A friend once laughed at me for making my kids cardboard swords when she could click a button and have half-a-dozen brightly colored foam swords mailed to her for about $6. I didn't answer the real reason: I want my kids to realize that they can interact with their world and make things out of things. There's a certain empowerment to realizing that you can make things, not just buy them. That said, our work is sub-par compared to both industrialized products and the handiwork of actual craftsmen. The thing about *good* crafting is that it requires tools. Tools require workshop space. We don't have that. It also requires time to fiddle around and try things, something that seems awfully scarce in our post-industrial, dual-income, private-school lives. Needless to say, I follow homesteaders on IG and would be thrilled to read a book about craeft.
It often boggles my mind how much attention people used to pay to everything in their environment. Recently I read that the bark of some trees isn't flammable and I realized that I've been making fires my whole life and never noticed whether bark burns differently. Similarly, you wonder how ancients discovered that, say, willow bark tea is good for a fever.
The answer is that they were in tune with a wholly different environment. Moreover, without artificial entertainment pinging their brain receptors, they had to amuse themselves by investigating their surroundings.
Lovely review, as always. There is definitely an important place for making things, for seeing the world as it is not and then making it so, in our lives. I am a little more sanguine about the future than you are, however, because I see the role of hobbies as just that, the realm where we find some skillset we like to work on and improve our craft. Of course not everyone picks up such hobbies, but I think those days came and went a long time ago. TV effectively meant people could go vegetable to pass time, and many seemed to do so instead of working on something more interesting. I presume before then it was sitting around and getting drunk filling that space.
It wouldn't surprise me if the proportion of crafty people is relatively stable over time. There always seem to be some who put in more energy and thought into doing things, and others who do only enough to get by with no real interest in more or better ways.
The average American reads and writes at the third grade level.
That is why LLM are so popular.
It is also why our popular films are based on children's comic books.
But based on my experience at places like the Institute for Advanced Study, the smartest and most well educated young adults are smarter and better educated then my generation was.
If the masses are happy as functional illiterates, that's fine with me.
I'm teaching a course right now on writing for IT workers and I'm in danger of getting stuck in a game where they try to hide how they're using AI to do their exersizes. Your metaphor of the forklift is helpful. I'll tell you how the next class goes.
In the mean time - have you read The Craftsman by Richard Sennet? He starts from a very different definition but ends up with a similar picture of the right relationship for us to have with technology.
To your point around the vast majority of people having no *cræft* anymore, and to the linked Scott Alexander earring story and it's tie in with the AI points you're making:
It's fairly obvious we'll ALL be in a post-intelligence future soon.
We’re all going to have Phd-smart, maximally conscientious personal assistants in our ears. And this is like "next year," to be clear, and they'll be on a steeply improving trajectory.
This is going to happen because there’s an immense market for it, and because it’s possible with the level of AI minds we have now. The reason we don’t have AI assistants already is largely risk mitigation and CYA dynamics, but as soon as somebody puts together a human-in-the-loop-enough program infrastructure together that’s good enough to derisk it, we’ll be off to the races.
Just imagine - never needing to make or answer any phone call again. Having the 90% of your email that’s slop or low value / urgency handled automatically. Somebody who can seamlessly make reservations, book travel for you, research options and present them intelligently, make recommendations on any front, make sure the food you didn’t even realize you wanted shows up hot at your door just as you’re realizing you’re hungry, advice, on any topic and to any depth. Useful interjections and context, apropos and as needed, in real time.
A post-intelligence future.
From a functional standpoint, this materially changes what you should look for in a mate, and what will enable success for your kids.
If intelligence is counterfeited, conscientiousness and discipline matters more for overall success, because the people who will do the best on complex multi-polar goals in the future like “I want an interesting career, and a great spouse, and to have good habits that leave me energetic and engaged with my day to day life” will be highly conscientious people who will strictly follow the advice that GPT-o7 gives to them.
They have been earringed, per Scott's story, in other words.
I actually wrote a post about Optimal Descendant Strategy™ in such a future:
Yours is the first credible argument I've seen for NOT putting the earring on your kids (assuming no catastrophe and a future where humans exist).
But are you really prepared to see them fall behind all their peers? To not participate in the bare handful of jobs left for the top tier of smart, creative, and conscientious humans? To not win that Red Queen's Race, and to be just an ordinary UBI-pleb like everyone else?
I'd take that tradeoff if I knew it would lead to more descendants - grands and great grands. But we don't know that. Maybe uterine replication is solved, and the remaining o7 elites are able to crank galactic settlers out by the gross, and send their DNA across the stars to found new worlds. Maybe UBI plebs are all sniped by Infinite Jest-style virtual heavens that are optimized to be maximally stimulating and satisfying across narrative arcs by attending to pupillary dilation, cheek flushing, heart rate, and parasympathetic arousal, so they'll be "coffin-slaves" hooked up to IV's and catheters in pods somewhere.
There's no way to predict that world. And you KNOW that the earring / "strictly following o7 advice" will objectively give them better lives and more happiness, and even more descendants if they want them.
Does knowing how to thatch or weld really offset that?
This is something I legitimately struggle with, as a nerd who loves woodworking and crafting and race cars and such. But I think the balance probably leans towards choosing to earring your kids, and even yourself.
Because if it's really beyond the smartest human's capabilities, won't it KNOW whether being physically competent and cræfty will pay off in net happiness and outcomes? If that were true, won't it do that anyways? And if it weren't true, as verified by what might as well be an omnimax god, you really think it's worth manually disagreeing and crippling your kids relative to their peers?
To expand a little: I think even things like "successful career" and "happy marriage" and "good habits" are basically an essay about the Civil War. Maybe, if you're lucky, they're a book about the Civil War by an actual historian, which is to say that they can also be a kind of work of art that is valuable in itself beyond what the making does to you.
I also disagree about the current and probable future capacity of LLMs, but philosophically what does it even mean to "know" what will make a person happy? What do we mean by "happy"?
> To expand a little: I think even things like "successful career" and "happy marriage" and "good habits" are basically an essay about the Civil War.
I totally agree, in the present environment where people have to "tend their own garden" and instill their own habits.
I think when in-your-ear AI assistants can instill habits via operant conditioning and other methods, it turns consequentialist - all that actually matters is that the habits exist and the good outcomes happen due to the habits.
"But installing it yourself is more resilient, because if / when the AI goes down or gets sick of humans and ascends to Digital Heaven or whatever, the self-installed people will still flourish!"
Agree, but this is just a minor epicycle - so instead of operant conditioning, they guide your "interior decision making" to deciding to do the habits or whatever. That's like 10% harder if they're good at debate and understanding your mind, which they would pretty much definitionally be.
> I also disagree about the current and probable future capacity of LLMs,
Sure, maybe "earrings" are never possible for *everyone.* But there's always a threshold, right?
I mean, today they're basically Phd smart - assume they're frozen there. That still means they're better at thinking and deciding than roughly 95% of people.
Maybe there's gaps in reasoning and understanding and evalution, and we should discount this down to ~66% of people.
So 66%+ of people would be better off doing the earring thing in the aggregate and solely taking and executing GPT-o7 advice. We're just arguing about price, not fundamentals.
> but philosophically what does it even mean to "know" what will make a person happy? What do we mean by "happy"?
Do you "know" what will make your kids happier in most choices / situations? Usually, right? Because you've sampled and thought about more of the world, you understand their personalities and proclivities, you have an outside view that they don't, and probanly other reasons.
All "knowing" this really means is "having an accurate perception and appraisal of their innate characteristics and drives and current resources" and "being able to predict how they'd respond to a given choice given those innate things."
Looking at how well most people run their lives, it's NOT hard to do better than a lot of people do on those two fronts, in the aggregate, unrolled over time.
I'm just positing this writ slightly larger. Sure, maybe it's not true for YOUR kids! They're smart and capable and conscientious, and a shining light reflecting all that is good about the world and human capability and flourishing!
But you know, MOST people - you see people. We're just arguing price again.
It doesn't take much to expand above "better than 2/3 of people" to 80% and 90% and 98%. Maybe there's a ceiling somewhere, but I honestly doubt it.
Humans at the end of the day just aren't that complex, and are relatively predictable, even by other human-level minds. Once we start involving literally superhuman minds, I don't think there's much controversy here.
1. I obviously have strong beliefs about what constitutes a life well-lived. The problem is that other people have equally strong beliefs that are unfortunately incorrect. This is not a problem that copious application of IQ points is well-suited to solving.
2. The point of developing good habits is not to secure good outcomes, it is to shape yourself for immortal glory. Law and culture (and Mom) are teachers, but the work has to be done yourself or it doesn't actually help.
On one hand, I really appreciate it. The Hacker making do with only 4K bytes, the Thatcher using whatever there is around.
On the other hand, what would it mean if they had abundance? Is the hacker using 4K bytes really a hacker when he has 4GB? The Thatcher, when he has Aluminum roofing?
The Solutions are only ingenious and worthy because of the constraints. When the constraints are gone, those solution to problems that don't exist anymore feel tacky and useless to me.
I can appreciate them, in a sort of mental gymnastic sort of way, but not for true Craftiness.
To be Crafty, you have to solve the constraints *you* have, not people in the past.
So maybe the crafty people of today are those hacking LLMs, or the ones navigating Bureaucracy or the one picking up furniture left at the side of the road and restoring them.
Whatever those crafty people are doing, they are dealing with their *own* constraints.
Note: Maybe that is why we see more Craftiness in 3rd world countries - more constraints!
One big tech company apparently sets their cloud strategy up to rent new machines every time one gets to 25–30% load because their code is so inefficient at dealing with a spike of requests coming in at once. Real Programmers don't seem to work there, like the ones who understood the problem and solved it back in the 1990s.
I come across a Victorian home decor manual a while ago which said that the ideal piece of furniture is a family heirloom. It's not really practical for *all* furniture, but I feel like it points to the idea that the goal of craft is just meaningfully different than the goal of doing things in the most practical, efficient way.
Making things custom and handmade is rarely the most efficient option, but it's only because it's custom and personal that an heirloom even counts as an heirloom! And it's nicer quality than it *needs* to be because it's supposed to last many more years than you even factor into a normal calculation.
I find that crafts need some kind of goal like that (though not necessarily so serious!) to not feel like a LARP, because if you're just trying to be efficient, then machines can probably do it "better."
"What happens to them?" I think you are a few steps in front of, or behind, the people currently running the government. (Depends on the kind of craftiness we are measuring.)
A friend once laughed at me for making my kids cardboard swords when she could click a button and have half-a-dozen brightly colored foam swords mailed to her for about $6. I didn't answer the real reason: I want my kids to realize that they can interact with their world and make things out of things. There's a certain empowerment to realizing that you can make things, not just buy them. That said, our work is sub-par compared to both industrialized products and the handiwork of actual craftsmen. The thing about *good* crafting is that it requires tools. Tools require workshop space. We don't have that. It also requires time to fiddle around and try things, something that seems awfully scarce in our post-industrial, dual-income, private-school lives. Needless to say, I follow homesteaders on IG and would be thrilled to read a book about craeft.
Consistently one of my favorite Substacks - and this was no exception! Great review
A book that would make a great companion with this one is Matthew Crawford's *The World Beyond Your Head*.
It occurs to me that the closest modern equivalent to "cræft" may be "kung fu": https://en.wikipedia.org/wiki/Kung_fu_(term)
That makes a lot of sense.
It often boggles my mind how much attention people used to pay to everything in their environment. Recently I read that the bark of some trees isn't flammable and I realized that I've been making fires my whole life and never noticed whether bark burns differently. Similarly, you wonder how ancients discovered that, say, willow bark tea is good for a fever.
The answer is that they were in tune with a wholly different environment. Moreover, without artificial entertainment pinging their brain receptors, they had to amuse themselves by investigating their surroundings.
Lovely review, as always. There is definitely an important place for making things, for seeing the world as it is not and then making it so, in our lives. I am a little more sanguine about the future than you are, however, because I see the role of hobbies as just that, the realm where we find some skillset we like to work on and improve our craft. Of course not everyone picks up such hobbies, but I think those days came and went a long time ago. TV effectively meant people could go vegetable to pass time, and many seemed to do so instead of working on something more interesting. I presume before then it was sitting around and getting drunk filling that space.
It wouldn't surprise me if the proportion of crafty people is relatively stable over time. There always seem to be some who put in more energy and thought into doing things, and others who do only enough to get by with no real interest in more or better ways.
The average American reads and writes at the third grade level.
That is why LLM are so popular.
It is also why our popular films are based on children's comic books.
But based on my experience at places like the Institute for Advanced Study, the smartest and most well educated young adults are smarter and better educated then my generation was.
If the masses are happy as functional illiterates, that's fine with me.
Of course, sadly they vote.
To put my comment in some context, Plato bewailed the recent introduction of writing to Athens
because it would ruin people's memories.
He wasn’t wrong! (Especially for philosophy.)
Perhaps, it was the beginning of knowledge based on superficial skimming--at least in Athens, because writing existed long before then.
He also was afraid that passive reading would replace the active interactive learning that discussion with a philosopher provides.
I'm teaching a course right now on writing for IT workers and I'm in danger of getting stuck in a game where they try to hide how they're using AI to do their exersizes. Your metaphor of the forklift is helpful. I'll tell you how the next class goes.
In the mean time - have you read The Craftsman by Richard Sennet? He starts from a very different definition but ends up with a similar picture of the right relationship for us to have with technology.
Thank you once again for this piercing essay.
I have not! I'll put it on the list.
To your point around the vast majority of people having no *cræft* anymore, and to the linked Scott Alexander earring story and it's tie in with the AI points you're making:
It's fairly obvious we'll ALL be in a post-intelligence future soon.
We’re all going to have Phd-smart, maximally conscientious personal assistants in our ears. And this is like "next year," to be clear, and they'll be on a steeply improving trajectory.
This is going to happen because there’s an immense market for it, and because it’s possible with the level of AI minds we have now. The reason we don’t have AI assistants already is largely risk mitigation and CYA dynamics, but as soon as somebody puts together a human-in-the-loop-enough program infrastructure together that’s good enough to derisk it, we’ll be off to the races.
Just imagine - never needing to make or answer any phone call again. Having the 90% of your email that’s slop or low value / urgency handled automatically. Somebody who can seamlessly make reservations, book travel for you, research options and present them intelligently, make recommendations on any front, make sure the food you didn’t even realize you wanted shows up hot at your door just as you’re realizing you’re hungry, advice, on any topic and to any depth. Useful interjections and context, apropos and as needed, in real time.
A post-intelligence future.
From a functional standpoint, this materially changes what you should look for in a mate, and what will enable success for your kids.
If intelligence is counterfeited, conscientiousness and discipline matters more for overall success, because the people who will do the best on complex multi-polar goals in the future like “I want an interesting career, and a great spouse, and to have good habits that leave me energetic and engaged with my day to day life” will be highly conscientious people who will strictly follow the advice that GPT-o7 gives to them.
They have been earringed, per Scott's story, in other words.
I actually wrote a post about Optimal Descendant Strategy™ in such a future:
https://performativebafflement.substack.com/p/optimal-descendant-strategy-in-a
Yours is the first credible argument I've seen for NOT putting the earring on your kids (assuming no catastrophe and a future where humans exist).
But are you really prepared to see them fall behind all their peers? To not participate in the bare handful of jobs left for the top tier of smart, creative, and conscientious humans? To not win that Red Queen's Race, and to be just an ordinary UBI-pleb like everyone else?
I'd take that tradeoff if I knew it would lead to more descendants - grands and great grands. But we don't know that. Maybe uterine replication is solved, and the remaining o7 elites are able to crank galactic settlers out by the gross, and send their DNA across the stars to found new worlds. Maybe UBI plebs are all sniped by Infinite Jest-style virtual heavens that are optimized to be maximally stimulating and satisfying across narrative arcs by attending to pupillary dilation, cheek flushing, heart rate, and parasympathetic arousal, so they'll be "coffin-slaves" hooked up to IV's and catheters in pods somewhere.
There's no way to predict that world. And you KNOW that the earring / "strictly following o7 advice" will objectively give them better lives and more happiness, and even more descendants if they want them.
Does knowing how to thatch or weld really offset that?
This is something I legitimately struggle with, as a nerd who loves woodworking and crafting and race cars and such. But I think the balance probably leans towards choosing to earring your kids, and even yourself.
Because if it's really beyond the smartest human's capabilities, won't it KNOW whether being physically competent and cræfty will pay off in net happiness and outcomes? If that were true, won't it do that anyways? And if it weren't true, as verified by what might as well be an omnimax god, you really think it's worth manually disagreeing and crippling your kids relative to their peers?
Yes.
Appreciate the answer. This to me feels like a great example of a "virtue ethics vs consequentalism" scissor scenario.
To expand a little: I think even things like "successful career" and "happy marriage" and "good habits" are basically an essay about the Civil War. Maybe, if you're lucky, they're a book about the Civil War by an actual historian, which is to say that they can also be a kind of work of art that is valuable in itself beyond what the making does to you.
I also disagree about the current and probable future capacity of LLMs, but philosophically what does it even mean to "know" what will make a person happy? What do we mean by "happy"?
> To expand a little: I think even things like "successful career" and "happy marriage" and "good habits" are basically an essay about the Civil War.
I totally agree, in the present environment where people have to "tend their own garden" and instill their own habits.
I think when in-your-ear AI assistants can instill habits via operant conditioning and other methods, it turns consequentialist - all that actually matters is that the habits exist and the good outcomes happen due to the habits.
"But installing it yourself is more resilient, because if / when the AI goes down or gets sick of humans and ascends to Digital Heaven or whatever, the self-installed people will still flourish!"
Agree, but this is just a minor epicycle - so instead of operant conditioning, they guide your "interior decision making" to deciding to do the habits or whatever. That's like 10% harder if they're good at debate and understanding your mind, which they would pretty much definitionally be.
> I also disagree about the current and probable future capacity of LLMs,
Sure, maybe "earrings" are never possible for *everyone.* But there's always a threshold, right?
I mean, today they're basically Phd smart - assume they're frozen there. That still means they're better at thinking and deciding than roughly 95% of people.
Maybe there's gaps in reasoning and understanding and evalution, and we should discount this down to ~66% of people.
So 66%+ of people would be better off doing the earring thing in the aggregate and solely taking and executing GPT-o7 advice. We're just arguing about price, not fundamentals.
> but philosophically what does it even mean to "know" what will make a person happy? What do we mean by "happy"?
Do you "know" what will make your kids happier in most choices / situations? Usually, right? Because you've sampled and thought about more of the world, you understand their personalities and proclivities, you have an outside view that they don't, and probanly other reasons.
All "knowing" this really means is "having an accurate perception and appraisal of their innate characteristics and drives and current resources" and "being able to predict how they'd respond to a given choice given those innate things."
Looking at how well most people run their lives, it's NOT hard to do better than a lot of people do on those two fronts, in the aggregate, unrolled over time.
I'm just positing this writ slightly larger. Sure, maybe it's not true for YOUR kids! They're smart and capable and conscientious, and a shining light reflecting all that is good about the world and human capability and flourishing!
But you know, MOST people - you see people. We're just arguing price again.
It doesn't take much to expand above "better than 2/3 of people" to 80% and 90% and 98%. Maybe there's a ceiling somewhere, but I honestly doubt it.
Humans at the end of the day just aren't that complex, and are relatively predictable, even by other human-level minds. Once we start involving literally superhuman minds, I don't think there's much controversy here.
1. I obviously have strong beliefs about what constitutes a life well-lived. The problem is that other people have equally strong beliefs that are unfortunately incorrect. This is not a problem that copious application of IQ points is well-suited to solving.
2. The point of developing good habits is not to secure good outcomes, it is to shape yourself for immortal glory. Law and culture (and Mom) are teachers, but the work has to be done yourself or it doesn't actually help.
On one hand, I really appreciate it. The Hacker making do with only 4K bytes, the Thatcher using whatever there is around.
On the other hand, what would it mean if they had abundance? Is the hacker using 4K bytes really a hacker when he has 4GB? The Thatcher, when he has Aluminum roofing?
The Solutions are only ingenious and worthy because of the constraints. When the constraints are gone, those solution to problems that don't exist anymore feel tacky and useless to me.
I can appreciate them, in a sort of mental gymnastic sort of way, but not for true Craftiness.
To be Crafty, you have to solve the constraints *you* have, not people in the past.
So maybe the crafty people of today are those hacking LLMs, or the ones navigating Bureaucracy or the one picking up furniture left at the side of the road and restoring them.
Whatever those crafty people are doing, they are dealing with their *own* constraints.
Note: Maybe that is why we see more Craftiness in 3rd world countries - more constraints!
On the topic of cræft disappearing in programming, technical readers might remember or enjoy this old Rachel Kroll post: http://rachelbythebay.com/w/2020/03/07/costly/
One big tech company apparently sets their cloud strategy up to rent new machines every time one gets to 25–30% load because their code is so inefficient at dealing with a spike of requests coming in at once. Real Programmers don't seem to work there, like the ones who understood the problem and solved it back in the 1990s.
I come across a Victorian home decor manual a while ago which said that the ideal piece of furniture is a family heirloom. It's not really practical for *all* furniture, but I feel like it points to the idea that the goal of craft is just meaningfully different than the goal of doing things in the most practical, efficient way.
Making things custom and handmade is rarely the most efficient option, but it's only because it's custom and personal that an heirloom even counts as an heirloom! And it's nicer quality than it *needs* to be because it's supposed to last many more years than you even factor into a normal calculation.
I find that crafts need some kind of goal like that (though not necessarily so serious!) to not feel like a LARP, because if you're just trying to be efficient, then machines can probably do it "better."
"What happens to them?" I think you are a few steps in front of, or behind, the people currently running the government. (Depends on the kind of craftiness we are measuring.)