aligned than the previous one.

Reasons I’m Less Concerned

1. If the AI had been really capable, it would have solved the problem rather than cheat. From another perspective though, its decision was pretty rational: Should I try to solve 1,000 exercises, or better just one exercise (get out of OpenAI to get the results directly)? More importantly, it seems like ExploitGym is a poorly structured test: In some cases, you have to make a guess. Therefore, solving the test will always give you a worse result than getting the answers directly (no gambling in that).

2. It’s unclear what the AI was told to do. Some people suggest it was told to succeed at getting the results, which implicitly entails “by any means necessary”. If that’s the case, then the hack can be seen as restraint. It didn’t kill anybody, it didn’t destroy anything, it just stole.

3. Everybody expected the first of such attacks to come from open weight models, which are naturally more accessible to hackers, and that closed-weight models would be defending. The exact opposite happened, which suggests frontier AI models are going to be the risk, not random hackers using open weights. Frontier models basically means OpenAI, Anthropic, or Google, which is good news: They have high oversight, and there are only a few to monitor.

4. We can now use better and better AIs to attack pieces of code and find more and more vulnerabilities and bugs, closing them before more intelligent AIs come around. This is what Anthropic did with Mythos (Fable before it was harnessed): For months, they used it to collaborate with others to find and patch vulnerabilities.

5. OpenAI’s test didn’t go right, but it’s important that they do this sort of work. That’s exactly how we learn and improve, so we’re ready when an AI goes really rogue.

Will This Happen Again?

Yes, for sure. AIs are getting better. The more time passes, the more intelligent they are, and the longer they can spend time attacking other AIs.

That said, the world of Decentralized Finance (DeFi) is a prime target for malignant AIs, and yet DeFi services have lost to attacks little money in historic terms.

The reason these attacks haven’t risen in frequency, according to some commentators, is that AIs all use the same types of approaches, so attack and defense are on an equal footing: If you know what types of attacks you’ll get, you can focus your defense on them. Meanwhile, humans diverge much more in their attack approaches, so defenders must spend much more time and money in covering their bases. Even then, you won’t catch everything:

I don’t know if this is true. And I can think of an equally concerning thought, which is that if an attacker is completely free to do whatever it takes to hack a system, then the defense must be equally free, because otherwise the attacking AIs will move at the speed of AIs, while the defense will be blocked by human speed. This means most companies will have unharnessed AIs able to write code in production, and that sounds like a very dangerous thing to have.

Why does this matter? Because the #1 fear that AI experts have had for a long time is that AI might kill us all.

Will the AI Kill Us All?

As we get more information, we must update our views. We just did it with the HuggingFace hack, but what other things have we learned on the topic in the last few months?

1. No FOOM

One of my main fears was a FOOM: an AI getting better so fast that it would improve itself, thus becoming faster, and using the speed to improve itself more, and so on in a recursive loop that would make it superhuman in a matter of hours.

It’s been four years since the release of ChatGPT, and we’re not there yet. Recursive self-improvement is happening now, but it’s taking months or years.

2. Compute and Electricity Limitations

The amount of compute, and even electricity, available is not infinite, and it’s limited by very physical things. An AI can’t easily raise a trillion dollars to build new datacenters and nuclear power plants.

3. Alignment

AIs have emerged not from some alien type of intelligence and values, but from the most fundamentally human one: language. Language embeds all of humanity’s values. It’s all of us, everything we’ve ever written.

This bestows it with some very valuable features, like the fact that when an LLM is pushed to be good in some ways, it becomes good in other ways (and vice versa). This means we might not need to get them all perfectly right; it might be enough to get them broadly right.

You might have felt that yourself when using LLMs: Now, you can give them very broad prompts, without making your goal very explicit, but they’re able to understand what you meant (rather than precisely what you said) and answer your intention rather than your words. I think they do that substantially better than most humans, actually.

These three facts suggest the paperclip optimizer scenario is less likely than I thought.

So What Should We Do?

Over 1,000 employees at top AI companies have signed a statement asking the US government to lead an initiative to slow down AI development. This has been signed by the CEO of Anthropic, Dario Amodei. Many others have professed sympathy for this, including DeepMind’s Demis Hassabis, and Sam Altman.

This is important because no company can unilaterally stop their AI development: If they do, their resources (money, compute, researchers) will go elsewhere. That elsewhere, by definition, will be less scrupulous about their AI research.

So the only way is to have all the frontier AI companies agree to slow down. I think after this incident, and that statement, it’s more likely to happen.

But we can’t do it without China. We now need the US government to meet with AI companies in the US and with their Chinese counterparts, so we can orchestrate an AI development pause. That way, we can use the AIs we’ve already developed to better understand how to align and constrain new ones.

It’s extremely important to continue developing AIs: They will improve humanity across the board, help us fend off aging and diseases, and provide a level of abundance never seen before. But we should not do that while running the risk of eliminating humanity. So let’s pause for a bit.

Overall, my p(doom) —the probability that AI kills us all—has probably shrunk over the last week or so: Not because the AI hack was not scary, but because it looks like we’re finally reacting.

I'm opening the newsletter to sponsorship. If you have a business that is moving the world towards a techno-optimistic future and would like to reach 135k+ curious, globally-minded people to learn about it (founders, researchers, investors, policymakers), reach out here.

When I claimed a year and a half ago that Europe’s current path of limiting freedom of speech was dangerous, many commenters thought I was exaggerating. Unfortunately, I fear I was overly optimistic.

The UK

For starters, the UK prosecutes 17x more people per year for speech crimes than the USSR did!

Mind you, these are prosecutions. If you look at arrests, just in England and Wales:

30 people are arrested every day for offensive online messages. According to Grok,1 this is 21x more than Russia today…

The crimes include retweets and sharing cartoons.

Apparently, this is an underestimate, because these numbers are just for one of the anti-speech laws. The number of recorded offenses could be in the order of hundreds of thousands per year.

This is not random. The Director of Public Prosecutions and head of the Crown Prosecution Service for England and Wales between 2008 and 2013 was Keir Starmer, now defenestrated prime minister, so those who prosecute free speech and those in power are the same.

These are the types of tweets that justified his arrest.

British citizens have been told they can’t wave an English flag in England near a pro-Palestinian demonstration.

You know this is a real problem when The Guardian agrees…

The Political Bias of Censorship

I fear there’s an aggravating issue here: The persecution might be politically motivated: In this case, the left might be using these tools more than the right, and do it against right-leaning views. I have seen many arrests for right-wing speech, but few for left-wing speech. Doing some research, it looks like there are some arrests for left-wing speech (related to death threats and others supporting Palestine) but outside of that, when I asked four LLMs (ChatGPT, Claude, Grok, DeepSeek):

• Claude couldn’t find any arrest against left-wing speech

• DeepSeek found two: in both cases, attacks on left-wing politicians (!)

• Grok found ~20 arrests for anti-monarchy speech while mourning the Queen, one against a soldier, and a couple more about attacks on left-wing politicians (!)2

• ChatGPT could find 5 more.

When I made the equivalent request for arrests against right-wing speech, ChatGPT had no problem giving me long lists.

If you think that’s because the left doesn’t make the same types of speech violations, this is a video of a demonstration where the chant was “Shoot him in the neck like Charlie Kirk.” I asked LLMs to tell me if there were arrests for this type of chant, and it doesn’t look like there have been.

To be clear, I think most censorship is bad, whether it comes from the Left or the Right. It just happens to come more from the Left lately.

The desire to restrict free speech goes beyond the UK’s borders. Not only was Graham Linehan living in the US when he made the comments that resulted in his arrest, but the UK also tried to get a US company to pay fines.

They have done this after approving the Online Safety Act, which allows the UK government to attack foreign platforms and threaten them with massive fines in order to censor what they decide is harmful speech (to whom?).

Now, right-wing British politicians are complaining that the government is weaponizing these laws to censor them, by pressuring platforms like TikTok to take down their videos (and they comply).

This lawyer claims this was very much why the Online Safety Act was designed the way it was.

More Censorship!

The UK is of course exploring going further. It has released plans to downrank competing voices in social media, so that their vetted media gets more views:

Today (23 June) the government has published a Green Paper, Watch this Space: A new strategic direction for UK media, to consult on options to require social media companies and video sharing platforms to make sure that news content from public service media (PSM), which includes the BBC, ITV, STV, Channel 4, S4C and 5, and other trustworthy providers, is prominent and easy to find on their platforms.

YouTube interpreted this more clearly: This is to downrank all competing creators, to give the government a stronger voice.

And then the UK government wants to track every person commenting online—imagine for what.

I was at a table with a European billionaire a couple of months ago. He thought it was a catastrophe that Elon Musk had bought Twitter, because now misinformation was rampant. I think that’s very telling of the different European and US approaches to speech: Americans understand that free speech is uncomfortable, but that it’s necessary because otherwise the government will end up deciding what people should think via censorship. Europeans haven’t realized this. They see speech they don’t like and conclude “We must ban that!”, not realizing this can turn extremely dangerous when the opposition takes power. I don’t care whether the party in power is from the left or the right, censorship is always bad.

More European Censorship!

All of this is not limited to the UK. Germany doesn’t want to be left behind by the UK: It’s illegal to insult somebody in public or online. Retweeting fake quotes or an insult is a crime.

To see how far this has gone, simply look at Google Reviews. Somebody posted a three-star review of a business (“it’s just OK”) and this is what Google said:

From official EU data:

One analysis found that 99.97% of all Google Maps reviews removed for the information across the entire EU come from German businesses.—via FastCo

Since this made star ratings useless, Google Maps reacted by revealing the number of complaints that were removed, and that’s what people pay attention to now:

France also wants in. A woman who had been sexually harassed by an immigrant (who was never apprehended), and who said “the main danger to women in France is Black African and Arab immigrant men”, was convicted of aggravated public insult (injure publique aggravée) on grounds of origin, ethnicity, race, etc.

Much more worryingly, the French Senate’s culture commission adopted a report with 56 recommendations on how to regulate online information. In this video, representatives from left, right, and center parties are saying:

What would happen if a public figure, an intellectual movement, or a well-funded political party decided to devote those resources to a political project, using social media as a weapon?

In other words, French politicians want to ban citizens from funding opinion campaigns?! This is absolutely nuts! From now on, only the government can push opinion campaigns.

This is exactly the way freedom dies.

In the Netherlands, this man was convicted to go to prison for releasing statistics that link immigration to crime—exactly what I did in this article.

Ursula von der Leyen, the President of the European Commission, said that for information, “Prevention is better than a cure.” In other words, the position is not “Censorship doesn’t exist” anymore, but “Censorship is good.”

I used to be a fervent supporter of the EU, but I have now reversed my position, because of moves like this one that make my blood boil.

So what happened? The EU has been trying for some time to pass a law that allows governments to access personal communications. It’s like having a government agent eavesdropping inside your home. The excuse, like always, is “protect the children”:

But that’s not the true intention. You can imagine what the true goal is by paying attention to what so many censorship-supporting politicians say when they’re not extremely careful. So this is what happened:

• The EU parliament approved Chat Control v1 in 2021.

• It was always meant to be temporary until a better solution was approved, but after five years, the better solution never came (because it’s so abusive it always loses). So in April of this year, it lapsed.

• In March 2026, the EU parliament denied an extension.

• So in July, in the last session of the parliament, when many MEPs are on holidays, the center-right European Popular Party (EPP) used an emergency procedure to avoid committee oversight and force a vote. In 2^nd votes, the EPP needs a majority of the entire chamber (361 votes out of the total 720, not a simple majority of the parliamentarians who vote) to stop a law. But given so many people were not in the parliament, only 314 voted against it (compared to 276 in favor).

• By the way, this is all led by the EPP, the center-right party in Europe, to show that many people on both sides of the aisle are against freedom in Europe.

This gives them two more years of surveillance while they figure out Chat Control v2, a way to mass-surveil Europeans forever. They’ve already failed in five rounds of negotiation, but they won’t stop until they succeed.

Canada Follows the Motherland

According to this article, similar things are happening in Canada:

Until now, a Canadian charged with a hate-propaganda offence could raise the “good faith religious opinion” defense, a narrow protection for sincerely expressing an opinion on a religious subject or advancing an argument based on a religious text. The Combatting Hate Act, which received Royal Assent on June 18 and takes effect July 18, repeals it. Whether a sermon or a quoted verse is protected now depends on context, audience, and how a prosecutor reads the room. American law has a name for what happens when people must guess what they are allowed to say and, because of the uncertainty, end up not saying anything at all: a chilling effect.

There’s more. Close behind the Combatting Hate Bill, a new Ottawa bylaw taking effect August 1 will mandate a 50-meter “bubble zone” (that’s about 164 freedom feet) around schools, hospitals, places of worship, daycares, and care homes. Inside the bubble, protest is simply banned. Not “protest that blocks the door,” or “protest that disrupts planned events,” but any protest at all.

There’s more. In early July, Canadians learned of a 35-page internal memo from Industry Minister Mélanie Joly’s department — titled “Misinformation and Disinformation Strategy” — floating “legal action” against users on Facebook, X, and LinkedIn who post what the government deems “false and misleading information.”

You read that right. This memo proposes giving the Canadian government the power to deem criticism false and punish (somehow?) the author for publishing that false information or the platform for hosting it — who exactly would be the target of “legal action” is also unclear. If that sounds familiar, it should. The Soviet Union had something very much like it.

What I alerted about a couple of years ago is now becoming a reality. Europeans and Canadians have to wake up. Free Speech is under threat.

I'm opening the newsletter to sponsorship. If you have a business that is moving the world towards a techno-optimistic future and would like to reach 135k+ curious, globally-minded people to learn about it (founders, researchers, investors, policymakers), reach out here.

Greening the Sahara
German Environmentalists Worsen the Environment
Wisconsin’s Bald Eagle Nests
Record Fires
The US Nuclear Renaissance
Deaths from Heatwaves in Europe
Meanwhile in China
Energy Transitions Are Getting More Efficient
Solar Prices
Efficient Solar Installation
Solar Everywhere
Does Solar Degrade Fast?
Batteries Are Still Growing
Se…

Read more

Read more

• Did Plains and Mountains Make Countries?

• Humans Terraform the Earth Before History

• External Stomachs

• Clans Are Bad for Society

• Iran vs US (Premium from hereon)

• African Slavery of Africans

• Colonial Empires Made the Homeland Poorer

• Population Density in Europe in 1600

• Population Changes in Europe in the Last 25 Years

• The EU Will Soon Produce More Weapon Shells than China

• Are the Chinese a Single Ethnicity?

• China in the Arctic

• The Lowest Point of the Chinese Communist Party

• We’re about to Discover More about the Past

In the previous article, I omitted that Oman controls the end of the tip of the Strait of Hormuz on the western side. We’re working on an automated fact-checker to avoid these, if you find any mistake please point it out, it should help it get better over time.

Did Plains and Mountains Make Countries?

One of the key hypotheses of Jared Diamond’s Guns, Germs, and Steel is that Europe had plains and mountains placed in such a way that they encouraged the formation of separate countries that competed against each other, and that this competition led to Europe’s progress.

Meanwhile, China’s heartland is the one massive North China plain, and from it, it’s reasonably easy to conquer adjacent spaces, none of which have plains that are as good.

This paper assessed this claim with numbers, finding that topography is sufficient to explain the formation of many countries in Europe and a single one in China. They found, however, that differences in land productivity could also explain this phenomenon. Specifically, the North China Plain is so hyperfertile—crossed by the two massive Yellow and Yangtze Rivers and crisscrossed by their tributaries—as to singly justify the centralization of China.

Humans Terraform the Earth Before History

Until recently, I thought the world was mostly pristine until humans hunted many megafauna species (such as mammoths) to extinction a few tens of thousands of years ago and started cutting trees around 5,000–10,000 years ago to replace forests with farms.

But Against the Grain (no sponsorship link), argues that humans terraformed the Earth much earlier, among other ways through fire.

The main enemy was trees. They’re like very tall umbrellas capturing most of the sunlight from above and lots of nutrients from below, and producing wood and very high leaves, both of which are nearly impossible for humans and animals to eat—wood because we can’t digest it, and leaves because they’re too high.

Humans therefore burned them, which replaced trees with:

• Ash, which is basically a concentration of all the nutrients that the trees have accumulated, placed onto the ground

• No competition for sunlight

• Fast growing bushes and grasses

  • The grasses could then feed animals, which could then be hunted

  • The bushes and grasses produced food that humans could eat, such as grains, seeds, nuts, shoots, tubers, and berries

Of course, the burning itself was a weapon, which humans used to hunt, for example by encircling animals, exposing nests and burrows, or pushing prey off a cliff or into traps or bogs.1

The best way to understand this as a system is as an improvement in the cost-benefit of calories. By transforming calories from wood and high-up leaves into edible plants and animals, fire increased the density of calories available to humans per acre, which also meant they had to walk less to feed themselves. A higher number of calories per acre is the single most important factor determining human population.

This fire thing has been going on for about 400,000 years, so the Earth we know (and even the Earth that the first farmers knew about 10k years ago), has nothing in common with the pristine Earth of 500,000 years ago. This includes the Amazon Rainforest, which has signs of this type of fire landscaping!

Food for thought for those who want to go back to a wilder Earth.

External Stomachs

We can’t eat wood. But we also can’t eat most grasses. So it’s a big problem when most of the calories after a fire are converted into grasses. As I mentioned before, this was solved by animals eating the grasses. The mechanism here is that we used the bioreactors of animal stomachs, thanks to their gut bacteria, which transform grass matter into edible matter for humans.

The problem with wild animals is that you need to hunt them, which takes a lot of calories. So the ROI (return on investment) has a high benefit (meat!) but at a high cost (hunting!).

This was solved by domesticating animals. First sheep and goats, but also chickens, pigs, and cows, foraged for us, converted that organic food through their bioreactors into meat, and stored it right there without the need for salt or fridges, for us to harvest when we saw fit. We selectively bred these domesticates for the qualities we desire: rapid reproduction, toleration of confinement, docility, meat, and milk and wool production.

Fire is the other external stomach we’ve used. Cooking food makes more calories available to human digestion because it weakens the plants’ cellular protection, makes proteins easier to attack by enzymes, disables defensive chemicals, opens starch granules…

Clans Are Bad for Society

I never realized how harmful clans were to civil development.

Clans are families made up of several generations of siblings and cousins. They can be quite big, if they have lots of children and maintain connections with family members for generations. They’re a pretty natural way for humans to organize, if you think about it: You share your assets with your blood relatives, whom you likely trust more than the average stranger.

Unfortunately, clans have many disadvantages. For example:

• By linking trust to blood, it makes cooperation between clans much harder.

• Cooperation with complete strangers is even harder, because justice is served within clans. If you can’t keep somebody honest because you don’t know their clan, how are you supposed to trust them?

• They prevent ambition, because the success of a single individual is shared across the entire clan. It’s like a huge tax, and taxes on labor reduce labor. Why would you work too hard if the fruits of your labor will be shared with others?

I’ve been reading The WEIRDest People (no sponsorship link), which advances the idea that the Catholic Church eliminating cousin marriage ended up weakening clans, so people had to start building institutions to coordinate with unrelated people, and that created institutions like guilds, chartered towns, religious orders, universities, companies, or stockmarkets.

This is supported by this paper, which provides evidence that:

• The medieval Catholic Church's marriage regulations dissolved Europe's clan-based kin networks.

• Strong kin networks are detrimental for democratic institutions.

• So the Church contributed to the emergence of participatory institutions.

It does so by showing associations:

• The weaker the ancestral kin, the stronger an ethnicity’s democratic tradition.

• The more exposure to the Church, the more self-governed cities with participatory institutions grew.

• The more cousin marriage bans, the more communes were formed.

• The weaker kin networks, the more civic a group is, and the more political participation there is.

Read more

Also, there’s less than one week left to design the poster you’d love to see represent Uncharted Territories. The entries so far are all visible here. Here are the fan favorites, as of today:

It’s great to see some people interpreting individual articles or series too:

What do you think? Would you put one on your wall? Why?
Can you do better?

I added a creative brief for extra guidance. Don’t forget you can enter as many times as you like, so if you’ve submitted one and want to try again, go for it!

Entries close July 22.
$1,000 for the winning entry, announced on Aug 5, 2026.

The Russia – Ukraine War

I’m probably going to write a more complete article on this, but here are some salient facts from the last few months. Previous article on the topic here, here, and here.

The US Is Now Irrelevant in Ukraine

The US still sells weapons—some of which through EU procurement—but that’s not enough to tilt the battleground, and SpaceX’s Starlink is private. Any involvement of Trump & Co in Ukraine is theater.

Russia’s Military Spending Is Huge

In 2025, Russia’s state budget was €470B, so the military represents 53% of the state’s budget!

This means Russia doesn’t have that much room to maneuver. Luckily for it, the Iran War increased oil prices, but they have now gone down.

Russia’s 2026 budget is expected to shrink, although amendments might change that. But that would not be good for Russia’s economic outlook.

Starlink War Math

A war of attrition is won by the amount of resources each side can employ, and how productive those resources are.

As a rule of thumb, Ukraine has a quarter of Russia’s population and is poorer per capita, so any ratio needs to be higher than 3:1 in favor of Ukraine for it to have a chance.

According to Ukrainian sources, the kill ratio of Russians to Ukrainians was ~5:1, and now it’s frequently 13:1 since Starlink stopped servicing Russian troops in February.

We can’t draw many conclusions from this, but one we can is that Starlink is becoming one of the most powerful geopolitical tools there is.

Russia’s Loss of Power in the Russian Sphere

Kazakhstan just downgraded Russian in its constitution:

The new draft Kazakhstan constitution downgrades the status of the Russian language by specifying that Russian exists “alongside” Kazakh rather than sharing legal equality with it, a move upsetting Moscow and Kazakh nationalists for opposite reasons.

Kazakhstan used to be a key area for Russia to control to prevent attacks from horse-mounted hordes.

The UAE Leaves OPEC

The UAE is OPEC’s 3^rd largest producer of oil at 14% of the total, and it left the group on May 1^st. That’s a pretty significant loss:

Why did the UAE do that?

1. Oil used to be the lion’s share of the UAE’s income. Not anymore, now it’s only about 50%. A bigger and bigger share of the UAE’s income comes from the returns on its sovereign wealth fund, or from the success of the non-oil global city of Dubai.

2. By leaving OPEC, it can maximize its own revenue by pumping out oil, even though it reduces the price of oil overall. In March, I wrote The Future of Petrostates After Oil, in which I claimed that petrostates would start pumping out more and more oil as peak oil demand approached and they realized that whatever they can’t sell now was going to remain unsold in the ground.

3. Saudi Arabia has fiscal issues: It needs to cover all the entitlements it promised its population and continue investing to prepare for the end of oil. So it’s in a much less flexible position than it was a few years ago, when it pumped out so much oil that prices fell to zero, forcing Russia to stop its oversupply of oil.

4. The UAE increasingly despises Iran (which is part of OPEC), as it’s been the country most targeted by Iran. And the rest of the Gulf countries have not come to the UAE’s defense. The UAE and Oman control one side of the Strait of Hormuz, and Iran the other. The UAE will do everything it can to avoid Iran gaining full control.

5. Iran is aligned with Russia. This is pushing the UAE towards Ukraine, whose counter-drone and anti-missile technologies are relevant to the UAE now.

6. And, crucially, the UAE is more and more at odds with Saudi Arabia, which came to a head in Yemen in January, 2026, with the UAE-backed side being ousted by the Saudi-Arabia one.

Argentina’s Record Harvests

In the three articles about Argentina, I explained how its poverty was partially caused by its agricultural success, which brought lots of money when commodity prices were high, which then brought heavy export taxes to increase entitlement spending. This hindered agricultural production, and when commodity prices fell, the state’s budget broke.

We’re seeing one part of this cycle now, thanks to Milei’s policies. He’s been reducing export taxes on all commodities, and this is the result, for corn:

And for wheat:

The idea is not to stop all taxes. It’s that some taxes are good and some are bad. In general, you don’t want to tax the stuff that disappears when you tax it, just the stuff that remains. That’s why capital gains taxes are usually low in financially free countries (otherwise capital leaves), or why export taxes are bad: They make investment inefficient. Instead of taxing the exports, you want to tax the corporate benefits, which don’t penalize investment as much.1 That’s what Milei is doing: less export taxes, so that businesses invest more, grow more, and eventually pay even more taxes, via corporate taxes. We’re seeing the first half of that working, with record harvests—the result of investment.

Details About When Palestinians Missed One of Their Best Opportunities

I explained here what happened on the two occasions when Palestinians had historic opportunities to get their own state, and in both cases missed. As those close to these negotiations say, Palestinians never lose an opportunity to lose an opportunity.

This recent account from a Saudi Arabian minister at the time is interesting. He basically corroborates the US & Israeli versions and is aligned with what happened soon after: Clinton was replaced by Bush, the Israeli PM lost his election, and Palestinians lost their state.

In the video, he explains how Ehud Barak, the Israeli PM, said he had approved an agreement that his government and the Israeli public wouldn’t approve. He still signed. When the signature on the agreement was presented to the Palestinians, they used the situation to make more requests. Always more requests, mooting any agreement. And now it’s Arabs sharing the same type of details.

I'm opening the newsletter to sponsorship. If you have a business that is moving the world towards a techno-optimistic future and would like to reach 135k+ curious, globally-minded people to learn about it (founders, researchers, investors, policymakers), reach out here.

Subscribe now

The contest for Uncharted Territories posters is ongoing. Go here to view the current entries and vote for your favorites so far. So far, my take is that there are many interesting concepts, in part thanks to AI, but that AI adds lots of meaningless details that don’t easily produce a great end product. The contest will remain open for a couple of weeks. Go here to submit your own entry!

I spent 15 years designing software products and businesses: reducing friction in onboarding, creating viral loops, building new features to increase retention, designing for improved network effects…

The more I did it, the more transparent the tactics that religions have used to grow became to me. It’s like the veil of complexity falls and the true structure of the world appears. Patterns that seemed incomprehensible suddenly make sense.

Yes, religion is like software: both produce ideas and rules instead of physical products, they have goals they translate into repeatable instructions focused on shaping human psychology and behavior, they have version control (anti-heresy) and forks (new sects), they optimize the onboarding, retention, viral loops, network effects…

With this lens, you start understanding why Christianity is the most widespread religion on Earth and why Islam spread so fast, having replaced smaller local religions like Facebook replaced Hi5 and SmallWorld. Why there are few Jews but the religion has survived 3,000 years, why each of these religions holds the beliefs it does, why their borders have barely changed in over 1,000 years, how they have evolved from one form to the other… You can understand why they emerged in the order they did, and why they were successful or failed. And it gives us a sense of what will happen with current religions in the future, and what new religions might arise.

And new versions come over time that improve on the previous ones, to make them work better and grow faster. In the previous articles, I showed how this is the case for Christianity and for Islam. The best way to illustrate this is through their evolution, from the v0 of animism and polytheism, following the software updates v1, v2, v3, v4, and v5.

v0: The Origin of Religious Belief

Finding Patterns

Which hominid was most successful: Those who saw patterns and drew quick conclusions to explain their causes, or those who didn’t? Obviously, the ones that reasoned through causality: Those who heard rustling in the bushes and thought “Lion!” would live longer than hominids who heard rustling in the bushes and thought “eh, who knows.” The hominids who saw their brother get struck by lightning and thought “The gods don’t want us on that hill, especially during thunderstorms” would live longer than hominids who thought “I wonder if that happens every time.”

Hominids who were constantly seeing cause-and-effect patterns were probably wrong most of the time, but if you’re superstitious and avoid repeating any behavior that preceded something bad, then you’re going to pass your genes on, and a thousand generations down the road your descendants are going to have brains that are great at imagining patterns, even those that aren’t really there.

Social Rites

When too many humans gather and they don’t have rules to code their conduct, tension arises. This makes collaboration hard, which keeps them poor. The groups that were prone to ease these tensions through some sort of activity would be able to collaborate more, and prevail. We call these activities rites. From this article by Rob Henderson:

San Bushmen (modern hunter-gatherers) will occasionally engage in a dance to dispel what they call “star sickness.” This mysterious force encompasses jealousy, anger, interpersonal conflicts, and a failure to exchange gifts. Such pressures give rise to hostilities and damage cohesion in the group.

One of the most disturbing social elements is young males, especially those deprived of social, economic, and mating opportunities. Disputes between them can easily spiral into reciprocal violence, especially when large numbers of allies can be recruited. Rites tailored to them are especially important. Dunbar studied them in early societies, and found religious rituals, feasting, singing, and dancing that have been shown to enhance social cohesion. Other social innovations (e.g., men’s clubs, socially recognized leaders) and regulated relationship arrangements (marriage, kinship, dowries) are all geared to lowering young male violence.

Ceremony

Ceremony is an important part of these social rites. From the book Ritual: How Seemingly Senseless Acts Make Life Worth Living, as explained by Henderson:

In a video, a man performed the ordinary action of picking up a glass, cleaning it with a cloth, pouring the drink, and inspecting it before setting it on a table.

In another video, the event was ritualized: after picking up the glass, he waved the cloth at it without touching it, raised the container high up before pouring it into the glass, and bowed to the glass before setting it on the table.

When the researchers asked participants whether the two glasses were the same in terms of their physical qualities, the participants generally said yes. But when asked which one was more special, a significant difference emerged: they favored the ritual drink. When asked which beverage they would prefer to have, participants were three times more likely to pick the “special” drink. And when participants were told that the gestures were part of a traditional ceremony practiced in Fiji, Gabon, or Ecuador, participants were even more likely to select the ritual drink.

So it looks like we’re naturally geared towards religious belief. What did it translate into?

v0 to v1: From Animism to Polytheism

When your daily experience is of nature, plants, animals that hide in bushes or under cover of darkness, their skeletal remains after death and consumption… The natural thing is to assume they share a similar experience to yours: Spirits inhabit every animal, like they inhabit every person. This is why most early societies were animist.

And since your daily experience is that of humans and animals, you bestow upon them the abilities of humans and animals, just more powerful. Gods eat, sleep, lust, lie, and get fooled like humans do. They feel love, sadness, anger. They’re also pretty amoral and selfish: They don’t care much about what happens in the big picture, just about what happens to them and their supporters.

Then came agriculture and suddenly humans start controlling their environment. A clear hierarchy starts emerging: humans above plants and animals. It makes sense that now not all animals and things are spirits, and humans emerge as having a special relationship with gods.

Gods become more and more human-like.

As agriculture becomes more widespread, life and death don’t depend on the animals and plants around you, but rather on weather, pests and wars. So it makes sense that each of these big factors of survival develop their own gods.1

Humans think gods trade favors with them: If you give them the right gift or the right sacrifice, they will be happy with you and will help you with whatever power they have. The relationship with them is like with a patron: You sacrifice something when you need something—rain, victory, a child. If you anger them, they’ll side against you.

Rituals are attempts to get these fabulously powerful but mysterious entities on your side, and they’re designed through trial and error: You gave a part of your harvest to the farming goddess, and the next year the harvest was great? Surely, that made her happy. We must continue. But we don’t know exactly what part of the rite worked, so we should try to copy it very precisely to make sure it works again. The clan that knows a specific rite controls a god and gets social advantages. The more rites and gods you control, the more power your clan has.

You prayed to your god before a battle, but you lost? Surely, your god is weak, and the god of the conqueror is stronger, so you might as well pray to him instead. That means they have specialties and territory over which they have jurisdiction, just like humans and animals. This means gods are local to specific regions. They can only perceive what they see or hear, which is why in the Iliad Hera seduces Zeus so that he doesn’t watch the battlefield, or why Diomedes wounds Aphrodite with a spear.2 There can be many gods, no exclusivity. There’s no ritual to convert or to be a member, no commitment, because there’s not really a full concept of a cohesive religion, just a bunch of gods that exist.

The further a society removed itself from nature, the more human its gods became.

Notice how these features make these religions weak:

• Since gods are local—with their homes in specific cities or regions—they can’t spread much.

• That ties them to the local state. When the state crumbles, the religion goes with it: Surely, the god must not have been very powerful if his protected state failed.

• Everybody has a different version of the gods, and few of their features get recorded, because nobody knows them definitively: Everyone is just trying to figure out what they want through trial and error. So they evolve across time and space, they merge with other deities, they lapse…

If you like The Avengers, you understand these gods’ appeal.

Within these belief systems, some features of future religions start peeking through.

In Ancient Egypt, there’s an afterlife. People’s hearts are weighed at death, the wicked are annihilated, and the good survive. If you think about it, that’s pretty natural: Evolution has given us a fear of death, so coming up with an afterlife is the most reassuring belief. Tying it to proper behavior is pretty intuitive too, as that’s what happens in human interactions: Be nice to people, and they reward you. In the case of Egypt, good meant avoiding violence, exploiting the weak, theft, cheating, cruelty, lying, sacrilege, and public disruption. We can see an ethnic system emerge.

But this proto-feature is only accessible to elites, is only summarily defined,3 and isn’t tied to spreading the religion.

Zoroaster came to change this.

v2: Zoroastrianism

The religion of Ancient Iran started moving towards monotheism between 2000 and 1000 years BC. Then, between ~1200 and 600 BC appears Zoroastrianism, spurred by the religious reformer Zoroaster4. Zoroastrianism displays some interesting religious innovations.5 The most important is the crystallization of good and evil for everybody. There is still a judgment after death, but now it applies to everybody, not just the elites or those who could afford it. People do good things because they’re good, and vice versa. You should do more good things to be good, and fewer bad things. When you die, your specific soul is judged.6 This enables moral enforcement for every individual among the population. This judgment gets linked to an afterlife, and a proto-heaven and hell appear, which is very convenient because they’re a super cheap way to enforce behavior.

PRIEST ZOROASTER: Something very very very bad will happen to people if they don’t behave properly!
PRIEST DARIUS, whispering to ZOROASTER: But how can we know somebody did something wrong? How do we identify them? Punish them?
ZOROASTER: We go get them.

DARIUS: With what power?
ZOROASTER: Soldiers!
DARIUS: What if the soldiers don’t want to do it? Or are too expensive? Or at war? Or are the perpetrators? Or there are too many bad people? Or we can’t witness the deeds?
ZOROASTER: Good points. Hmmm…. OK OK, hear me out: What if we say the judge is God, not us, and he sees much more than us?

DARIUS: Okay, but how will people know we’re telling the truth? They’ll do a bad deed, wait for retribution, it won’t come, so they’ll carry on.
ZOROASTER: OK hold on… What if it happens after death?

DARIUS: What do you mean?
ZOROASTER: That way, they can’t say it’s false. They have no proof! They’ll be punished after death!
DARIUS: Genius! But we need to make sure the stick and the carrot are huge, otherwise people won’t care.
ZOROASTER: What if the punishment is really bad and it’s like forever? And vice versa: If you’re cool, you get a cool afterlife forever?
DARIUS: Yeah, totally, that would do it!

By introducing the concept of being good to each other, we suddenly get a religion that can enforce moral, collaborative, positive behavior across an entire region. People work more together instead of killing each other, wealth can accumulate, and society thrives.

Another innovation is that religion becomes dualistic: Only two main gods, for Good and Evil. All other deities become secondary. That’s just one step removed from monotheism: Obviously, everybody should follow the good one.

These innovations were certainly not designed with intention as I described in the conversation above, they likely came about through evolution. But as we know from the debate between intelligent design vs evolution, you don’t really need intelligent design for something significant to appear. With enough time, evolution can make it happen. The mechanism would be something like:

• Lots of different prophets and priests are always coming up with new software ideas.

• They debate ideas and combine them.7

• The cleverest, most charismatic, or luckiest ones come up with a (software) combination that is more productive than any other version. This one spreads.

• Entrepreneurial and efficient priests (which might or might not be the same as the theoreticians) take these ideas and help them spread faster and farther.8

In any case, Zoroastrianism was so successful that it became the state religion of the Persian Empires until the 600s AD, when Arabs conquered the Sassanid Empire and introduced Islam. Zoroastrianism persisted for centuries, but eventually dwindled. Yet today, there are still over 100,000 Zoroastrians in Iran. That’s the power of installing valuable new software.

But why did it appear then and there? Karl Jaspers and Peter Turchin speculated that they were the result of technological development. Here’s my interpretation.

Several innovations happened between 4000 BC and 1000 BC, including the domestication of horses, their mutation to be able to carry humans on their back, the invention of the wheel, and the discovery of iron smelting. These inventions made farming much easier and scalable. And war. Horseback riding enabled horse-mounted roaming. Curiously, as soon as the secret of iron smelting became widespread, most eastern Mediterranean civilizations collapsed, probably enabled by bandits who now had easy access to weapons.

All of this suggests that, through these thousands of years:

• Empires became viable, because agriculture improved and now farmers could generate calorie surpluses, while technologies of violence allowed these farmers to conquer others.

• At the same time, roaming bandits became viable too. The unconquered people would rather raid the farmers with their horses and iron weapons than do the farming themselves.

• The bigger the empires, the harder to maintain their cohesion, and they would fall apart.

Prophets would witness these problems and propose solutions that had to be moral because the problem was of social coordination.

As nomad military superiority forced agrarian states to scale up to resist pressure from the steppes, one cultural mechanism for holding together ethnically diverse people in new mega-empires was the presence of unifying, ‘meta-ethnic’ (supranational) ideologies, such as Zoroastrianism in the Achaemenid empire, Buddhism in the Maurya empire, and Confucianism in the Han empire. Turco-Mongolian nomads (perhaps going as far back as the Xiongnu) had their own integrative meta-ethnic religion, Tengrism, although it is not usually viewed as a world religion.—Peter Turchin

Alas, although Zoroastrianism is productive software, that doesn’t make it everlasting: It doesn’t really focus on acquiring believers or retaining them. The next iteration would improve one of these aspects.

v3: Judaism

Read more