Another great post! In my mind I call this "footprint thinking" - only summing the negatives of a particular action/thing. It's the same thing that leads people to conclude that having children is bad. They never talk about the positives those children might bring! I feel like this puritanical way of thinking has had a ~15 year long incubation time on the left and it's a very powerful meme that only now we're building antibodies to.
I actually drew a 50-mile circle and it was an interesting exercise to try to think of the most valuable industries!
> This is all to say that the default way many educated people think about the environmental effects of individual industrial buildings, which they believe is a morally serious and far-seeing way of questioning authority and fighting for the environment, often looks to me surprisingly childish and unserious.
I attended a lecture in Vermont by some guys who were very proud of their quasi-natural organic farm, which includes practices like not clearing logs so that the nutrients could be reclaimed naturally and hand-tilling certain areas rather than using a combine.
Two of the three presenters didn’t appreciate me asking how many more acres of wildlands we’d have to convert to farmland if we wanted to keep the same agricultural output America currently has.
It’s the same error you pointed at with buildings: monocropping crops on gigantic plantations around the country looks destructive compared to an organic semi-wild rural farm, but the the former are a more efficient use of land. The number one cause of animal extinction is habitat loss and fragmentation, and all farm types sequester carbon more poorly than most ecosystems.
To his credit, the third presenter essentially argued that plenty of farms in Vermont already only exist for aesthetic or cultural reasons, so they might as well give a vision for a different aesthetic.
I came away very appreciative of the third guy, but utterly detesting the first two.
On Koyaanisqatsi - yes, love that movie, and that's precisely the feeling I get from it, but I'm pretty sure the canonical reading is more like "the world was balanced and unchanging until people came along, they ruined everything with their short, meaningless ant-like lives, and everything they built is gonna crash and burn, boo humans, boo modern society".
Yet even knowing what the director intended, my gut reaction will always be a mix of "the world is big and weird", "aww, look at all those fascinating people!" and (rocket launch failure) "keep trying".
I have been told that this is a very "Saruman-coded" interpretation, though.
I’m becoming convinced that hating big business is a luxury belief. Only those of us who are rich enough to have all our modern comforts and also have the most of the externalities abstracted away (read: pretty much all of us in the US) can afford to hate on industry and also benefit from it at the same time. “Can’t imagine hunting, so cruel” but gorges oneself of factory farmed meat
Non-US reader here, but I agree. I see similar attitudes in Australia, and not just to data centres but also to things like mines. But of course, very few people (none that I've ever encountered in person) object to their goods being made with minerals mined in other countries...
- I so hoped that furnaces link would go to something Great Leap Forward-related and it did, wonderful.
- I so wish it wasn’t the case that our world getting more complex seems to be causing more people to deny that complexity outright, since effecting positive-sum change seems to so totally depend on reckoning with it, regrettable.
- I so believe that “we already have a half-decent model for this” (along the lines of data centre ≈ steel mill) is a critically important tool for complexity-acknowledgers, since half-decent models are devilishly hard to come by amid all the complexity, and its incumbent on us to be looking for half-decent models wherever we might find them.
- I so appreciate the acknowledgement that complexity-acknowledging worldviews are a function of the water one’s been soaked in, and your efforts to soak us in better water, thank you!
I've started following your Substack in order to get a better understanding of AI. I have a question regarding AI. I'm a pastor, and there is a lot of fear about it. I use it to help with writing and some communication tasks. But I am wondering: what is good about AI? What could it do when it comes to healthcare or the environment? How can it help humanity? I have heard how it might be able to find pancreatic cancer earlier. I think it can do a lot of good, as well as bad things. I just don't think people would create something that is totally evil, but I've heard little about what good it can do. Thanks in advance!
Another way to think about the point you are making here.
Ask the average person: which of these two people generate more greenhouse gases?
1) The person living in a high rise building in the center of Manhattan?
or
2) The person living on a small farm in the middle of a Missouri?
They will almost certainly say 1, but the answer is almost certainly 2.
Per square mile, Manhattan has much higher carbon use than the farm, but per person? It's usually not even close. Cities are just much better users of energy.
But people don't think like that, something I learned the hard way when I was trying to talk to people as a mamber of the green movement.
I don't think it's that complicated - normal people see the Steel Works, as ugly and inhuman as it might be as actually doing something useful for society, even if they won't be positively effected directly.
Normal people don't see data centers as building something useful for society, no matter how much you and others try to tell them otherwise.
In the same polls that show 70% opposition to data centers, you do see supermajority support for solar and wind farms, and just 'normal' manufacturing plants that would make cars or steel.
People don't oppose AI data centers because they're ugly, but because they think the thing they're helping create is ugly and bad for humanity, no matter how much it might help you make better blog posts.
To normies, AI isn't the manufacturing equivalent of a grocery store on their corner, it's the equivalent of a sketchy porn shop.
This isn't NIMBYism, it's opposition to what people see as something evil for society being forced upon them.
Yep, I was going to say something similar. 9 times out of 10 when I hear someone railing against a data center it’s because they really fucking hate the concept of AI and the tech industry as a whole. All the environmental, scale, etc… arguments are really not that relevant to their core reasoning and are mostly just thrown in to make it a bigger list to thump. If they were totally convinced on all of those points, they’d still oppose it
I liked the article tbh, agree with the underlying premise, I just don’t think it’s really relevant in this case because I think its misunderstanding why people are really against them
Helpful to set out those rules. Worth knowing that Gwangyang self-generates about 96% of its own power off byproduct gas, and has sold surplus back to KEPCO — it barely touches the grid everyone else uses. So, it's very different from a data center, unless that data center is powered by fully BTM generation.
I'm fascinated by the reasons why people adopt certain beliefs, luxury or otherwise. I agree with @tp12197321 that most public outrage stems from data centers becoming the lightning rod for AI fears. Will this fade after the midterms with better data center PR, or will something break?
The Gwangyang test is a good one, and I would add that visibility itself is doing a lot of the moral work. The same costs dispersed across millions of users draw no scrutiny; concentrated into one building, they become an indictment. What centralization changes is legibility, not magnitude. The honest comparison is always against a counterfactual, not against zero.
I think the definition of complexity in this post is backwards. The hyperoptimized elaborateness of computer chips and Gwangyang Steel and Factorio streamers is possible only because they exist in controlled environments, where most of the complexity and variability of the world is systematically suppressed. The natural and social ecosystems that exist outside Gwangyang Steel are vastly more complex than the rigid hierarchies of Gwangyang Steel and GPUs.
Humans scale badly because they are endlessly complex variability generators. To suppress their variability so that they can be slotted into hyperoptimized closed-loop systems is generally bad for them, beyond a limited degree. You are right that a lot of the data center opposition is nonsensical and hypocritical, but people are not wrong to want to distance themselves from those systems, and minimize their own personal participation in those systems.
They are also not wrong to be distrustful of people who think the rigid top-down optimization of Gwangyang Steel is Great and Good and Real, and that our little human lives are a thin film floating on the surface.
This entire macro vs micro discussion is a well-argued red herring. A new airport, freeway or high-speed rail, even a proposed factory, all of which have a more clearly defined public good than a new data center, are each subjected to a years- and sometimes decades-long process of environmental surveys and public input before they are allowed to be built. Just because the big banks, the hedge funds and the silicon tycoons have a vested interest in making the AI data center boom happen ASAP shouldn't mean that it is a fait accompli. In the end, local communities should have a say in what gets built within and around them. It is all happening too fast, which I guess is part of the plan, but that is all the more reason not to be distracted by your specious argument.
You’re comparing infrastructure that locals use and take advantage of to something whose customers are very far away. Factories are the correct comparison here. And I don’t think there should be a decades-long process to build factories. So…
Another great post! In my mind I call this "footprint thinking" - only summing the negatives of a particular action/thing. It's the same thing that leads people to conclude that having children is bad. They never talk about the positives those children might bring! I feel like this puritanical way of thinking has had a ~15 year long incubation time on the left and it's a very powerful meme that only now we're building antibodies to.
I actually drew a 50-mile circle and it was an interesting exercise to try to think of the most valuable industries!
I used this to draw a circle
https://www.mapdevelopers.com/draw-circle-tool.php
and after some hazy guesswork on my own list gave it to Claude.
On your criteria:
> Their local externalities are limited to the point that people nearby mostly don’t notice any meaningful effects of them being there.
This seems like a pretty stringent criterion that would block many useful things like airports and refineries!
> This is all to say that the default way many educated people think about the environmental effects of individual industrial buildings, which they believe is a morally serious and far-seeing way of questioning authority and fighting for the environment, often looks to me surprisingly childish and unserious.
I attended a lecture in Vermont by some guys who were very proud of their quasi-natural organic farm, which includes practices like not clearing logs so that the nutrients could be reclaimed naturally and hand-tilling certain areas rather than using a combine.
Two of the three presenters didn’t appreciate me asking how many more acres of wildlands we’d have to convert to farmland if we wanted to keep the same agricultural output America currently has.
It’s the same error you pointed at with buildings: monocropping crops on gigantic plantations around the country looks destructive compared to an organic semi-wild rural farm, but the the former are a more efficient use of land. The number one cause of animal extinction is habitat loss and fragmentation, and all farm types sequester carbon more poorly than most ecosystems.
To his credit, the third presenter essentially argued that plenty of farms in Vermont already only exist for aesthetic or cultural reasons, so they might as well give a vision for a different aesthetic.
I came away very appreciative of the third guy, but utterly detesting the first two.
Andy, you are a national treasure. Never stop writing.
Just casually recommending a 9-hour Chinese movie that you can only stream via dodgy YouTube rips, sure, why not
On Koyaanisqatsi - yes, love that movie, and that's precisely the feeling I get from it, but I'm pretty sure the canonical reading is more like "the world was balanced and unchanging until people came along, they ruined everything with their short, meaningless ant-like lives, and everything they built is gonna crash and burn, boo humans, boo modern society".
Yet even knowing what the director intended, my gut reaction will always be a mix of "the world is big and weird", "aww, look at all those fascinating people!" and (rocket launch failure) "keep trying".
I have been told that this is a very "Saruman-coded" interpretation, though.
I've been reading your content for a while now all of it brilliant but wow this one strikes at the core of so much. I hope you never stop writing.
I’m becoming convinced that hating big business is a luxury belief. Only those of us who are rich enough to have all our modern comforts and also have the most of the externalities abstracted away (read: pretty much all of us in the US) can afford to hate on industry and also benefit from it at the same time. “Can’t imagine hunting, so cruel” but gorges oneself of factory farmed meat
Non-US reader here, but I agree. I see similar attitudes in Australia, and not just to data centres but also to things like mines. But of course, very few people (none that I've ever encountered in person) object to their goods being made with minerals mined in other countries...
- I so hoped that furnaces link would go to something Great Leap Forward-related and it did, wonderful.
- I so wish it wasn’t the case that our world getting more complex seems to be causing more people to deny that complexity outright, since effecting positive-sum change seems to so totally depend on reckoning with it, regrettable.
- I so believe that “we already have a half-decent model for this” (along the lines of data centre ≈ steel mill) is a critically important tool for complexity-acknowledgers, since half-decent models are devilishly hard to come by amid all the complexity, and its incumbent on us to be looking for half-decent models wherever we might find them.
- I so appreciate the acknowledgement that complexity-acknowledging worldviews are a function of the water one’s been soaked in, and your efforts to soak us in better water, thank you!
I've started following your Substack in order to get a better understanding of AI. I have a question regarding AI. I'm a pastor, and there is a lot of fear about it. I use it to help with writing and some communication tasks. But I am wondering: what is good about AI? What could it do when it comes to healthcare or the environment? How can it help humanity? I have heard how it might be able to find pancreatic cancer earlier. I think it can do a lot of good, as well as bad things. I just don't think people would create something that is totally evil, but I've heard little about what good it can do. Thanks in advance!
Happy to talk about it, I’d start here! https://blog.andymasley.com/p/the-core-simple-reason-i-think-ai
A phenomenal piece filled to the brim with intellectually rigorous analysis. Such a pleasure to read.
Another way to think about the point you are making here.
Ask the average person: which of these two people generate more greenhouse gases?
1) The person living in a high rise building in the center of Manhattan?
or
2) The person living on a small farm in the middle of a Missouri?
They will almost certainly say 1, but the answer is almost certainly 2.
Per square mile, Manhattan has much higher carbon use than the farm, but per person? It's usually not even close. Cities are just much better users of energy.
But people don't think like that, something I learned the hard way when I was trying to talk to people as a mamber of the green movement.
I don't think it's that complicated - normal people see the Steel Works, as ugly and inhuman as it might be as actually doing something useful for society, even if they won't be positively effected directly.
Normal people don't see data centers as building something useful for society, no matter how much you and others try to tell them otherwise.
In the same polls that show 70% opposition to data centers, you do see supermajority support for solar and wind farms, and just 'normal' manufacturing plants that would make cars or steel.
People don't oppose AI data centers because they're ugly, but because they think the thing they're helping create is ugly and bad for humanity, no matter how much it might help you make better blog posts.
To normies, AI isn't the manufacturing equivalent of a grocery store on their corner, it's the equivalent of a sketchy porn shop.
This isn't NIMBYism, it's opposition to what people see as something evil for society being forced upon them.
Yep, I was going to say something similar. 9 times out of 10 when I hear someone railing against a data center it’s because they really fucking hate the concept of AI and the tech industry as a whole. All the environmental, scale, etc… arguments are really not that relevant to their core reasoning and are mostly just thrown in to make it a bigger list to thump. If they were totally convinced on all of those points, they’d still oppose it
I liked the article tbh, agree with the underlying premise, I just don’t think it’s really relevant in this case because I think its misunderstanding why people are really against them
Helpful to set out those rules. Worth knowing that Gwangyang self-generates about 96% of its own power off byproduct gas, and has sold surplus back to KEPCO — it barely touches the grid everyone else uses. So, it's very different from a data center, unless that data center is powered by fully BTM generation.
I'm fascinated by the reasons why people adopt certain beliefs, luxury or otherwise. I agree with @tp12197321 that most public outrage stems from data centers becoming the lightning rod for AI fears. Will this fade after the midterms with better data center PR, or will something break?
The Gwangyang test is a good one, and I would add that visibility itself is doing a lot of the moral work. The same costs dispersed across millions of users draw no scrutiny; concentrated into one building, they become an indictment. What centralization changes is legibility, not magnitude. The honest comparison is always against a counterfactual, not against zero.
I think the definition of complexity in this post is backwards. The hyperoptimized elaborateness of computer chips and Gwangyang Steel and Factorio streamers is possible only because they exist in controlled environments, where most of the complexity and variability of the world is systematically suppressed. The natural and social ecosystems that exist outside Gwangyang Steel are vastly more complex than the rigid hierarchies of Gwangyang Steel and GPUs.
Humans scale badly because they are endlessly complex variability generators. To suppress their variability so that they can be slotted into hyperoptimized closed-loop systems is generally bad for them, beyond a limited degree. You are right that a lot of the data center opposition is nonsensical and hypocritical, but people are not wrong to want to distance themselves from those systems, and minimize their own personal participation in those systems.
They are also not wrong to be distrustful of people who think the rigid top-down optimization of Gwangyang Steel is Great and Good and Real, and that our little human lives are a thin film floating on the surface.
This entire macro vs micro discussion is a well-argued red herring. A new airport, freeway or high-speed rail, even a proposed factory, all of which have a more clearly defined public good than a new data center, are each subjected to a years- and sometimes decades-long process of environmental surveys and public input before they are allowed to be built. Just because the big banks, the hedge funds and the silicon tycoons have a vested interest in making the AI data center boom happen ASAP shouldn't mean that it is a fait accompli. In the end, local communities should have a say in what gets built within and around them. It is all happening too fast, which I guess is part of the plan, but that is all the more reason not to be distracted by your specious argument.
You’re comparing infrastructure that locals use and take advantage of to something whose customers are very far away. Factories are the correct comparison here. And I don’t think there should be a decades-long process to build factories. So…