I would push back on the W3Techs data, because a fair amount of the argument rests on it. That metric counts websites, domains, each weighted the same whether it holds one page or ten million. It does not measure pages published, traffic, time spent, or anything one could call attention.
The absence of Chinese from that top ten should be read as evidence against the metric rather than evidence about Chinese. China is close to a fifth of the world's population with near universal internet access, and most of its content sits inside WeChat, Douyin, Weibo, Bilibili and Zhihu (walled platforms that a domain-level crawler cannot see).
The same distortion applies to Hindi and Arabic, where consumption is overwhelmingly mobile, video and in-app.
As you may know, English proficiency in Thailand is rather low (except for Bangkok, the city you are writing from, which is more cosmopolitan). Most Thai use apps like Facebook, Line, TikTok, Pantip so it would be counted as English based on W3Techs methodology, whereas the content is in Thai and should be counted as Thai.
The number that actually matters is the gap: English accounts for roughly half of websites but only about a quarter of internet users.
That divergence tells you English dominates publication (the open, crawlable, SEO-optimised, machine-translated, commercially-motivated layer of the web) while consumption remains far more fragmented and has been fragmenting further as the major platforms localise.
This matters for the thesis rather than just the footnote. Your argument is a network effect argument: English wins because everyone else is already there. But a network effect operates on where people actually spend their attention, not on where pages are hosted. If the world's largest online populations are conducting their lives in their own languages inside their own platforms, the network is more segmented than the W3Techs chart suggests and the mechanism that is supposed to dissolve national linguistic communities is weaker than you have assumed.
Maybe, but that doesn't change my point. Today, most of the content available online is in English. Someone who limits themselves to their native language has access to a much less diverse range of content, even if you factor in Chinese, and even if you factor in conversations in apps.
And while we can't ask a Polish or Japanese person to speak French or Italian, we can ask them to speak English : this simple fact shows that English has won out.
I think we are actually closer than it may seem, but I would still push back on the framing.
English probably represents closer to 17–20% of global online attention, not 50%; W3Techs’ 49.5% is a website-count metric and is therefore a poor proxy for the internet as people actually use it.
Its share has also been declining, although much more slowly today than during the dramatic shift of the early 2000s.
That does not mean English is no longer dominant: it remains the only truly global cross-border language, while the other 80% is fragmented across hundreds of languages but with China still representing about 15-20%.
It is not a surprise that users generally prefer their own language when a sufficiently good local ecosystem exists.
e.g. China built WeChat, Douyin, Weibo, Bilibili, etc…, and Russia built VK, Yandex and RuTube etc…, those are large parallel ecosystems requiring very little English.
So English dominance (when you say ‘most of the internet’) should not be confused with majority usage.
Its real strength is increasingly supply side: science, technical documentation, finance, software, research and other high value information remain disproportionately produced in English.
That forces users toward English when equivalent material does not exist locally, regardless of preference or proficiency. But that is likely to change.
AI translation may further reduce visible English consumption (as your footnote suggests), and maybe (and paradoxically) still reinforcing English as the hidden source or pivot language behind translated content…?
So overall, I would describe the trend as : English is declining as a share of attention, but remains structurally dominant (about 20%) in cross-border communication and high value information.
As I see the destructive power of nationalism as one of the major problems that faces humanity, I would love to agree. But while your view makes sense, it does not take into account some rather important aspects.
First, wealth is mostly not digital but material (toilet paper, sewage system, water, food, energy, etc).
Second, wealth is not flow but stocks and stocks need to be built by some organization (or nature). Who will build the public goods on which free enterprise depends if not nation states?
Third, voting with their feet would require a right to migrate, which is not on the horizon in any conceivable way.
Fourth, and most importantly, let us divide the world into two groups, those who have enough assets that they can live a free and secure life protected from upheavals like wars and inflation and shortages etc, and those who dont and who are hit hard each time a natural disaster or a war or an economic recession makes their lives more difficult.
What are the relevant percentages in this world of the owning class and the not-owning class? 10% v 90%, 1% v 99%, 0.1% v 99.9%?
Surely the not-owning class will have to organize to protect themselves. You are writing from the pov of somebody who can protect himself. But who will protect all the members of the not-owning class?
My prediction is that nationalism is on the rise again because the not-owning class has nowhere else to go than to their nation state. And no amount of digital technology will be able to change that.
2) So how did states protect their resources before the creation of nation-states? I get the impression that you’re confusing “State” with “Nation-State.”
3) On the contrary, voting with your feet is recognized as a fundamental human right : Article 13 of the United Nations Universal Declaration of Human Rights states that "Everyone has the right to leave any country, including his own"
4) The argument I make in https://disruptive-horizons.com/p/digital-nomadism-disrupting-nation-states is that in most countries, the wealthiest 10 percent of the population - and thus those with the greatest mobility - pay between 60 and 70 percent of the taxes, and that it therefore only takes a very small proportion of them to leave their country to completely destabilize public finances. I think this is partly what we’re seeing in many Western countries.
So, whether we like it or not, the disruption caused by nationalism will have an impact even on people who remain staunch patriots and stay in their own countries.
5) "My prediction is that nationalism is on the rise again because the not-owning class has nowhere else to go than to their nation state" : I agree, that is why I wrote "Expect relapses along the way. Nationalism will flare up here and there, carried by populists elected by desperate populations who cannot understand why everything keeps getting worse, and who fall for impossible promises built on a world that no longer exists. These resurgences will be real, loud, and temporary. A counter-reformation, not a reversal." :)
At some stage the not-owning class may go much further than looking to the nation state, but turn instead to revolution against the owning class. It is notable that many of the owning class have decamped to territories where their wealth and wellbeing is protected by highly authoritarian regimes.
I'm sure there was a degree of Schadenfreude among the moderately well off stuck in one country when Dubai turned out not to be such a safe place for the rich nomads, thanks to the blunderings of the US President in his dealings with Iram.
Great piece Olivier, but I think it comes from the perspective of a European digital nomad dreaming of utopia. The US and China are getting more nationalist. Countries in conflict like Russia, Ukraine, Iran and Israel too. Our best hope is that Iran continues to humiliate the US, and that has the same impact as the Vietnamese communists in constraining their impulse towards aggression.
Yes, but, as I wrote, "Expect relapses along the way. Nationalism will flare up here and there, carried by populists elected by desperate populations who cannot understand why everything keeps getting worse, and who fall for impossible promises built on a world that no longer exists. These resurgences will be real, loud, and temporary. A counter-reformation, not a reversal." :)
I would push back on the W3Techs data, because a fair amount of the argument rests on it. That metric counts websites, domains, each weighted the same whether it holds one page or ten million. It does not measure pages published, traffic, time spent, or anything one could call attention.
The absence of Chinese from that top ten should be read as evidence against the metric rather than evidence about Chinese. China is close to a fifth of the world's population with near universal internet access, and most of its content sits inside WeChat, Douyin, Weibo, Bilibili and Zhihu (walled platforms that a domain-level crawler cannot see).
The same distortion applies to Hindi and Arabic, where consumption is overwhelmingly mobile, video and in-app.
As you may know, English proficiency in Thailand is rather low (except for Bangkok, the city you are writing from, which is more cosmopolitan). Most Thai use apps like Facebook, Line, TikTok, Pantip so it would be counted as English based on W3Techs methodology, whereas the content is in Thai and should be counted as Thai.
The number that actually matters is the gap: English accounts for roughly half of websites but only about a quarter of internet users.
That divergence tells you English dominates publication (the open, crawlable, SEO-optimised, machine-translated, commercially-motivated layer of the web) while consumption remains far more fragmented and has been fragmenting further as the major platforms localise.
This matters for the thesis rather than just the footnote. Your argument is a network effect argument: English wins because everyone else is already there. But a network effect operates on where people actually spend their attention, not on where pages are hosted. If the world's largest online populations are conducting their lives in their own languages inside their own platforms, the network is more segmented than the W3Techs chart suggests and the mechanism that is supposed to dissolve national linguistic communities is weaker than you have assumed.
Maybe, but that doesn't change my point. Today, most of the content available online is in English. Someone who limits themselves to their native language has access to a much less diverse range of content, even if you factor in Chinese, and even if you factor in conversations in apps.
And while we can't ask a Polish or Japanese person to speak French or Italian, we can ask them to speak English : this simple fact shows that English has won out.
Thanks for your reply !
I think we are actually closer than it may seem, but I would still push back on the framing.
English probably represents closer to 17–20% of global online attention, not 50%; W3Techs’ 49.5% is a website-count metric and is therefore a poor proxy for the internet as people actually use it.
Its share has also been declining, although much more slowly today than during the dramatic shift of the early 2000s.
That does not mean English is no longer dominant: it remains the only truly global cross-border language, while the other 80% is fragmented across hundreds of languages but with China still representing about 15-20%.
It is not a surprise that users generally prefer their own language when a sufficiently good local ecosystem exists.
e.g. China built WeChat, Douyin, Weibo, Bilibili, etc…, and Russia built VK, Yandex and RuTube etc…, those are large parallel ecosystems requiring very little English.
So English dominance (when you say ‘most of the internet’) should not be confused with majority usage.
Its real strength is increasingly supply side: science, technical documentation, finance, software, research and other high value information remain disproportionately produced in English.
That forces users toward English when equivalent material does not exist locally, regardless of preference or proficiency. But that is likely to change.
AI translation may further reduce visible English consumption (as your footnote suggests), and maybe (and paradoxically) still reinforcing English as the hidden source or pivot language behind translated content…?
We could also use better measures:
- Page-weighted (Common Crawl, which runs language detection across billions of actual documents): English is 40.6%. It was 43-45% in 2018-20, so it's slightly declining. https://commoncrawl.github.io/cc-crawl-statistics/plots/languages
- Attention-weighted (internet users x hours online x language actually consumed): roughly 20%. https://www.obdilci.org/projects/main/
So overall, I would describe the trend as : English is declining as a share of attention, but remains structurally dominant (about 20%) in cross-border communication and high value information.
As I see the destructive power of nationalism as one of the major problems that faces humanity, I would love to agree. But while your view makes sense, it does not take into account some rather important aspects.
First, wealth is mostly not digital but material (toilet paper, sewage system, water, food, energy, etc).
Second, wealth is not flow but stocks and stocks need to be built by some organization (or nature). Who will build the public goods on which free enterprise depends if not nation states?
Third, voting with their feet would require a right to migrate, which is not on the horizon in any conceivable way.
Fourth, and most importantly, let us divide the world into two groups, those who have enough assets that they can live a free and secure life protected from upheavals like wars and inflation and shortages etc, and those who dont and who are hit hard each time a natural disaster or a war or an economic recession makes their lives more difficult.
What are the relevant percentages in this world of the owning class and the not-owning class? 10% v 90%, 1% v 99%, 0.1% v 99.9%?
Surely the not-owning class will have to organize to protect themselves. You are writing from the pov of somebody who can protect himself. But who will protect all the members of the not-owning class?
My prediction is that nationalism is on the rise again because the not-owning class has nowhere else to go than to their nation state. And no amount of digital technology will be able to change that.
To answer your points :
2) So how did states protect their resources before the creation of nation-states? I get the impression that you’re confusing “State” with “Nation-State.”
3) On the contrary, voting with your feet is recognized as a fundamental human right : Article 13 of the United Nations Universal Declaration of Human Rights states that "Everyone has the right to leave any country, including his own"
4) The argument I make in https://disruptive-horizons.com/p/digital-nomadism-disrupting-nation-states is that in most countries, the wealthiest 10 percent of the population - and thus those with the greatest mobility - pay between 60 and 70 percent of the taxes, and that it therefore only takes a very small proportion of them to leave their country to completely destabilize public finances. I think this is partly what we’re seeing in many Western countries.
So, whether we like it or not, the disruption caused by nationalism will have an impact even on people who remain staunch patriots and stay in their own countries.
5) "My prediction is that nationalism is on the rise again because the not-owning class has nowhere else to go than to their nation state" : I agree, that is why I wrote "Expect relapses along the way. Nationalism will flare up here and there, carried by populists elected by desperate populations who cannot understand why everything keeps getting worse, and who fall for impossible promises built on a world that no longer exists. These resurgences will be real, loud, and temporary. A counter-reformation, not a reversal." :)
At some stage the not-owning class may go much further than looking to the nation state, but turn instead to revolution against the owning class. It is notable that many of the owning class have decamped to territories where their wealth and wellbeing is protected by highly authoritarian regimes.
I'm sure there was a degree of Schadenfreude among the moderately well off stuck in one country when Dubai turned out not to be such a safe place for the rich nomads, thanks to the blunderings of the US President in his dealings with Iram.
Great piece Olivier, but I think it comes from the perspective of a European digital nomad dreaming of utopia. The US and China are getting more nationalist. Countries in conflict like Russia, Ukraine, Iran and Israel too. Our best hope is that Iran continues to humiliate the US, and that has the same impact as the Vietnamese communists in constraining their impulse towards aggression.
Yes, but, as I wrote, "Expect relapses along the way. Nationalism will flare up here and there, carried by populists elected by desperate populations who cannot understand why everything keeps getting worse, and who fall for impossible promises built on a world that no longer exists. These resurgences will be real, loud, and temporary. A counter-reformation, not a reversal." :)