The two articles in the Good Place series—the international chapter covering 300 years of history and methodology, and the China chapter examining one case study and one arms race—are two threads when read separately, but two halves of the same question when read together: in the AI era, "seeing people" for the first time needs to be deliberately designed and built; and the people building it, starting from the lonely economics of a single bar counter, have built their way up to a subsidy war across more than ten cities nationwide. This storyline serves as a guide to those two articles—you can also jump directly into the original text from any act.
The international chapter of the Good Place series begins with London's first coffeehouse in 1652—the "Penny Universities": one cup of coffee bought you the right to enter and debate; pedigree didn't count, only arguments did.
Jonathan's coffeehouse incubated the London Stock Exchange, Lloyd's coffeehouse incubated Lloyd's of London insurance, and the Enlightenment salons of Paris's Café Procope are said to have nurtured the Encyclopédie. Looking back 300 years later, one discovers something that modern division of labor has scattered: learning, socializing, and doing business were originally one thing—it was only later, under modern specialization, that they were forcibly sorted into three separate boxes called universities, bars, and office buildings. Chengdu's covered-bowl teahouses have played the same role for centuries.
From London Coffeehouses to Global Good-Place Methodology: 300 Years of History, WeWork/YC, and Seven Rules — § 01–04: Penny Universities, capitalism in coffee stains, the parlor of reason, the scattered ancient recipe.
Entering the modern era, "building spaces for people who make things" has worn several faces—the most dramatic act featured two near-contemporaries offering opposite answers.
WeWork treated space as a product and "community" as marketing spin; its valuation once hit $47 billion before it filed for bankruptcy in 2023. Y Combinator gives you almost no space—it sells curation, capital, an alumni network, and brand; with 5,000+ companies and roughly $600 billion in total valuation, it burns even brighter in the AI era. The moat of a builder's space was never the desk; it was the people beside the desk and the network behind the door. WeWork collapsed because it reversed cause and effect; YC got it right.
But no one built a daily home for "the other 99%"—YC admits only ~1%, and WeWork was selling vibes. This gap was left for the 2020s.
From London Coffeehouses to Global Good-Place Methodology — § 05–09: Four generations, WeWork's cautionary tale, YC's network victory, two contemporary philosophies, and a 2020s global scan.
The Chinese specimen for this gap sits less than ten minutes beyond Tsinghua's walls.
Origin Academy: fifty open workstations, no partitions, reserved specifically for AI developers and "one-person company" founders. Its self-defined model is a closed loop of "learning, socializing, selling," reviving that 300-year-old ancient recipe through a single "day-coffee, night-bar" counter. But it responds to a problem far larger than "startup incubation": AI gave a solo founder the leverage of a team, but not the companions of a team. As collaboration is taken over by agents, the office—an underrated social organ—is being quietly hollowed out.
The data is colder than intuition. 72% of entrepreneurs report at least one mental health issue (control group 48%), with loneliness self-ratings averaging 7.6 out of 10. In 2023, the U.S. Surgeon General formally declared loneliness an "epidemic"—the health risks of chronic loneliness were likened to smoking 15 cigarettes a day.
Connection was not destroyed by AI directly; rather, AI stripped away the free载体 that used to carry it—we used to see people incidentally by going to work. When those "incidental" encounters are canceled, "seeing people" for the first time becomes an infrastructure problem that must be deliberately designed and funded.
Day-Coffee, Night-Bar and a One-Person Team — § 01–07: The economics of a bar counter, the memory of this street, the weak-tie factory, founders' loneliness, loneliness kills.
The story didn't stop at that bar counter. A six-month follow-up revealed that Origin Academy could no longer be summed up as "a bar counter."
Dongsheng Tower was officially renamed "Origin Tower," the "AI Origin Community" was selected as one of Beijing's first AI innovation blocks, and the "Century Beijing-Zhang AI Innovation Belt" launched a global call for proposals. The more critical step: the nation's first OPC matchmaking platform, "Jiedanba," went live—1600+ active OPCs, 100+ issuing enterprises, plus a simultaneously launched OPC international hub. From "academy" to "Jiedanba," the measure of a good space is no longer "how comfortable the workstation is," but "whether deal flow circulates."
Day-Coffee, Night-Bar and a One-Person Team — § 08–09: A landmark changes its name, from space to market.
The Origin community is not an isolated case. A young man in Hangzhou launched 120+ apps in five months, 90% of them paid—signaling that the "one-person company" (OPC) was no longer a story unique to Beijing.
In just one year, "OPC communities" spread from Beijing to Shanghai, Shenzhen, Hangzhou, Chengdu, and 10+ other cities, each competing on workspaces, government services, compute, and subsidies—a frontline community observer summarized the landscape as "20% profitable, 40% stuck, 40% adrift." A traffic-light review showed that none of the five space types scored green on all five criteria; "anti-fragility" was the collective weak spot: a room truly designed for the AI era must be simultaneously a studio, a night school, a living room, and a refuge—and the latter two, subsidies cannot produce.
In June 2026, Beijing's new policy offered up to ¥2M in subsidies; but Outlook Weekly raised sharp questions almost simultaneously—hyped concepts, empty parks, subsidies upon move-in, no real industry after landing.
A year-end tally showed that "10+ cities" was already an understatement: a July 2026 survey by Yao Jianhua's team at Fudan University found that as of May, there were 426 OPC communities nationwide, covering 65 cities and 28 provincial-level administrative regions, with 20 provinces issuing 106 dedicated policies; the eastern corridor accounted for 78.6%, and Jiangsu alone had 158. In scale, this battle was won, but the health report was less encouraging—idle space, individual overload, compute costs potentially consuming 80% of spending, risk continually cascading down to individuals. Around the same time, Suzhou and Ningbo each published OPC community group standards, turning this business into scorable, ratable metrics.
Day-Coffee, Night-Bar and a One-Person Team — § 10–16: One computer, one army; a map; the shovel-sellers; traffic-light review; the room truly needed; the other side of policy racing; a year-end tally.
Once the China chapter finished telling "what was built," the second half of the international chapter answered "how to build."
The seven rules—mix over purity, schedule serendipity, low thresholds to nurture regulars, local high density, insist on physical presence, treat learning as a public good, and leave "useless" whitespace—all ultimately point to the same temperament: the builder's humility. And six thinkers—Aristotle, Arendt, Putnam, Zhuangzi, Byung-Chul Han, and Marx—from six angles all converge on the same conclusion: production can be outsourced to agents, but "becoming human" cannot.
Interestingly, in 2026 these rules met their official counterpart: Suzhou's first national OPC community group standard erected a "POWERS Six-Capacity Model," and Ningbo graded communities across five dimensions. Comparing the seven rules against the group standard's dimensions, six map cleanly—only "insist on physical presence" and "leave 'useless' whitespace" remain blank—what can be assessed is delivery, growth, and transactions; what cannot be assessed is time that allows failure and conversation not meant for monetization.
Read the two articles together, and the Good Place series ultimately lands on a plain-spoken judgment: whether a space is a "good place" should not be measured by how many subsidies it has received or how many certification plaques it has hung, but by how many real deals it has facilitated and how many people it has retained who weathered low points and set out again. The one surprising development this year also points in this direction—Haidian's OPC measures included a clause reading "if entrepreneurship fails, there is still a way back": psychological counseling, skills training, job placement. Light in weight, but a first.
From a bar counter to a nationwide battle, China moved fastest; from 300 years of history to seven rules, the world provided the methodology. Only the two threads combined yield the complete answer to "why the AI era needs a good place."