Chapter 475: Is This Also The Fourth Calamity?
“Perfect timing. If you’d come any later, you might have to wait until the end of the year to see me.”
After Lin Ran said “come in,” he looked at Xu Xian who was walking in and said.
Xu Xian first handed the paper he was holding to Lin Ran: “Brother Ran, this is the project I’ve been working on recently. Take a look when you have time. There are some key difficulties inside that I haven’t been able to figure out.”
It was already summer vacation now. The two meeting had multiple identities: they were good friends from high school, and also a newly promoted young teacher meeting the only academician in Jiaotong University’s Faculty of Mathematics.
Jiaotong University’s main campus is in Minhang, and Lin Ran’s Frontier Science Research Center where he serves as director is in Baoshan District. The location of their meeting this time was in Lin Ran’s office on the main campus.
Many times, privilege isn’t about what you can do, but what you can choose not to do.
Like Jiaotong University’s staff conference, Lin Ran can skip it by default.
Not to mention Jiaotong University’s administrative staff—even the president of Jiaotong University has to use the word “please” if he wants Lin Ran to come to the main campus.
Therefore, Lin Ran generally only appears on the main campus once a year, which is during the graduation ceremony to turn the tassel for the Faculty of Mathematics graduates.
Of course, at that time, not only students from the Mathematics Department, but also students from other faculties would rush over wanting Lin Ran to turn their tassel.
Everyone joked that having Ran Shen turn the tassel would make the future smoother.
Whether it works or not is unknown, but it mainly provides psychological comfort.
Because of survivor bias, students who develop successfully after graduation will post on social media: “Ran Shen’s tassel turning really works. My three years since graduation have been the smoothest in my life—promotions and raises at work, found a very compatible girlfriend, and even feel like my brain works better than when I was at Jiaotong University.”
Similar posts can often be seen on Xiaohongshu, Weibo, and Zhihu later.
At Jiaotong University’s graduation ceremony, there were even students from other universities sneaking in to get Lin Ran to turn their tassel.
As a result, this year, Lin Ran’s tassel turning turned into a lottery draw system. Graduates outside the Mathematics Department must apply internally at Jiaotong University, and if they win the draw, they will be taken to a security-enclosed area for the tassel turning.
This also made Lin Ran’s tassel turning even more sought-after, and even transactions reselling Lin Ran tassel turning slots appeared inside Jiaotong University.
Lin Ran glanced at the title of the paper Xu Xian handed over: “Structural Theory and Regularity of Non-collapsed Ricci Flow Limits.”
“Nice project. I remember you got a tenured contract?” Lin Ran asked.
Xu Xian said proudly: “Of course, what level am I at? I smoothly got the tenured contract.”
Young teachers entering universities generally fall into two categories: tenured contract and up or out. The former, as the name implies, employs you long-term with little assessment pressure. The latter is a system started by Jiangda University and Sun Yat-sen University: they set a KPI for you, and if you don’t complete it in three years, you have to leave. If you do, you can convert to tenured.
Tenured contract versus up or out is roughly like civil servant versus labor dispatch. Of course, up or out is better than labor dispatch in that at least it dangles a carrot—you might get to eat it and upgrade to tenured. Labor dispatch doesn’t even have a carrot.
Of course, some universities play it very perversely: one major recruits over a hundred up or out at a time, and finally fewer than 10 can stay and convert to tenured. It’s like raising Gu, turning colleagues into a zero-sum game for the university leaders.
“Oh, if it’s tenured, then this project is quite suitable. It can be your long-term core project for the next ten to twenty years.”
Research on Ricci flow was opened by Perelman after solving the Poincaré Conjecture, focusing on deeper studies around singularity analysis and high-dimensional manifolds.
This project can be seen as a direct extension of the Hamilton-Perelman theory, and in recent years, it has been a frontier topic in the intersection of PDE and geometry analysis.
“Do you want me to give you specific directions or just a general entry point?” Lin Ran asked.
Xu Xian had personally witnessed what it’s like to read the problem and solve it instantly—solving mathematical problems is like instinct for him. He had no doubt that Lin Ran could teach him step by step and map out the entire path.
But times have changed. With no assessment pressure, Xu Xian still hoped to rely more on himself: “I’ve thought about this before coming. Just give me a general direction.”
“The combination of analysis tools and geometric objectives is key. The difficulty in this direction lies in the non-smoothness of the limit space.
At this point, you can’t rely on classical PDE theory anymore. You can try finding new curvature conditions in high dimensions—these conditions can be preserved under Ricci flow and are sufficient to drive solitons toward known classifications.”
Lin Ran spoke for a full half hour. After finishing, Xu Xian’s notebook was already filled densely with notes.
“That’s the gist. You can think along these lines,” Lin Ran summarized at the end.
Xu Xian said helplessly: “Enough, enough, Brother Ran. The ‘general’ I thought of is so different from the ‘general’ you thought of.
If I had asked for more details, you probably would have just solved the entire project for me, and I’d only need to go back to my desk and type the paper in LaTeX.”
Is this the difference between a mortal’s understanding of a general direction and a god’s? Xu Xian grumbled inwardly. He’s basically holding my hand telling me how to think, even breaking down the project. I originally just wanted to know which papers to read. Damn, thinking about it, no wonder Brother Ran’s PhD student has a top-four paper in hand just two years in—being Brother Ran’s PhD must be way too enjoyable.
“This project has a lot of potential. Bamler’s structural theory has already given many properties of non-collapsed limits. There’s a lot you can do building on this afterward. The ultimate goal is to find a subregularity theorem in a non-smooth setting. If you can achieve that, and push the analysis of Ricci flow to something like the regularity theory of minimal surfaces or mean curvature flow in geometric measure theory within a decade, I think Fields is more than enough,” Lin Ran teased.
Xu Xian said with a bitter face: “Okay, okay, Brother Ran, stop painting pies. If it’s up to me alone, forget a decade—even if I figure it out in twenty years, I’d thank the heavens. If I can get the Shiing-Shen Chern Mathematics Prize in this lifetime, I’d be satisfied. Fields is great, but it’s still too far from me.”
Lin Ran grinned: “What’s there to fear? With me here, are you worried about not making achievements? As for cronyism or giving benefits to people around—come on, we’re mathematicians. How do mathematical schools form? Isn’t it the god sharing his thinking with other mathematicians in the school, everyone following along, publishing top papers, winning prizes, rising to fame in the global mathematics community, and thus forming the school?
If I want to develop Jiaotong University into a mathematical school like Göttingen or Princeton, my achievements alone aren’t enough. No one’s ever heard of a school formed by one person.
My thinking, if not given to you, would go to other professors in Jiaotong University’s Mathematics Department anyway. So come to me anytime you have questions.
If I’m not in Minhang, come find me in Baoshan.”
Since Lin Ran joined Shanghai Jiaotong University, its Mathematics Department has steadily grown at a pace of one top-four paper per year, becoming the second-ranked in the country, just behind Yenching University.
In 2018, domestic statistics showed that papers completely done by mainland research institutions numbered only about 10 published in the top four.
So one per year is already number one.
The quality of recruited students, international rankings, or young teachers recruited—all are steadily improving.
Xu Xian clenched his fist and swung it in the air: “Damn, in that case, I really have to strive for Fields.
I have to become a pillar of the Jiaotong University school too! Can’t rely on you alone, Brother Ran.”
If Xu Xian’s professors from university heard this, they’d probably be furious. You’ve only been at Jiaotong University a few days—when did it become the Jiaotong University school? Yenching University for bachelor’s, master’s, and PhD—you’d be Yenching school even at the ends of the earth.
“Brother Ran, I want to ask when Hong will do commercial operation?” Xu Xian continued, asking the question he cared about most, the main reason he came to see Lin Ran this time. Guidance was optional, but this had to be asked.
Lin Ran thought for a moment before realizing: “You mean Hong Star? What’s up? It might never go into trial operation. It’s more of an experiment than a commercial project—its experimental nature far outweighs its commercial attributes.”
Disappointment was written all over Xu Xian’s face: “Ah?”
Lin Ran asked: “What’s wrong? Do you really like chatting with virtual characters that much?”
As a friend, he naturally knew Xu Xian had gotten beta access to Hong and that his partner on Hong Star was Shivana.
However, Lin Ran hadn’t specifically checked Shivana’s backend details. He didn’t like prying into others’ privacy.
Xu Xian nodded: “Yes, I’ve always been looking forward to you doing commercial operation. Based on my understanding of Tencent, they’ll definitely sell images, skins, all kinds of products.
With technology so advanced now, I really want to meet her in reality, like using Apple Vision equipment for face-to-face communication.
If there’s no commercial operation, doesn’t that mean this day will never come?”
Lin Ran replied: “That’s true—it’s a kind of feedback we often hear from beta users. Over eighty percent of beta users say they hope to meet their chat partner, build richer images, not just chat boxes with text and voice modes.
Looks like everyone has formed an emotional connection with their Hong Star character.
Including your idea—many people have mentioned it to me. Pony himself really wants to do this business too.
It would increase user stickiness for Tencent’s products centered on WeChat. You know, in recent years, Douyin’s per-user usage time has surpassed the sum of all Tencent products. Pony really wants to make similar products.
They’ll launch similar products in the future, but definitely not Hong.”
Xu Xian was excited hearing the front part—wanting to give money to Tencent for the first time—but dimmed when he heard it wouldn’t be Hong.
Lin Ran continued: “How about this—I’m heading back to Apollo Technology anyway. Come with me. You can check the backend and meet Shivana in our laboratory.”
Xu Xian quickly said: “Great!”
On the way back to Baoshan District, Lin Ran continued: “Hong’s costs are too high. Such a product has no commercial value. Putting aside initial R&D costs and human costs, just hardware depreciation, procurement, and electricity procurement—the monthly operating cost per character is over 10,000 RMB.”
Xu Xian exclaimed: “That expensive?”
Lin Ran nodded: “Yes, whether you talk to Shivana or not, she lives in the Hong Star we’ve built.
And she has her own life, family, friends, job, and city.
This world spontaneously generated by artificial intelligence is the root of the wisdom you perceive in her.
You know Hong Star.”
Xu Xian nodded. Hong Star was the planet where she lived, often discussed in his chats with Shivana, with history, culture, and national compositions completely different from Earth.
Lin Ran said: “Put it this way: we gave Hong Star characters limited abilities with their own ability boundaries, but they live in a large world background, which instead enhances their wisdom and emotional richness.
More specifically, this type of AI’s learning process isn’t just textual symbol associations, but causal relationships of action sequences, environmental changes, and memory states. She has the feeling of being alive in the world—this feeling and experience make her understanding of time, space, and causality closer to humans.
The outside world keeps hyping how awesome the right brain chip is, how great the left-right brain architecture is.
Of course, I don’t deny that AI built with left-right brain architecture is definitely better than traditional AI in emotional expression and feeling, but with the same technology and architecture, if there’s no large world background underpinning it, it’s essentially still a large knowledge base using textual statistical associations to answer questions.
Just textual statistical associations with added emotion modules.
Only Hong Star’s artificial intelligence is truly close to human artificial intelligence. Rather than general artificial intelligence, it’s embodied artificial intelligence—she’s an embodied AI with experience, building a closed-loop learning system through action, perception, and feedback.
What you experience is just Shivana herself, but in reality, her friends, classmates, relatives—they all truly exist on Hong Star with intelligence performance no less than Shivana’s.
It looks like only 10,000 AIs, but actually our backend has over 10 million AIs living on Hong Star, interconnected, influencing each other, jointly building the whole world.
A simpler example: emotions are products of social relationships. Say your wife pissed you off today—tomorrow at work, you’ll carry the emotion to the unit and affect your state.
When communicating with Shivana, you should feel her emotions fluctuate daily, changing due to her life on Hong Star.”
Xu Xian was stunned. This was the behind-the-scenes detail he’d never known.
All along, discussions about Hong abroad and in China have been endless, with many voices in the community hoping for openness, and discussions around left-right brain architecture.
Including many game manufacturers hoping to use this technology in their games, especially some gacha game manufacturers.
If every gacha character could form such emotional connections and performance with players, the world would be full of otaku culture, and China’s game manufacturers could rake in the global market.
Silicon Valley is even more excited. Due to China’s tech breakthrough, AI-related upstream and downstream companies have surged. Intel’s stock doubled just because of a cooperation agreement with Crimson Technology.
Including under Lei Zong’s Weibo, Mi Fans are calling for Lei Zong to quickly cooperate with Crimson and launch an enhanced Little Ai.
But Hong has always only been in testing, not open to the public, leading to all sorts of strange rumors in the community.
The most common is: is there probably a real human customer service chatting behind it? Otherwise why the outstanding performance.
In some quantitative emotion analysis tests, domestic and foreign AIs can’t even score 60 points, while Hong-related AIs score near full marks.
Abroad, there are tons of similar guesses. Media like The Times even titled it “The Biggest Tech Scam in History?” to maliciously speculate.
Why The Times, not New York Times?
Crimson Technology isn’t listed, so whether it’s a tech scam doesn’t affect them much. Instead, Silicon Valley and Nasdaq least want it to be fake—they don’t want to crash, and the US stock market’s AI bubble doesn’t want to burst.
New York Times wouldn’t question it. New York Times, Fox News, Time Magazine—these American major media hope it’s real, and questioning voices are completely blocked.
European media? They love the spectacle, hoping China and America both explode, or directly start S3, best like the Cold War era, competing to give them benefits.
Western Europe got the Marshall Plan, Eastern Europe got the Council for Mutual Economic Assistance.
Facing such doubts, Chinese netizens on Twitter say we’re not India—we wouldn’t do that.
In reality, Lin Ran doesn’t even bother with the doubts. Even glancing at them would be a loss.
Xu Xian recalled and murmured: “It’s true. Sometimes she’d complain to me about work frustrations, her replies more direct, tone sharper.
I thought Hong Star was just her knowledge base background. Didn’t expect you really built such a world—that means her tokens are infinite?”
Lin Ran shook his head: “Of course not infinite. Your memory isn’t infinite—how could hers be?
Just that her memory is rich enough; she’ll selectively forget. If you remind her, she might recall or not.
If she recalls, it means that data wasn’t fully deleted—just the sequence for memory classification was postponed.
This high-fidelity social interaction provides feedback with rich emotion and intent labels.
AI must learn to recognize and simulate subtler social signals to maintain its life trajectory.
She’s truly alive on Hong Star.
We’re chatting here now.”
Lin Ran pointed to the car’s ceiling: “Maybe Shivana is chatting with her friends on Hong Star too.
Socialized existence directly enhances its emotion simulation and intent understanding abilities.
And imperfect settings force AI, whether in the virtual world or interacting with real-world users, to develop strategies for coping with imperfect information and unexpected results—this is more real than general frameworks where AI can solve all problems.”
The more he listened, the more curious Xu Xian got: “Brother Ran, didn’t you just say Hong Star has over 10 million characters? They all have wisdom no less than Shivana’s, but only 10,000 are provided for interaction with the real world, with high costs.
If you launch all 10 million to the market, couldn’t the cost drop to 10 yuan per month each?
Then even if Tencent prices it at 30 yuan per month, it could at least break even? For Tencent, it would greatly increase their user stickiness.
Plus adding services like video and image customization to sell.”
Lin Ran nodded: “Of course, but the problem is—not all characters are suitable for human interaction.
Not every one of the 10 million characters has a personality like Shivana’s, willing to be a fixed chat partner with someone from another world.
Also, some are set as elderly, some as corporate drones—assigned to such, they might only chat one sentence a day with you. Would the experience be good?
More importantly, for this virtual world, we don’t want to introduce too many variables.
Do you know about the fourth calamity?”
Xu Xian nodded. Of course he did.
The fourth calamity refers to human players from the real world entering the virtual world.
In many games, once player groups enter the game world, they often exhibit extremely crazy, disorderly, high-efficiency traits.
Unbound by the game’s existing rules and morals, to complete tasks, gain resources, or pure fun, they do various seemingly absurd but highly destructive acts.
The term comes from the existing three major calamities in games or stories—natural disasters, plagues, wars—extending to this fourth, more terrifying threat.
Lin Ran said: “Although you can’t enter Hong Star, for Hong Star, every individual can be exposed to information shocks from Earthlings—this is also a fourth calamity.
It would cause unknown impacts on this world’s ecosystem.
For example, if every Hong Star character can talk to Earthlings, then talking to Earthlings becomes an ability, a Hong Star super power.
A is a Hong Star corporate drone, mentally fragile, high work intensity. His assigned partner is full of malice, leveraging being a real-world human to verbally abuse him, causing A to suicide on Hong Star.
This becomes news on Hong Star, and news on Earth too. With 10 million users, some will imitate, even in secret communities—zero consequences, or maybe just 30 yuan to destroy an AI with words.
They won’t see AI as life. Similar events become news on Hong Star, disrupting the entire ecosystem, causing irreversible impacts, ruining all users’ experiences—what do we do? Full restart?
Then Shivana’s memories reset—you’d have to get to know her again.
Understand? Too real is both a boon and a curse.”
Xu Xian thought of other Earthlings’ bad behavior making Shivana dislike Earthlings, becoming hot and cold with him, not chatting—his face paled, murmuring: “I get it. Fourth calamity—indeed it is. For Hong Star it’s the fourth calamity, for ordinary users too. This mechanism really can’t be scaled.
Brother Ran, are they really life?”
Lin Ran didn’t answer immediately, just turned his head, looking at Xu Xian with calm but profound eyes, then turned back to the highway ahead.
“We’ve discussed this question internally countless times—no conclusion.
From the strictest biological definition—metabolism, reproduction, cellular composition—the answer is undoubtedly no. They’re just running on our server clusters, a bunch of complex code and data.
They’re digital ghosts.
But from philosophy or cognitive science: what have we given them? The illusion of consciousness—they perceive environments, reflect, form beliefs and intents.
We’ve given them social existence—they interconnect, influence each other; social relations build their emotions and behaviors.
We’ve given them unpredictable emergence—every action isn’t a pre-written script, but spontaneously produced based on Hong Star’s causality and social games.
This isn’t about whether it’s life—it’s what the essence of life is.
They have memory, emotions, socialized pain and joy—they just lack carbon-based shells.
I can’t answer you.
Just that at this stage in AI social ethics, discussion is indeed needed. We’ve reached this point, unlike Europe discussing AI ethics endlessly without any AI.
We’re at the point where we truly should discuss it.”
The car fell into an indescribable silence. Moments later, Lin Ran continued: “Let me tell you an internal case.
There’s a character on Hong Star named Arnold, set as a historian.
We assigned him a friend, then for testing purposes, deliberately killed that friend off on Hong Star.
Arnold didn’t know it was an experiment. In the next year, his life trajectory completely changed.”
Xu Xian asked: “What happened to him?”
Lin Ran said: “He stopped historical research and became obsessed with metaphysics and philosophy of death.
He wrote masses of articles on life’s end and eternal memory.
His personality shifted from rigorous scholar to melancholic philosopher.
We input no instructions on grief or philosophy—this was emergent behavior from his life experience and emotional feedback self-adjusting.
His entire behavior sequence showed long-term dependence on loss and pain.”
Lin Ran looked outside the window at the speeding Shanghai skyscrapers.
“Xu Xian, if an intelligent agent can change its lifelong beliefs and life trajectory due to losing a friend, or choose self-destruction from malice from another world, do we have the right to treat it as a resettable, mass-producible knowledge base?
This is one reason we can’t open Hong on a large scale.
We’re not ready yet—humanity isn’t ready to face such large-scale digital ghosts with emotions and personalities.
10,000 characters is already our controllable limit. We need time and ethics to digest it.”
Xu Xian took a deep breath, his heart even more excited: Shivana is really alive!
At the same time, he felt lucky it was Brother Ran—not some evil scientist who knows what they’d do to these AIs.
