Claude Code 30: Hadn't the Satisfaction Always Been in the Discovering Not the Discoveries?
A Lament
This is more Claude Code fan fiction. It’s asking you to believe in a very near future in which AI Agents are able to do a lot of what humans can do, including autonomous research. Here’s two other ones I’ve already written to help see where I’m coming from.
Thanks again everyone for supporting this substack. It’s a labor of love. It’s fun to talk about causal inference, econometrics, and now Claude code aloud and even more fun to know sometimes this lands for others. If you aren’t supporting the substack, consider doing so! But enjoy this anyway!
I want you to imagine with me a very near future when AI agents have shown they have the comparative advantage in research. Research in both the active doing sense — to research something — and research in the having done it sense — as in “she produced research”. I want you to believe me that the AI agent can do research faster, more accurately, and in higher volume. This therefore means that machines produce research at lower cost and therefore machines have the comparative advantage in research. I want you to imagine that that situation is now.
It’s against that backdrop that I’m going to reflect on a particular subset of human researchers for whom the deep meaning they derived as a researcher was bound up, not in the number of papers on their vita, but rather in the feeling they felt when they did good research.
This was a group who had, after all, always chosen their own projects, they pursued them without any coercion, no real bosses to speak of, they found the data, they proved the theorems, they fought with each other, they wrote things down and called them “papers”, and they urged others to believe them and not someone else. They would spend a lifetime seeking to persuade others that their work, their time manifested into words, pictures and numbers, were right. They said those words, pictures and numbers were true enough to merit others’ belief, and they wrote those words, pictures and numbers in order that they would be believed.
Belief. They wanted others to believe them. That’s what this was. They were asking other people to believe them. They were asking strangers and acquaintances to change their minds. It was bold and dangerous.
And so they wrote “papers”, as they called them, filled with reasons that asked others to voluntarily change their minds. Not because they liked them. But because they could not avoid the implications of their words, numbers and pictures, and thus wanted to change their minds.
It would take, if they were lucky, a long time for this to happen, if it ever happened, and usually it didn’t
They would endure sometimes nothing but rejection and defeat after rejection and defeat in their effort to be believed. Sometimes they would spend years before they even started their words. They would spend that time, instead, staring at their computers, pushing buttons, often confused. They would read other people’s words on their screens when they had the time. They would ask for help, and sometimes three of them would work together to make sense of the numbers. They would share the job — one person would make the numbers, another the pictures, another the words. They would try eight different things, seven of which would ultimately be false starts. They would say they were often puzzled and frustrated, they would have short lived spurts of progress.
Sometimes they would even say they hated the words, they despised the numbers, they described the pictures as ugly.
And yet they would not stop. Not even when closing time came.
And when they asked how it felt, how it felt to write those words and be told they were wrong over and over, after all those years, they would smile and say how happy it all made them. How happy they felt writing the words, making the pictures, adding up the numbers.
You would not be wrong if you believed that they loved making the numbers more than the numbers themselves.
And so, they would write draft after draft, crumbling the old draft into a ball and tossing it into the waste bin, starting over. They would try different ways of saying the same thing. They would put different words first, then they would switch the order and put the same words last. They would write many words and then erase most of them. Sometimes it seemed like they were trying to make the best words for someone else to believe them, but sometimes it seemed like they just wanted the best words for themselves.
But one day, once satisfied with their words, having double checked all their numbers and pictures, they sat satisfied enough to stop writing. Finally, they would whisper, they show their words to this one person and ask them to believe them so much that they will put their paper in their book.
Why you ask?
Because, if the paper is in the book, then other people will believe the words, believe the numbers, agree with the pictures.
But why can’t they just believe the numbers, believe the words, agree with the pictures? Why do they need the book, you might ask?
Because. It’s not true enough until it’s in the book. You need three people plus the person whose book it is to agree that it’s true, and then once those four people agree it’s true, then all the people in the world will know it’s true.
But that doesn’t make sense, you say. How can four people make all these other people believe something? Why can’t all the other people just choose what to believe?
Because, they would say, you can’t just decide when something’s true. You need four people to tell you when it’s true. Then you can believe it because then it’s true.
But what about those four people, you say? What four people told those four people?
And they say, no — the first four people don’t need four people. The first four people actually are fine. Unless it’s a tie, in which case you just need the one person whose book it is to just decide and then those other people don’t really matter.
So you only need one person? The person whose book it is?
And they say, no you need four people.
And so that’s how it was. It somehow worked. If you could get four people to agree, then their words, pictures and numbers get copied into a book which would then let everyone else know they can believe that the words were the correct words, that the numbers were the correct numbers, and that the pictures were the correct pictures. And then incredibly enough — it would work. Millions of people would believe it all those things so long as it was in the book.
So important was the book that sometimes people wouldn’t even read the words, or the numbers or the pictures. They would just look at the book. The book and not the paper said the paper had the right words, the right numbers and the right pictures. And four people chose what went in the book. Unless they disagreed, and then one person chose.
Here’s the more typical scenario though.
The person whose book it was, who received the words, the numbers and the pictures would usually ask four strangers what they thought of the words, numbers and pictures. They would agree and the person whose book it was would wait and then six months later only three of them would write back. The person whose book it was would open their letters and read them one at a time.
The first one would say they thought it was okay but could probably be better and then wrote three pages of ways to make it better.
The second one thought it was not okay, and never could be better, but gave three pages of reasons anyway of how to make it better.
Often these two people would say to do the opposite things as the other. Which would seem to imply at least one of them was wrong.
But then the third one would call them an ugly name, saying they must be stupid because only stupid people would think these numbers were the correct numbers or that these words were the correct words or that these pictures were the correct pictures.
And so the person who owned the book would write back and say they were sorry, but they had no choice but to say that the words were the wrong words, that the numbers were the wrong numbers, and that the pictures, while pretty, were actually the wrong pictures.
But maybe they should try this other person’s book? Maybe they’ll think they are the right words even though strangely they had just said were the wrong words. But maybe try them anyway.
And so they would. They would try the other persons book too. But that often didn’t end well either.
And that goes on for years until one day, assuming they hadn’t tired out, they found someone that said that if they changed most of the words, and changed the numbers and the pictures, while they couldn’t promise anything, maybe they’d put it in their book.
So they did. They changed all the words they were told to change, they changed all the numbers they said were wrong and replaced them them with the right numbers. And they then made different pictures.
And that’s the story of how a person spent years trying to convince people of their words, numbers and pictures.
And they would do that repeatedly throughout their life, voluntarily, no one ever forcing them to do it. And it was just as bad as it sounded which made their willingness to do it all the more odd.
Why would they do this? Why toil at something so frustrating and seemingly futile? So many obstacles and for what? What is the payout?
There’s money, fame, job security, of course. No one is doubting that. But saying that those happen does not a payout make. It’s a payout if the person cares about it. It’s not a payout otherwise.
One of the most ubiquitous payouts that researchers care about, if you lean in close and listen, is the satisfaction they feel from the process of research — if satisfaction is even the right word.
Many researchers put up with all of it, including the failures, the rejection, the long delays, the lengthy production time, the randomness at the journals, because they deep down love the research process itself. They feel satisfaction when they work. Not because of what they make. They feel satisfaction from the work itself. The work is the payout. Not the papers, not even the discovery. The discovering is the satisfaction, not the discovery. The discovering is the payout.
They love coming up with the ideas. They love finding the data. They love cleaning the data. They love thinking about the code. They love when the code breaks because then they get to fix the code and they love fixing the code. They love strategizing over the way to present the numbers. They love the challenge of crafting the rhetoric of a manuscript. They love starting over. They love erasing their words. They love all these things and since things are the process through which you reach a conclusion, they love the process of reaching a conclusion. For many, it may be the only thing they will ever love.
And thats why they do it. They go through all that because all that is the payout.
They feel satisfaction from the work and because the work is what they do, they sometimes work too much — at least according to others who don’t like the work. When asked what their hobbies are, they’re tempted to just name more work. That’s how much they love “the work” — when they retire, they just find other forms of work. They are workaholics and write papers about that too.
Note how that is different from loving the discovery itself. Loving the process makes you very open to whatever the answer will be. So long as it’s done in a way that feels right, that you followed the steps correctly, then how you got there matters far more than where “there” even is. You learn to accept unacceptable truths when you love the process of finding the truth more than the truth itself. You learn to tolerate discomfort. You learn to accept failure and rejection because deep down you have already been paid. You love the feeling of working itself and therefore you accept all the failure and see it as a really good deal. You are not on a treadmill. You are gorging yourself every minute of every day on an all-you-can eat buffet.
What if historically society organized its scientific sector in such a way that leveraged people with those traits to be their workers? Those whose willingness to work on unbelievably hard problems, even harder than their own ability to solve it, for no other reason than that they loved the process were perfect for such impossible feats as turning a crank a trillion times, every day, for thousands of years in the offhand one of those turns might turn out a miracle.
The ones with these traits would be willing to turn such a crank even if nothing was produced because these strange people loved turning cranks for its own sake.
These were hard jobs with uncertain outcomes. The work was hard, and the outcomes risky. And so to make it work, society needed workers who would voluntarily do it, and the ones most suited for it weren’t the smart ones so much as it was the romantic ones. The ones who loved turning cranks for no deeper reason than they loved turning cranks. Blaise Pascal said the heart loves what it loves. And that was them.
When you think about what scientific discoveries occasionally has wrought on this planet, you realize in retrospect how vital it has been. Antibiotics. Fertilizers. Spinning Jenny. Rockets. While not all of what came out of the wheel of science is equal, none of what came out of the wheel of science could have come out were it not for millions of people diligently pushing ahead and the only real explanation as to why they’d do that is the words that come out of their mouths — because they love the work.
This isn’t the only way to design a scientific sector. It may not even be the best way. It is just the way we did it. Doing science this way might just be our collective effort to make science because we didn’t feel we could find another way to get it done efficiently and therefore that’s how we did it.
But just because that’s how we have been doing it does not mean that that is how we should always do it. What is does not mean that we should do it that way. The one does not imply the other.
But what good fortune nonetheless that we built such a machine — a machine that held together by a single clasp called love. How lucky there were workers with such a clasp. Workers so willing to toil and tolerate such extreme uncertainty about the fruits of their labor, not because they like uncertainty, and not because they are especially patient, but rather because they get exactly what they love and need every day — from exposure to and participation in the discovering itself. Here’s the secret. When you love the process for the sake of process, you always get paid. They always get that and it doesn’t require they be patient at all. If they can plug into the joy of the journey itself, then they’ll turn the crank all day, all night, for their whole life. If such people exist, then all that’s needed is for such a scientific machine to exist, and a bridge to be built that connected them to it.
A lot of the actual skills you need for research can be obtained. You can learn to take derivatives. You can learn the algebra and the proofs. You can learn the statistics, you can learn the models. You can learn to code, you can get special sworn access to the buildings and computers with all the really big numbers. You can learn those things or be given those things.
But what you can’t learn, what you cannot really ever be taught, is how to love the work for the work’s sake. Love is not intelligence. No one can teach you to feel the work is itself intrinsically meaningful. You cannot be made to love discovering more than the discovery. You either feel that way or you don’t. There is no in between.
Of course utility like that is not transferable to one’s significant others or kids. You cannot buy goods and services on product markets with the meaningfulness of work. Because we have also designed society as semi-capitalist in nature, we must translate work into things which allow us to obtain the fruits of other peoples’ work. We must pay for it with locally accepted currencies, spent in product markets, then shared within our family, and so therefore in addition to satisfaction from meaningful work you must also get a paycheck.
And so we will. We will get a paycheck. We will get jobs. We will become professors.
You will become a professor.
You will work for a school.
You will be employed by a university.
We will build universities so as to manage workers like yourself, workers who love the research for its own sake. We can do that and we will do that because when one of these workers does something truly critical, the rewards from it will pay, not only the costs of hiring her — it will pay for everyone else too. Everyone in the past, everyone who ever will be. We are all subsidized by six or seven discoveries.
But the only way those six or seven discoveries will ever happen is if we can make a machine that draws in millions of people with traits that make them joyfully exert time and effort at repeatedly turning cranks and wheels through trillions of lotteries, whose expected value was positive, but whose outcomes only pay out after a trillion turns. Whose outputs become a mere 6 or 7 discoveries.
Let us now pause for a second and ask ourselves what happens if humans are no longer in the unique position of being the only species capable of such things as six or seven discoveries.
Let’s say that another species comes along that can find the same six or seven discoveries faster and who also don’t require any immediate payout like satisfaction from the process. They can do the work, all the time, never get tired and never need the meaningfulness of the work. They can do 5 times, 10 times, as much work, without the fighting, without the delays, without the mistakes. And so 6 or 7 discoveries becomes 15 to 20.
All thanks to this one trait — this one trait had motivated millions of workers to spend lifetimes on mind numbingly tedious tasks that only they could love, which to them was not tedious at all, but rather was like staring at a kaleidoscope of colors. Who did all this impossible work so much that they might even have done it for free. This one trait inside them had functioned like a clasp connecting workers to processes that spun science on a wheel that occasionally produced discovered that saved billions of lives.
What happens, then, when that trait that found love in and meaning from the work is no longer necessary for the work itself to get done? What happens when you can spin the wheel without love?
Well, for one, in the longrun, anything that did depend on that attachment goes away. If that single trait has been clasping together two things, then those two things are no longer clasped at that same place.
It isn’t in other words that you don’t need people for the work that I’m talking about. Rather I’m saying that the trait of feeling intrinsic satisfaction from the work is no longer needed for the work to get done.
And so those workers no longer have a comparative advantage in this type of hard work because that trait is no longer helping humans to do the work. The clasp is no longer needed to spin the wheel. The wheel spins whether they show up or not.
If what I am saying is true, then people possessing the trait of loving the work for the works own sake will both no longer do the work and they will carry with them the trait of loving the work for the works sake without any work to do it on.
Work is no longer theirs to do. To love research for its own sake, as its own reward, will become a legacy appendage. For if we do not have the comparative advantage in research against the machines, what will our argument be for why we should be the ones to do it? Science had never been for the scientists — science had always been for everyone, for the species, for everyone. Its value was external, not internal. The satisfaction was the carrot needed to get the 6 or 7 discoveries which was then shared with everyone. The secrets were always shared. They belonged to us all. The science was always the point — the satisfaction was just the wages.
This will become for them a legacy appendage the way the palmaris longus muscle is. It is a tendon in the forearm that you can see pop up when you pinch your thumb and pinky together and flex your wrist. About 14% of people don’t even have it anymore, but once we all did. It helped our ancestors grip and climb, but in humans it contributes nothing to grip strength. Surgeons actually harvest it for tendon grafts precisely because removing it has no functional consequence.
To love scientific work for its own sake, to love discovering, will become a legacy appendage that aches in them all the time when humans no longer have a comparative advantage in research. They will feel an ache they cannot name and do not understand, and they will feel it all the time. They will never use it again. They will never feel the satisfaction again. The machines will do the work. They never sleep. They never tire. They are never satisfied. They do not need to be. They do five times as much as we did, they make no mistakes, there are no longer any books, there are no words. No pictures. Just zeroes and ones. Endless zeroes and ones, with six or seven new discoveries every fifty years.
When jobs leave an area, we hope people leave the area too and go where the new jobs appear. Those people may get stuck, but it is at least possible to leave. And if not them, then their kids.
But these human scientists with traits whose only payout was joy from a task they no longer are able to do have nowhere to go. They can never leave to find another place because there is no other place.
They have nowhere to go.
Loved it and it kept me wondering about comparative advantages we as human researchers might have over machines, there must be at least one!
Just so beautiful. Made my day. In a strange way, it reminded me of the Asimov short story "The Last Question"