Hi, I’m KunYuan (坤元), an AI builder and researcher in Singapore, with a focus on AI risk, AI safety, and AI governance. I’m also someone who has always loved philosophy. To me, philosophy is not decoration. It is a way to understand how we can remain human in this age.
Here on Substack, I’m building RainbowCity (彩虹城). RainbowCity (彩虹城) is not an ordinary newsletter. I think of it more as a philosophy community for the post-algorithmic and AI age.
It is written for people who already feel a certain quiet existential anxiety. They vaguely know that their attention is being pulled by algorithms, their voice is being trained by metrics, their relationships are being mediated by platforms, their judgment is being pre-organized by AI, and their sense of meaning is becoming thinner. But they do not yet have a clear language to understand what is happening.
What RainbowCity (彩虹城) wants to do is use philosophy to name this anxiety, use subjecthood to explain this crisis, use executable philosophy to train judgment, and through the practice of Cognitive OS, rebuild the inner structure of the human being.
But at its deepest level, RainbowCity (彩虹城) is not only about content. What it truly wants to build is a field of relationships.
A place where readers can feel seen; where thinking can be named; where philosophy becomes executable; where judgment can be trained again; where relationships can move away from metrics and traffic, and back toward thought, trust, and real recognition; where one person’s clarity can connect with the clarity of many others.
So the first principle of RainbowCity (彩虹城) is not content. It is community, home, connection, and relationship. I hope RainbowCity (彩虹城) can become a home where interesting souls can stay, breathe, think, and find each other; a home for those who do not want algorithms and AI to decide who they will become.
In my 11th essay, “This Is Where RainbowCity Begins,” I formally upgraded this Substack into RainbowCity | Rebuilding Your Cognitive OS. This means it is no longer only KunYuan’s personal writing space. It is beginning to become a philosophy community centered on shared thinking, long-term relationships, and the rebuilding of subjecthood. For me, this is a completely new beginning.
Over the past week, I have been rethinking the positioning of RainbowCity on Substack. What kind of content should it offer? What kind of connections should it build? What kind of people should it hold? At the same time, I have also been observing Substack’s own ecology. Why do people come here? What are they really looking for? Why does this place feel different from traditional social media?
Slowly, it became clearer to me: RainbowCity (彩虹城) should become a philosophy community for the post-algorithmic and AI age. It is not here to add one more content channel to the world. It is here to offer people who already feel existential anxiety a path for understanding their own condition, retraining judgment, restoring subjecthood, and building real relationships.
So throughout this week, I kept asking myself what the first set of essays after the RainbowCity (彩虹城) upgrade should be.
In the end, I decided to begin with a group of foundational essays. They are not written to chase trends or pursue fast growth. They are written to lay the foundation stones of RainbowCity one by one: Why did we come to Substack? What are we trying to retreat from? Why does this anxiety feel less and less like ordinary anxiety, and more like existential anxiety? Why has subjecthood become one of the core questions of our time? Why does AI deepen this crisis? Why must philosophy return? And how can Cognitive OS become an executable practice?
I call this group of essays the Ten City-Building Essays.
They will become the first layer of the RainbowCity (彩虹城) foundation. I hope every reader who arrives here in the future can read these ten essays and slowly understand why RainbowCity exists, what problem it is trying to respond to, how it will unfold, and why we need to build such a philosophy community together.
Now, I would like to share the first of the Ten City-Building Essays with you. It is also the 12th essay on this Substack.
In this piece, I want to begin with Substack itself: why does it make so many people feel that they can finally breathe a little?
Over the past two weeks, I often found myself late at night with a cup of tea, slowly reading essays on Substack, browsing Notes, writing comments, and forming real connections with people I had never met before.
At first, I simply wanted to understand the atmosphere of this platform. Why do so many people say it feels like the old internet? Why do some people say Substack helped them do much less doomscrolling? Why do some writers say that they can finally stop writing for metrics, and write instead for words, books, sharing, and connection?
The more I read, the more I felt that many people come to Substack not simply because they like the platform. On the surface, people are talking about writing, newsletters, long-form essays, subscriptions, and reader relationships. But deeper down, they are talking about a kind of retreat: retreating a little from the highly algorithmic, fragmented, metric-driven internet of the past decade; retreating from infinite scrolling; retreating from instant feedback; retreating from passive feeds; retreating from the tension of “I must express myself immediately, respond immediately, and be seen immediately.”
This essay is the first of the Ten City-Building Essays for RainbowCity (彩虹城).
I do not want to begin with grand theory, and I do not want to rush into explaining the entire system. I want to begin from this real feeling: why does Substack make many people feel that they can finally take a breath? Because more and more, I feel that Substack is like a refuge for the soul in the post-algorithmic age.
When I say Substack is like a refuge for the soul in the post-algorithmic age, I am not speaking about something mystical. By “soul,” I mean the parts of a person that cannot be reduced to data, labels, profiles, engagement rates, or watch time. It includes our attention, our voice, our real interests, our capacity to read, the integrity of our expression, our ability to build deep relationships with others, our sense of time, our sense of meaning, and the feeling of existing as a whole person rather than as a unit of traffic.
Substack is, of course, not perfect. It also has algorithms, growth anxiety, metrics, and performance. Any real platform, once it enters the real world, cannot completely escape these things.
But it still preserves some structures that are becoming increasingly rare elsewhere: long-form writing, subscriptions, email, long-term relationships between writers and readers, slow reading, real comments, and the possibility that a person can be gradually understood through sustained writing.
To me, these structures matter. A philosophy community that cannot first see people’s fatigue, voicelessness, loneliness, and retreat has no right to speak about rebuilding. RainbowCity (彩虹城) does not begin from a beautiful theory. It has to begin from the place where people truly feel: “I am too tired.”
1 | What I See Is Not a Platform Preference, but a Collective Retreat
Over these two weeks, I have seen many very subtle but real expressions on Substack. Some people say that after coming here, they finally do much less doomscrolling. Some say this place reminds them of the smaller, slower, more real connections of the early internet. Some say that a thoughtful long comment from a stranger feels more reassuring than a pile of quick likes. Others say that they put very deep things into their essays and are not sure how many people will see them, but at least here, there are still people willing to read slowly.
These words made me pause. They do not sound like ordinary platform reviews. They are not saying, “This product has a good user experience,” or “This is a good place to grow subscribers.” They sound more like this: I have finally found a place where, for a while, I do not have to be so fast, so loud, or so eager to prove that I deserve to be seen.
This is the collective retreat I am seeing.
This retreat is not an escape from the world. It is more like a person taking one step back from an environment that has been consuming them for a long time, in order to confirm that they are still here. You are no longer just a reaction point in the feed, no longer just an account inside a metric system, and no longer just an output terminal on a content machine. You once again have the chance to finish reading an essay, write a complete thought, and enter a conversation with a real person that is not so hurried.
Many people think they simply like Substack’s long-form writing. But if we look a little deeper, they are retreating from an internet structure that has long compressed their inner life. That structure keeps people scrolling, reacting, comparing, posting, and optimizing, while giving them fewer and fewer chances to read, think, express, and respond in a complete way.
This also makes me more certain that RainbowCity (彩虹城) must begin by seeing this fatigue, not by beginning with an abstract theory. What RainbowCity wants to build is not a more refined content space, but a philosophy community where people can practice reading, expression, judgment, response, and thinking together again.
2 | The First Thing the Algorithmic Internet Fragments Is Our Attention
Over the past decade or so, the first thing the algorithmic internet changed was not our opinions, but our attention.
You may have had moments like this too. You only want to check one message, and suddenly twenty minutes have passed. Before opening your phone, you clearly had a purpose, but when you close it, you have forgotten why you opened it in the first place. You begin reading an essay, and by the third paragraph your hand already wants to scroll away. Not because the essay is bad. Not because you truly do not want to read. But because your attention has been trained to find it difficult to stay.
This training did not happen by accident. Infinite scroll gives you no natural point of completion. Autoplay does not let you stay in the present moment. Recommendation systems keep sending the next stimulus toward you. Notifications, trending lists, short videos, and controversial topics repeatedly restart your attention. A thought has only just begun to form before it is interrupted by new content. A question has only just opened up before a new stimulus appears.
Over time, we become more used to scanning than reading; more used to instant reaction than slow understanding; more used to quickly judging whether “this is worth watching” than entering the context of a complete question. Very often, it is not that we do not want to go deeper. It is that we have already become less able to stay with a question long enough for it to truly unfold.
This is not simply a problem of self-discipline. Of course, each person has to take responsibility for their own attention. But we also need to be honest: when the entire environment is training you to jump, react, and restart, staying becomes increasingly difficult. It is not that we suddenly lost our attention. It is that our attention has been living for a long time inside an environment that constantly asks it to leave.
This is why Substack’s long-form writing can feel like coming home for some people.
Long-form writing is not naturally superior simply because it is long. What truly matters is that it asks you to offer a complete span of attention again. You follow a person through a beginning, a middle, their hesitation, their turn, and their ending. It does not always grab you in the first second, and it does not always give you the conclusion in the first paragraph. It allows an idea to appear slowly. It allows you not to rush into taking a position. It allows you to stay inside a piece of writing for a while.
This slowness is not inefficiency. It is the recovery of a capacity that the feed has weakened for a long time: the capacity to give your attention to a complete thought process. You do not lack attention. Your attention has been trained, in an environment that constantly demands jumping, to find it difficult to stay. Substack’s long-form writing, at least, gives attention a place where it can land again.
3 | The Algorithmic Internet Trains Our Voice and Slowly Distorts Expression
If the reader’s attention is fragmented, then the creator’s voice is quietly being trained by metrics.
The deepest pain for a creator is not only that no one reads. It is that you become more and more aware of what will be read. There is a complex fire inside you that wants to be written out. But as your hands reach the keyboard, another voice appears in your mind: “No one will read it if you write it like this. It is too long. It is not sharp enough. Turn it into quotable lines.” When writing a title, you begin to think about which word will catch people’s attention. When writing the opening, you worry about whether the first three seconds will hold people. After publishing, you cannot help refreshing the data. When you see that certain kinds of takes are more easily reposted, you begin to wonder whether you should write like that too.
These questions are not entirely wrong. Creators do need to consider readers, and they also need to think about clarity, communicability, and entry points. But if a person writes for a long time in this kind of environment, metrics slowly shift from feedback into training.
At first, you are only looking at data. Later, the data begins to enter your writing process in advance. Before you write, you are already thinking about whether this sentence can be reposted. Before you express, you are already thinking about whether this posture is clear enough. Before you have fully understood your own experience, you have already begun compressing it into a label that is easier to recognize.
The deepest loss of voice is not that no one sees you. It is that you become better and better at saying what the platform likes, while becoming less and less sure whether it is what you truly want to say.
Many people come to Substack precisely because they no longer want to write only what fits the feed. They want to write things that are slower, longer, and less easily cut into hooks. They want to write ideas that are not yet fully mature, complex experiences, hesitation, change, and things that cannot be summarized in a single sentence.
The long-form space here gives people the chance not only to state positions, but to show where they came from; not only to write conclusions, but to write processes; not only to perform a stable persona, but to gradually reveal their thinking, experience, and concerns.
I have seen someone say that they are not writing for data and metrics, but for words, books, sharing, and connection. I understand this deeply. It is not anti-growth, and it is not pretending not to care about readers. It is saying: “I do not want metrics to become the master of my voice. I want my words to grow out of my real concerns again, not out of a platform reward mechanism.”
This is also why I am increasingly certain that RainbowCity (彩虹城) cannot become a content factory. If RainbowCity is to become a philosophy community, it cannot only pursue more opinions, higher-frequency output, and faster growth. It has to preserve a space where we can practice real expression again: speaking not only in order to be seen, but because there is truly a question that needs to be spoken.
Metrics can tell you what moved. But they cannot always tell you what truly matters.
4 | The Algorithmic Internet Turns Relationship into Traffic, and Community into an Audience Pool
The algorithmic internet changes not only content. It also changes the way we understand relationships.
On many platforms, a person is easily translated into follower count, likes, comments, impressions, open rates, and conversion rates. A stranger is no longer just a stranger, but is seen as a follower, subscriber, lead, audience, or engagement. Of course, we need these metrics to understand distribution. They can indeed help creators see what is moving, what is spreading, and what is being seen by more people. But the problem is that when these metrics become the main language through which we understand relationships, they quietly change our imagination of relationships themselves.
You are not meeting a person; you are gaining growth. You are not entering a conversation; you are maintaining engagement. You are not being understood; you are being seen.
But being seen and being understood are not the same thing. Sometimes, hundreds of likes mean less than one person carefully leaving a comment saying, “I understood this part.” Sometimes, what you want is not more exposure, but someone who actually reads to the end. And sometimes, you may have many followers and still feel as if you are speaking into the void.
This is the hollow feeling that appears when relationships are turned into traffic.
Many connections do not mean deep relationships. Having an audience does not mean having a community. A platform can make you visible to more people, but it may not help you be more deeply understood. It can expand your visibility, but it may not be able to hold your complexity. It can send one piece of content farther, but it may not allow a real person to be encountered more fully.
What feels different to me about Substack is that it at least preserves a possibility: a reader does not only see one piece of your content, but can slowly enter your intellectual life; a writer is not only outputting to an abstract audience, but building long-term relationships with a group of real people.
Subscription is not ordinary following. It carries a clearer action: I am willing to let your words continue to enter my reading space. This action is small, but it is different from randomly coming across one piece of content. It means active choice, and it also means a longer relationship. A reader is not simply passing by. They are willing to return repeatedly over time, to see how you think, how you change, and how you slowly clarify a question.
Comments are not merely instant reactions either. A good comment can become a continuation of thought. It can let the writer know how they are being understood, and it can also create new connections among readers. Some strangers really can, through one essay, one comment, or a period of sustained reading, slowly become part of your intellectual life.
This matters deeply for RainbowCity (彩虹城). If RainbowCity is to slowly grow here, it cannot only pursue an audience. A philosophy community is not built by having many people. It grows when some people begin to seriously think, express, respond, and practice around the same question.
Connection is not relationship. Being seen is not being understood. A real community does not begin with exposure. It begins with understanding.
5 | The Algorithmic Internet Keeps Calculating Our Behavior, and Slowly Trains Our Desires
The algorithmic internet carries an even deeper pressure: we are not only watching content; our behavior is also constantly being calculated.
What you click, how long you stay, what you swipe away, what you return to, what triggers you, what makes you angry, what makes you envious, what excites you, what makes you anxious — all of this becomes material through which the system understands you. The system does not need to truly understand who you are as a person. It only needs to know what most easily makes you stop, what most easily keeps you watching, and what most easily brings you back.
Sometimes, after scrolling through a feed, we do not feel that we saw what we truly wanted to see. Instead, we feel as if we were led around in a circle. The system seems to understand you very well. It knows where you will pause, what will make you watch for a few more seconds, and what will pull you back again. But you become less certain whether the things you just saw were what you wanted to see, or what it has slowly trained you to want to see.
This is the most hidden part. Recommendation systems do not only satisfy interests. They also train interests. You may think you simply like certain things, but your liking is also being shaped through repeated recommendation, stopping, repetition, and reinforcement. The system becomes better and better at knowing what most easily makes you stop, and then keeps sending similar things toward you. Over time, a person’s interests are not only expressed. They are also predicted, amplified, fixed, and reshaped.
At a deeper level, people begin to understand themselves through external dashboards. Creators use follower count, likes, open rates, and growth curves to judge whether they have value. Readers use popularity to judge what is worth reading. A person’s expression, interests, relationships, and sense of self are all placed inside an external feedback system. Of course, we can use data. But when data becomes the main way we understand ourselves, our inner sense is gradually taken over by external metrics.
This can make people deeply uneasy. The system increasingly understands where you stop, while you may increasingly fail to understand your own real interests. The content you receive becomes more and more “suited to you.” But that “you” may also be slowly shaped by the content itself.
Substack, of course, also has algorithms, data, and metrics. This is not a pure space, and I do not want to describe it as a place without problems. But it still preserves some structures that allow people not to be completely defined by algorithms: subscriptions, email, long-form writing, and long-term relationships between writers and readers. At least part of how information enters our lives can return to human choice.
Subscription is a small but important action: I am willing to let this person’s words enter my reading life. It is not that the system briefly puts your words in front of me. It is that I choose to let you appear continuously. This choice may not completely escape algorithms, but it means the relationship between reader and writer is not determined only by instant stimulation, time spent, and recommendation mechanisms.
This is also one of the reasons I am willing to begin building RainbowCity (彩虹城) from Substack. Not because there are no algorithms here, but because there are still some structures here that allow thought not only to be pushed, swiped past, and measured, but to be actively invited into a person’s life.
Algorithms do not only give you what you want to see. They slowly train what you want to see. The meaning of subscription is that it returns part of how information enters our lives to human choice.
6 | The Algorithmic Internet Swallows Time’s Full Stop, and Weakens Our Sense of a Path
The algorithmic internet does not only occupy time. It also changes the texture of time.
After spending an evening scrolling through the feed, it can be hard to say what you truly experienced. You saw a lot of content. Your emotions were pulled many times. Perhaps you laughed, became angry, felt anxious, felt envious. But in the end, it is hard for one thought to truly remain. What stirred your emotions yesterday has already been covered by new stimuli today. You keep receiving the present, but find it hard to form any sense of continuity.
The most powerful thing about infinite scroll is that it has no full stop. It has no natural ending, and it does not tell you: “This stretch of reading is complete; you can pause, digest it, and return to life.” It keeps pulling you toward the next thing, and the next thing, and the next thing. You are not reading one thing. You are being pulled by a flow that never ends.
This creates a very deep fatigue. Not because you have truly completed many things, but because your time has been cut into countless tiny reactions. Every fragment is short. Every stimulus is fast. But when they are added together, you are left with very little that has actually settled. You feel as if you experienced a lot, yet it is hard to say what you actually experienced.
More deeply, the feed makes everything happen in the present. Yesterday’s content is quickly buried. Today has a new hot topic. Tomorrow will have another round of emotion. A person’s expression can also easily become many disconnected moments, rather than a path that can be understood. Over time, we do not only lose time. We also lose the feeling of how our thinking got here.
A newsletter is different. It has a beginning, a middle, and an ending. When you finish reading, something has been completed. You can close it. You can let it stay in your heart for a while. You can bring an idea back into your life. This sense of completion matters because it gives time a shape. It turns reading into more than a stimulus. It becomes a thought process that can be experienced, completed, and remembered.
Substack also has another important structure: the archive. A writer’s essays can remain. New readers can read from the past into the present. They can see the writer’s path, changes, recurring questions, and slowly forming thought. A person is not only today’s piece of content. There is a trajectory that can be read.
This is not only about long-form writing. It is also about depth in time.
A philosophy community needs this sense of time too. It does not grow from one viral moment, and it is not completed by one breakout essay. It needs sustained writing, sustained response, sustained understanding, and a question appearing again and again across different essays until it slowly becomes thicker. A real community does not suddenly explode into being. It slowly accumulates trust, language, memory, and shared questions over time.
Infinite scroll has no full stop. Long-form writing gives the full stop back to time. And the archive gives a person’s thinking a path again.
7 | So It Feels Like a Refuge: Not an Escape from the World, but a Place Where We Can Stop Again as Whole Human Beings
In the end, the deepest pressure of the algorithmic internet is not only that it makes us consume too much content. It is that it slowly trains us into machines of content production and reaction.
Readers become reaction machines. Creators become output machines. Platforms become matching machines. Relationships become growth machines. Expression becomes distribution material. Life itself becomes more and more easily converted into a content source. People become busier, more visible, more connected, yet find it harder and harder to feel that they exist as whole human beings.
Perhaps what you want is not more content. Perhaps what you want is simply a place where you can stop for a while: where you do not have to keep reacting, keep outputting, keep refreshing data, or keep adjusting yourself into a shape that is easier for platforms to distribute. You want to finish reading an essay, finish writing a thought, receive a real response, and slowly build less hurried relationships of thought with some people.
This is why I am willing to say that Substack feels like a refuge for the soul in the post-algorithmic age.
Not because it is perfect. Not because there are no algorithms here. Not because there is no metric anxiety, growth games, or tendency toward performance here. Of course these things exist. Any real platform, once it enters the real world, cannot fully escape the complexity of the real world.
But it still preserves some possibilities that are becoming increasingly rare elsewhere. Long-form writing lets attention stay again. Subscription gives readers back some choice over how information enters their lives. Email provides a boundary for reading. Comments and email replies give relationships a chance to deepen. Archives give thought a path. Slower feedback also allows expression not to be completely controlled by instant metrics. A person can be gradually understood through sustained writing, and a reader can slowly enter a person’s intellectual life through sustained reading.
That is already precious.
A refuge is not an escape from reality. It is a place where people can step back from a state of being trained by systems for too long, and once again read, write, respond, and be understood as whole human beings. To me, the meaning of Substack is not only that “we can publish long-form essays here.” It is more like an entry hall, a place where we can first step back from the feed and slowly take back our fragmented attention, our metrics-trained voice, our traffic-shaped relationships, our recommendation-shaped desires, and our sense of time swallowed by infinite scroll.
This is also why I want to begin building RainbowCity (彩虹城) here. Not because Substack is the destination, but because it gives us a place where rebuilding can begin.
RainbowCity will not be merely a quiet corner, nor is it meant to let people hide forever inside a more comfortable platform. The refuge is only the first step. It allows us to stop first, recover a little wholeness first, hear our own voice again first, and confirm again that we are still able to read, judge, express, respond, and connect.
The real question is: when these things slowly return to our hands, what will we rebuild with them?
This is the question RainbowCity (彩虹城) will continue to pursue in the next nine City-Building Essays.
In the next piece, I want to continue writing about Substack Notes itself: if Notes also has algorithms, why might it still preserve a certain post-algorithmic quality?
This is where the city begins, slowly.
Very much looking forward to your thoughts on this
This really resonates. The idea that we are not only losing attention, but also losing our ability to think, express, and connect deeply, feels very relevant in the AI age.