In addition to highlighting the increasing risks of climate change to Himalayan communities, the collapse and flooding have also drawn attention to Chinese authorities’ unrelenting efforts to clamp down on and control the tone of information in the wake of natural disasters. Chinese and overseas netizens alike have noted the stark asymmetry between the regular updates and open information sharing from Nepalese authorities, and the tight press controls, limited information sharing, and lack of first-person survivor accounts on the Chinese side. Recent reports that ten people in Tibet have been punished for “spreading rumors” online about the flooding only serve to reinforce these observations.

CDT Chinese editors have been tracking online censorship of video and images of the flooding, as well as search censorship on various platforms and muting of Weibo topics related to the flooding, death toll, number of missing, and other disaster-related information. Their observations closely match the well-established playbook for handling of accidents and disasters, in which narratives are aggressively steered toward "positive energy" about government-led relief efforts and away from independent accounts or scrutiny of human factors. Translated below is a CDT summary of online information controls tracked and observed in the days after the disaster:

CDT Editor’s Note: On the morning of August 26, 2026, a catastrophic mudslide struck Gyirong County in Shigatse, Tibet, destroying the main building of Gyirong Port [a major border crossing between Nepal and Tibet]. Chinese officials subsequently attributed the cause to a flood triggered by a high-elevation glacial collapse upstream. As of 1:00 a.m. on August 29, Chinese official reports put the toll at 7 dead and 554 missing in Tibet; an earlier report had broken this figure down to include 260 foreign nationals.

In the aftermath of the flooding, numerous clips of surveillance footage spread across the Chinese internet, showing the mudslide swallowing heavy trucks, vehicles reversing to escape, half a highway sheared away, and aerial panoramas of the scene. Two clips in particular—surveillance footage recording the moment Gyirong Port’s border gate was swept away—were restricted on domestic social-media platforms.

1. The Footage That Was Singled Out

Both clips appear to be from fixed surveillance cameras, showing the port’s freight parking lot and customs-and-border inspection area. A roiling flood surges down from the gorge above, water and mud sweeping away the multi-story customs building and inundating everything in sight.

The videos were first published via overseas channels such as the X account "Teacher Li Is Not Your Teacher." Internet users tried repeatedly to forward these two clips via WeChat, without success. Meanwhile, surveillance footage of the same disaster from other angles, filmed during the same window of time, circulated widely on WeChat, Weibo, Douyin, and Xiaohongshu (RedNote). A considerable share of this content was picked up and reposted in various formats by mainstream Chinese media accounts.

The disparity in censorship of video content seems to be occurring at the level of the images themselves, rather than the accompanying text. The platforms clearly aren’t using keyword filtering to block these videos. Whatever censorship method is being used—video fingerprinting, frame hashing, or model-based recognition of particular scenes—the result is that specific videos are being placed on a blocklist. We haven’t yet been able to determine whether this results from the dichotomy between domestic WeChat and the overseas version of the WeChat app, or from some other technology aimed at identifying these two particular videos.

The expansion of censorship from words to images is a trend CDT has documented with increasing frequency in recent years, in the course of our ongoing work tracking sensitive-word lists.

2. “Sanitized” Search Results on Xiaohongshu (RedNote)

As of August 28, CDT editors found that searching for "Tibet mudslide" on the social-media platform Xiaohongshu (RedNote) returned the following top video results:

• "Live Footage! Gyirong County, Tibet Suffers Mudslide Disaster, Resulting in Large Numbers of Casualties and Missing Persons" (CCTV.com)
• "Gyirong Port Mudslide in Tibet Leaves 3 Dead, 265 Missing; Experts: Flood Possibly Due to ‘Glacial Lake Breach’" (Chengdu Economic Daily’s “Red Star News”)
• "Before-and-After Comparison Images of the Gyirong Port Mudslide Disaster"
• "Gyirong Port Buildings Covered in Silt"

The following content appears further down in Xiaohongshu’s video search results:

• "Caused by High-Elevation Glacial Collapse! Cause of the Gyirong Port Mudslide Announced" (China News Service)
• "Mudslide Disaster in Gyirong, Tibet: Latest Figures Are Three Dead, Two Rescued, 558 Missing" (Tibet Headlines)
• "Five Warning Signs a Flash Flood Is On the Way" (a disaster-preparedness guide)

Most of the content visible in the search results consists of tabloid-style infotainment with no additional investigative reporting—a genre that has become increasingly popular in China in recent years, commonly described as "new yellow journalism" (新黄色新闻, xīnhuángsè xīnwén).

Image and headlines from Sichuan Observer (四川观察, Sìchuān Guānchá, affiliated with Sichuan Radio and Television) on Xiaohongshu
Top: Gyirong County, Tibet Suffers Mudslide Disaster, Resulting in Large Numbers of Casualties and Missing Persons
Middle: 200 technical experts and others from Beijing, Lhasa, Chengdu, Chongqing, and elsewhere have been dispatched to the disaster-stricken area
Bottom: Gyirong County, Shigatse City, Tibet

These cause-of-disaster briefings, limited casualty figures, before-and-after image comparisons, and disaster-preparedness guides are bundled together to form a complete, presentable, and internally consistent “information package.” The original surveillance footage of the border gate being swept away did briefly appear on Xiaohongshu, before vanishing from the search results page entirely.

3. Weibo’s “Trending Topics List” Shows Only Content From Verified Accounts

Based on repeated monitoring from August 27-28, Weibo’s trending topics list saw 714 instances of topics related to the Gyirong disaster falling off the list; of these, 306 topics disappeared abruptly. Among the 12 topics we chose for more in-depth research and cross-checking, several Gyirong-related subtopics displayed the same pattern: only posts from verified Weibo accounts (“blue V” accounts) remained visible on the topic page, while ordinary users’ content did not appear in the results at all. The following are some of these subtopics, with their initial position on the list, amount of time spent on the list, and their “heat score,” which is Weibo’s proprietary, algorithmically computed popularity score for each hot-search entry.

• "Tibet mudslide: 3 dead, 558 missing" at one point occupied the #1 spot on Weibo’s trending topics list. It remained on the list for 20.1 hours, and had a heat score of 704,000.

• "Tibet mudslide: 265 missing, 3 dead" likewise occupied the #1 spot on the list, where it remained for 10.9 hours, with a heat score of 1.35 million.

• "Gyirong, Tibet mudslide: 260 foreign nationals missing" and "More collapses hit the slope above the landslide dam at Gyirong Port." Both of these topics disappeared from the trending topics list in less than an hour.

• "PLA and People’s Armed Police deploy over 500 rescue personnel" once ranked #3, with a heat score of 1.79 million, and remained on the list for two hours.

• "Port police continued directing evacuation even as crowds fled" likewise stayed on the list for under an hour.

Please note that despite search results being "only from verified accounts," this is an editorial observation on our part. At present, it cannot be taken as definitive proof of censorship or soft censorship; we are including it for readers’ and researchers’ reference only.

The following is an inventory of images and news about the disaster that remains visible on Chinese domestic online platforms:

• CCTV’s before-and-after comparison images of the Gyirong disaster zone
• Before-and-after comparisons of Gyirong Port
• High-definition footage of the mudslide disaster zone relayed by a Wing Loong drone
• Footage of a drone transporting firefighters
• Footage of People’s Armed Police soldiers clearing roads
• Fire and rescue teams arriving at the core disaster zone
• Port staff moving against the crowd to evacuate people during the disaster
• Police continuing to direct evacuation even as crowds fled
• News that 30,000 packages of relief supplies were sent to support Tibet’s disaster relief
• News that Apple would donate to Tibet
• News that Premier Li Qiang arrived in Gyirong to direct emergency rescue efforts

4. A Short-Lived Clue

At 3:00 p.m. on August 27, a topic briefly appeared on the trending list — "Gyirong Port only reopened to cross-border traffic six months ago," with a heat score of 347,000 — before disappearing.

The following sequence of dates explains the backstory: on July 8, 2025, Gyirong Port suspended cross-border operations due to a catastrophic flash flood. It resumed operations on January 1, 2026, on which day it processed 425 person-crossings and 64 vehicle-crossings. On August 26, 2026, it was destroyed again by flooding and mudslides. In other words, within fourteen months, the same border crossing was cut off twice by flash flooding and mudslides.

Equally short-lived were the topics "Investigating the recurrence of geological disasters near Gyirong Port" (heat score 230,000, remained on the list for only one hour) and "Tourists say Gyirong Port used to be busy and bustling" (heat score 469,000, and only remained on the list for under an hour).

From this, we can see that questions about site selection, border infrastructure, glacial-lake and ice-collapse monitoring, and early-warning systems failed to sustain any ongoing discussion on the largest Chinese-language trending-topics list.

5. The Trajectory of the Numbers

We also tracked the trajectory of officially reported numbers of the missing and the dead.

At 11:00 p.m. on August 26, the official figures from the Chinese side were 265 missing and three dead. By the morning of August 27, this had changed to three dead and 558 missing (including 260 foreign nationals), a figure that “held steady” for roughly 36 hours. At 5:00 p.m. on August 28, it was updated to five dead and 558 missing. At 1:00 a.m. on August 29, per the latest report from Xinhua, two residents of Rexuo Village had been rescued, and there were seven dead and 554 still missing.

Within that same window of time, reports from the Nepalese side of the border, as relayed on Weibo, climbed steadily from 157 dead, to 270, to 289, to 359. As of 6:00 a.m. on August 29, per the latest data from Nepali police, the death toll had risen to 616.

We believe this looks less like a mass takedown campaign, and more like a finely calibrated operation: one in which public-opinion guidance is accomplished through a saturated information environment, rather than through wholesale post deletion. [Chinese]

A still from the film features animation so bad it beggars belief. The subtitle at bottom reads, “Niu Lai.”

A five-year labor-of-love produced on a shoestring budget by mother/son team Sun Lifang and Xin Yumeng, “The Cow Comes” is a dreamlike coming-of-age tale about a calf named Niu Lai who grows up to be a brave warrior. In its first ten days of theatrical release, the film sold fewer than 300 tickets nationwide, earning about 7,000 yuan ($1,040). But when reviewers and social media users started circulating clips of the film’s astonishingly bad graphics, awkward dialogue, and bizarre plot, audiences began flocking to see it in theatres. Thus far, the film has officially surpassed 40 million yuan ($6 million) in box-office revenue, and is projected to gross over 100 million yuan (nearly $15 million).

“The Cow Comes” has inspired a flood of memes and media coverage.

Moviegoers and theater employees alike have described a carnival-like atmosphere during screenings of the film: audiences laugh uproariously, mockingly recite the dialogue, talk back to the screen, and take videos which are often later shared online. Many theaters, unprepared for the film’s success and lacking publicity posters, resorted to displaying crudely hand-drawn posters, which themselves became objects of online ridicule.

Poster at left: “In Ultra HD 4K: The Cow Comes. Limited-time screening. Utterly bizarre. Caveat emptor: No refunds!!! No refunds!!! Badasses come to watch this badass cow! (The director didn’t give us a poster, so we had to draw our own.)”
Poster at right: “The Cow Comes—to the Big-Screen International Theater, Fusheng Branch, Changsha. This film is all about the lulz. No substance whatsoever.”

There has been considerable backlash to the film’s success, including criticism from some CCTV-affiliated outlets, an article in The Beijing News calling the film “nonsensical,” and reports that authorities in Dalian (where the creators are based) were worried the film’s success would harm the city’s reputation. There have also been rumors that Chinese authorities might actually ban the film, a kill-the-vibe approach that calls to mind 2024’s short-lived “night cycling to Kaifeng” trend. Some netizens have accused China’s film authorities and critics of hypocrisy. One Xiaohongshu user pointed out the anomalous treatment of state-backed and grass-roots film successes: “‘Ne Zha’ was called ‘a blockbuster,’ but ‘The Cow Comes’ is being called ‘an anomaly.’”

CDT Chinese editors have archived nine articles exploring various aspects of the Niu Lai phenomenon. Only one of the articles appears to have been censored—blogger Xiang Dongliang’s humorous reworking of Wen Yiduo’s famed 1925 poem “Dead Water.” Xiang’s final stanza preserves most of the original, but replaces the word “ugliness” with the titular cow: "In this ditch of stagnant water / No Hope nor Beauty can be found / So why not let the Cow Come at it / And see what kind of World comes out."

An archived article from WeChat account 难得君 (Nándéjūn) discusses negative reactions to the film’s success from Chinese directors and other film industry professionals, some of whom argue that it represents a death-knell for Chinese cinema, a glaring example of “bad money driving out good money” at the box office. The author refutes this argument by giving examples of supposedly “good” cinematic fare that turned out to be tax dodges, Ponzi schemes, critical failures, or box office bombs. One example is director Guan Hu’s 200 million yuan ($30 million USD) "A Man and a Woman," a pandemic-era quarantine drama set in Hong Kong. When it was released in 2026, after two years of delays, it earned a paltry 3.85 million yuan (about $560,000) in ticket sales, prompting one reviewer to call its bloated production and promotion budget “legalized robbery” (合法抢劫, héfǎ qiǎngjié). A portion of Nandejun’s article, “The Success of ‘The Cow Comes’ Isn’t Bad Money Driving Out Good; It’s the Audience’s Silent Protest,” is translated below:

Audiences are flocking to theaters with one clear purpose: to see just how bad this film could possibly be.

“I watched for an hour,” said one filmgoer, “and laughed the whole hour." One theater in [Hebei’s provincial capital of] Shijiazhuang was about 80% full, with laughter echoing throughout the venue.

A viewer in Wuhan said that by the end of the movie, “My mind had gone blank. Is this really the state of animation modeling technology in 2026? Even AI couldn’t dream up such an outlandish plot.”

Another said, “You’ll regret the 86 minutes you spent watching it, but if you don’t watch it, you’ll regret it for the rest of your life.”

Tickets cost 20 yuan ($3). For the price of a fast-food meal, you get to be part of a carnival that’s taken the internet by storm, with the added bonus of an evening filled with laughter. For many people, that’s well worth the price of admission.

Nobody actually cares about the plot. What matters is the sense of participation.

[…] The audience isn’t stupid. When they buy tickets to see "The Cow Comes," there’s likely a subtext to their laughter.

The supreme irony is that—in an industry where big directors, big stars, and big investors pile into "blockbusters" that nobody watches—this little cobbled-together mess of a film is playing to packed houses.

Some claim that "The Cow Comes" is a case of "bad money driving out good." But if "good money" means burning through 200 million yuan only to earn 3.85 million at the box office, shouldn’t we rethink our definition of "good money"?

Film is an art form for the masses. The survival of the industry and everyone in it depends on the public’s willingness to shell out to see their films. But at a time when film-awards bodies treat audiences like fools, when producers treat investors like ATMs, and when disastrously bad amateur animation becomes the most talked-about thing in cinema … it probably isn’t a victory for “The Cow Comes” per se, so much as the audience voicing their profound discontent in the most absurd way possible. [Chinese]

Another archived piece is “‘The Cow Comes’ Is So Bad It’s Good: It No Longer Matters Whether Movies are Good or Not,” from WeChat account 冰川思享号 (Bīngchuān Sīxiǎnghào, recently rebranded as “Iceberg Institute,” formerly translated by CDT as “Glacier Think Tank”). The article compares the viral animated hit to other “so-bad-its-good” films and songs, and discusses what this says about the state of the Chinese film industry and the tastes of Chinese audiences. Later in the article, the author addresses rampant online rumors that Chinese film authorities might pull the movie from theaters, and if so, the “Streisand effect” this will have:

A lot of people are demanding to know how a film this bad could possibly pass muster with the censors. I suspect this question reflects a misunderstanding of how the film censorship review system actually works.

As long as a film’s content doesn’t cross any red lines, and it meets the technical safety standards required for theatrical release, it will generally pass review. As a fairytale coming-of-age tale of a young calf, “The Cow Comes” naturally poses no problem on that front.

So a film can have schlocky visuals, tragic animation, and a ridiculous plot, but none of that stands in the way of it getting the "Dragon Seal" of approval. The minimum threshold for passing review is “whether it can be shown in theaters,” not “whether it’s any good.”

The fact that a film as insanely bad as "The Cow Comes" even made it into theaters is, in itself, insane. But there’s even more insanity in store: if the film gets pulled from theaters on the grounds of "public mockery,” then it will no longer be a movie, but a crowd-sourced work of modern performance art.

The irony is that removing the film from theaters would instantly cede the moral high ground to those who were heretofore just goofing around and making memes. Niche consumers of bad movies would morph into staunch defenders of freedom of expression. The recent "bamboo cicada" incident serves as a cautionary tale. [This refers to Huawei drawing unwanted attention after the telecom titan tried to crack down on a meme-driven wooden toy it felt was mocking its corporate image.]

That’s why the cost of pulling "The Cow Comes" from theaters would be far greater than simply letting it fade away on its own. Allowing it to quietly finish its theatrical run and be forgotten is the safest course of action. Any meddling would hand it a significance it doesn’t deserve. [Chinese]

In a similar vein, lawyer and legal-affairs blogger Yan Senlin talks about online rumors that authorities might curtail screenings of the film, fueled by members of the public who are calling for an outright ban, and the important distinction between “criticism” and “bans”:

I’m not certain whether the rumors about “The Cow Comes” being pulled from theaters are true or not, but I want to reiterate this: As long as the production and approval processes were legal and compliant, then no matter how bad the film is, it has the right to continue being screened. It can be mocked, but it mustn’t be banned.

I would go a step further and say that only under a more relaxed and tolerant institutional framework is it truly possible for “100 varieties of film to bloom,” or for the film industry to achieve genuine prosperity.

One thing I’ve always worried about is people’s tendency to demand government intervention at the drop of a hat, whenever anything displeases them. This phenomenon is particularly distressing in the cultural and publishing fields. Objectively speaking, the creative space in those fields is already quite limited, so if the public goes on making complaints and clamoring for regulation to resolve aesthetic differences of opinion, it will only further curtail that space.

Today, "The Cow Comes” might be banned for being crappy; tomorrow, a film you like might be subsumed by the same logic, banned because it doesn’t suit the aesthetic preferences of some other group. It’s a game in which no one is safe, and no one ever truly wins.

That we so often fall into this confusion may, in some sense, be due to the lack of a stable middle ground in our public discourse. Between singing the praises of something and vilifying it, there ought to be an ample middle ground that allows people to say, “I don’t like it, but I’ll defend its right to exist”

This “buffer zone” is the fertile soil essential for nourishing cultural criticism, and its existence is a sign of a society’s intellectual maturity.

Let criticism be criticism, let the market be the market, and let power, if it does insist on intervening, at least be governed by law. [Chinese]

Lastly, an article from blogger and social commentator Wang Yongzhi, who writes under the pen name Wang Wusi and publishes on the serially reincarnated WeChat account New New New Silence. Wang, who once worked as an animator himself, argues that “The Cow Comes” has a kind of clumsy sincerity, an authenticity that audiences seem to appreciate—especially compared to the often preachy, disingenuous, or inauthentic state-approved films frequently promoted at the box office. Wang characterizes the success of this little-film-that-could as an expression of long-simmering public frustration with Chinese cinema, a “revenge vote” by Chinese filmgoers:

Box office receipts for the film “The Cow Comes” keep climbing. As a veteran animator, I can tell at one glance that this film is bad, glaringly bad. But compared to the multitude of big-budget blockbusters with shiny packaging and vapid plots, its badness is sincere: it isn’t trying to deceive anyone, or masquerade as anything other than what it is. It’s so bad that it’s transcendent: in other words, in a Chinese film industry so packed with bad movies, it is so far ahead of the pack that the descriptive “bad” doesn’t do it justice.

The viral popularity of “The Cow Comes” is not, at its heart, a ringing endorsement of the film itself, but a product of pent-up frustration with Chinese cinema, a "revenge vote" on the part of the moviegoing public. In recent years, Chinese filmmakers have flooded the market with all manner of bad movies, churned out on an industrial scale. There are slapdash scripts, casts crammed with celebrities, and wall-to-wall marketing blitzes, but when we show up to the theater, what we often find are shoddy, phoned-in, straight up crappy films. And to make matters worse, they won’t even let us say they’re crap. “The Cow Comes” is the exact opposite of this: it’s openly, unreservedly, guilelessly bad. It doesn’t try to use dim lighting to hide its flaws, or fast cuts to disguise how low-budget it is: every crude polygon, textural glitch, and clip-art error is on open display, projected onto a twenty-meter-wide screen. You could call it garbage, and the producers wouldn’t even argue with you.

This is who Chinese moviegoers are. No one loves Chinese cinema more than they do. They don’t ask for much, just sincerity. As long as you don’t lie to them, they’re patient enough, and willing to wait for Chinese cinema to “come of age.” But they’ve been lied to so many, many times over the years, and treated so very, very badly. Chinese moviegoers are like Chinese soccer fans or Chinese stock-market investors: arguably the best in the world, yet the object of their love is the worst in the world: the worst directors, the worst actors, the worst scripts, and the worst regulatory authorities. For so long they’ve waited for the chance (a chance like the one offered by “The Cow Comes”) to love Chinese cinema once more, to love it wholeheartedly. “Dear Mama,” they cry, [like the son in the 1988 Taiwanese tearjerker “My Beloved”] “Love me one last time!” before their hopes turn to ash.

[…] These days, most of the relationships we encounter in daily life are extractive, with the underlying logic being “I want something from you.” Platforms want your attention, businesses want your wallet, and social relationships want your “emotional engagement.” Even when you’re watching a piece of content online, there’s a whole algorithmic system behind the scenes, running calculations on you. A genuine relationship in which “I don’t want anything from you, you don’t want anything from me, and we both just feel good about it” has become a rare commodity, indeed. That’s why a phenomenon like “The Cow Comes” feels so precious.

[…] I suspect that another reason people like “The Cow Comes” is that it doesn’t try to lecture its audience or force-feed them values, and it isn’t tasked with the responsibility of “disseminating culture.”

[…] You could call this a peasant uprising, or a win for the underdog. It tells us that the script can be rewritten, and the nonsensical shattered. It reminds us that unreasonable restrictions aren’t set in stone, that privilege can be mocked, and that those “sacred cows” aren’t really so sacrosanct after all. [Chinese]

I’m going to be bold for a minute, and address this. Investigative reporter Li Wei’ao has written about two incidents in the last few months of officials assaulting women at banquets.

There’s already been media reporting on the first case, so there are tracks to follow.

I think they’re still trying to work out the right phrasing on the second case, because I haven’t been able to find an official notice.

Someone online asked why the media isn’t investigating and reporting on it.

Great question. Don’t ask it again. Nowadays, in cases like this, if local authorities haven’t issued an official notice of their own accord, any outlet that tries to report on it will basically just be kicked around like a football. Even central state media will get snapped at: "Which paper are you from? Why don’t you report on something positive?"

So in that second case, the only information that can currently be found on Deputy District Head Yu Tingdong is that his "biography was removed from the official government website," and it’s impossible to ferret out anything about his current position.

Jinan Times put out a piece earlier, but it’s gone now.

There’s not much to say about Deng Fuguo, [the accused] from the first case. I read about it a while ago in a Caixin report, which said a banquet had been arranged, and then dot, dot, dot …

As for what will happen next, whether there will be an official notice and how much information it will contain—all that depends on their whim. What I want to write about today is the six characters Li Wei’ao closed with: "Why does this keep happening?"

I’ve been thinking about the psychological mechanisms of power: what I’d call "expansive power" and "contractive power."

Think of it as a kind of ecosystem in which the subordination of power varies from place to place.

In some places, when a person gains power, it’s "spoils of war." It’s a means of subjugating others: "I’m powerful now; you all have to do as I say." And if you don’t, there’s violence—coercion to obey.

Then there are other places where, when someone gains power, their first thought is trepidation: "Everyone’s watching, it’s unnerving." So the person is cautious and restrained, not wanting to give people anything that could be turned against them. In other words, power is like a hot potato.

In the first case, power is expansive; in the second, it is constrained.

Imagine, you become an official today. Would you dare assault a woman at the banquet table tomorrow?

Ordinary people will think, "Absolutely not." But stay in that role for a while, and it seems less of a big deal, and you might start to entertain the thought. Because you’ve started to expand, and after your daily practice of expansion you feel, when you meet refusal, that this is a challenge. Then you expand the boundaries of your power, showing others, "I can assault you, and there’s nothing you can do about it."

As Montesquieu wrote: "Constant experience shows us that every man invested with power is apt to abuse it, and to carry his authority as far as it will go." [Source]

The point I want to make is that many people denounce this as "lechery." But at their core, these cases have nothing to do with sexual appetite, and if we keep looking at it in those terms, then things will only get worse in the shadows. Each time a case is discovered, it will just be like spotting a single cockroach.

Things like official banquets are actually a theater of power in microcosm. How can it be a simple matter of dining, when even the sequence of toasts reflects an intricate power heirarchy? And some of the people there are just supporting players, or even props.

I think the differences in what stands above power can be categorized into three tiers:

The lowest tier is expansive power. Examples omitted.

Second tier is contractive power. Singapore’s "high public salaries to foster integrity" are one example: they give officials high pay, but with zero-tolerance accountability. A corrupt minister isn’t just risking his pension: absolute disgrace is at stake.

It’s a bit like loan collateral. Officials put their property, dignity, freedom etc. up as collateral: there are no "spoils of conquest." So even a box of chocolates received as a gift has to be reported, and they’re constantly on edge, because the cost of holding power is high.

The top tier is flattened power. I think this is the best kind, but even some of the most established developed countries can’t pull it off. Only a few small countries in northern Europe have hints of it, with "power" stripped of its halo. For example, prime ministers have no chauffered vehicle or security detail, but regularly commute to work by bike. Power doesn’t contract or expand—it’s been "disenchanted."

But this is the most difficult, because it doesn’t just rely on institutions: these must be matched with moral self-discipline. As we all know, that kind of moral character is an aberration.

So that’s what I think. Don’t curse lechery. When power expands into monstrous forms at the banquet table, refusal is interpreted as "provocation," and oversight must always wait for the official notice, then the problem isn’t someone’s inability to control his lower half, it’s our fundamental lack of an iron cage strong enough to keep power contained. So it expanded. [Chinese]

While playing the popular role of “mad monk” Ji Gong in a theatrical performance in Wuhan earlier this month, Guo improvised some new lyrics to “Playing My Beloved Pipa,” a tune best known for its inclusion as a musical interlude in the 1956 revolutionary film “Railway Guerrilla” (Chinese title: 铁道游击队, Tiědào Yóujīduì). Guo changed the line "All is quiet on Weishan Lake" to "All is quiet in the Forbidden City," and played the scene for laughs, adding a bit of gibberish (妈咪妈咪哄, māmi māmi hōng) that is a comedic mangling of the Buddhist mantra “Om mani padme hum.”

Following a complaint from an unknown member of the public who accused Guo of “distorting” and “vulgarizing” a revolutionary classic, the Wuhan Municipal Bureau of Culture and Tourism announced that it was investigating Guo and the Qilin Opera Society, a group he cofounded, for violating regulations by performing content that had not been submitted for prior approval. Guo, arguably China’s most famous comedian, has been accused of overstepping red lines before: in late 2025, he was reportedly called in for a scolding by the Beijing Xicheng District Bureau of Culture and Tourism for his crosstalk routine "Great Skill, Little Courage," which made unflattering allusions to state-affiliated theatrical troupes.

The online reaction to this latest investigation into Guo Degang was withering. Many Chinese netizens characterized the investigation as a politically motivated overreaction, a harbinger of China’s shrinking cultural space, and a throwback to Cultural Revolution-era persecutions of performers, writers, musicians, and other creators. One commenter repeated a quote attributed to Guo to describe the political pressure and betrayal he experienced during a 2010 “anti-vulgarity” campaign: “Those who wrong you understand even better than you just how wronged you are.” Another alluded to a joke that got comedian House (Li Haoshi) banned back in 2023.

CDT editors have archived eleven articles, five of which were later censored, about the controversy over Guo’s performance and subsequent investigation, as well as a compilation of comments from Weibo and X criticizing the investigation into Guo Degang, a selection of which are translated below:

你永远不懂的流年: What era are we in, anyway? They’re being way too sensitive.

KingC天: He “distorted” the lyrics? That’s a bit much.

严小猥c: Sure, let’s stick to model operas from now on, you can’t go wrong with those.

自来火wx: It was just some random song in a movie. Elevating it to a major political offense is pathetic.

王胖子大明白: The entertainment space keeps shrinking.

毛豆蔻800: “Those who wrong you understand even better than you just how wronged you are.”

村口的蓝猫: “The Forbidden City is silent” — isn’t that true? The Qing was China’s last feudal dynasty.

元帅V587: We’re still doing this sort of persecution?

小花大咖秀: Hard to believe this is happening in 2026 …

ge0sn0w: When the body politic is all festering sores, wherever you poke it, it’s going to hurt.

qSMwoZXfW1y64WW: From now on, the CCTV Spring Festival Gala can just be 20 different performers singing the same national anthem.

HLCtoC: This outdoes the Qing Dynasty’s literary inquisition.

cloud_bingo: The Cultural Revolution never ended.

Ktan4A: Two dogs on a leash: one’s named “Fight to Win” and the other’s “Exemplary Conduct.” [A reference to a joke riffing on two PLA slogans that got comedian Li Haoshi banned back in 2023.]

XPD_Schenn83: The moment he sang those words “Forbidden City,” you knew exactly who was going to get pissed off behind the scenes. [Chinese]

One of the five deleted pieces in CDT’s archive about Guo’s latest tussle with cultural censors comes from WeChat account I Come From the Next Life. In “Might I Meekly Say a Few Words in Guo Degang’s Defense?" author Fan Dang describes the chilling effect that such comedy crackdowns have on all forms of self-expression. The author’s intentional use of trepidatious language and textual attempts to dodge platform censorship (for example, writing "sensitive word" as "敏g词," rather than the typicall "敏感词," mǐngǎn cí) vividly illustrate how exacting and exhausting self-censorship has become. A portion of Fan’s essay is translated below:

I’m honestly not trying to make excuses for Guo Degang, or to show any disrespect toward our history or the forebears represented in these "red" works. I just … can’t help feeling that if a comedian cracking some harmless little joke onstage can escalate into a formal investigation, doesn’t that mean that the space for all of us ordinary people—the space to talk, chat, or joke around—is just going to keep getting narrower, too?

Nowadays before I post anything to my WeChat Moments, I have to draft it, read and reread it, delete any word that could possibly be misinterpreted, and even agonize over which stickers or emojis I use, just in case they convey some unintended meaning. After all that, and only after running it through a "sensitive-word" checking tool, do I finally dare to hit “send.” When I’m out eating with friends and the conversation touches on anything even slightly sensitive, I find myself glancing around and lowering my voice, like I’m some underground Party operative meeting a secret contact.

I always think that when it comes to culture, there’s got to be some room for harmless joking around, otherwise the bow is strung too tightly—ready to snap at the slightest touch.

By this point, I’ve already gone back and revised this piece at least seven times. I’ve double-checked every sentence, changed the punctuation more than a few times, and even closed and reopened the document on three separate occasions, out of fear that some line might come out wrong. If there’s anything inappropriate, I’ll fix it right away, even delete the whole thing if I have to. I’m about as ordinary a person as they come, and today it took me nearly two hours to work up the nerve to finally hit "send" and timidly speak a few honest words I’d been sitting on for a long time. That’s all I meant to say, truly.

Truly. [Chinese]

Nevertheless, the notice that has replaced it online says that "this content cannot be viewed due to rule violations. Following relevant complaints, it was found to violate the Cybersecurity Law of the People’s Republic of China." Automated keyword scanning may also have played a role: see other recent translations for more on collateral damage from this approach, and efforts by users to evade it.

Now that Zhengyangmen is open to the public, I felt compelled to make the obligatory visit to Tiananmen Square again.

I was already a bit apprehensive about the security checks after my parents’ trip to Beijing last year, when they’d decided to visit Mao’s mausoleum.

This time happened to coincide with the peak of the summer break. The heat was stifling, and it seemed like things would be even worse than before.

My goal, Zhengyangmen, is by Exit A of Qianmen Station on Line 2, but unfortunately, as I only realised later when I was trying to leave, that’s a one-way exit.

So after leaving through Exit C, I couldn’t just go straight across the road; I had to cross twice, through two separate underpasses.

This involves one ID-card swipe, and a security check.

They had to look through my bag at the security check, so I quickly got my things out. That day, that meant a book, a notebook for collecting tourist stamps and, because it had been raining a lot, an umbrella. That was about it.

I was extremely cooperative, hoping they’d finish up and let me through as quickly as possible, and of course that’s what happened.

I could see Zhengyangmen ahead of me on the right and, wanting to get inside quickly, I rushed forward without thinking.

The security checker must have sensed my impatience, because they drawled said, "No point in hurrying, there’s more security up ahead."

With hindsight, that first security check was just a canapé before the banquet.

As I crossed the intersection of Qianmen West and East Avenues and turned north onto West Guangchang Road, the crowds became denser.

At this point, Zhengyangmen was directly to my east. I was about to start lining up in that direction when I was courteously informed that that was the line for the group security check, and I needed to keep going.

So I kept heading north. The crowd had slowed to a crawl, and started to divide again into streams: those with bags, and without.

Those without bags moved on a bit quicker, while those of us with bags continued to wait.

By now, the sun was beating down on us mercilessly, and a lot of people were using parasols. Some were chatting about other places they’d visited over the past few days.

The Numismatic Museum was to the left, the building a fairly impressive sight from that angle. To the right was the mausoleum, looking solemn and dignified as always.

By then, we’d been kept waiting so long and had grown so impatient that it was hard to appreciate the sights.

The kids in the crowd started to get rowdy, and were variously consoled or scolded by the adults, who were suffering every bit as much as the little ones.

Kids are just more straightforward about voicing their annoyance.

The whole security line had stopped, and we had no idea what was going on up ahead.

I didn’t realize how stringent the security checks were until I actually experienced it myself. It’s even stricter than at the airport.

There are two stages to the personal security check. In the first, they pass a handheld sensor wand over your whole body and pat down your pockets, so I had to take off my sun jacket and also remove my keys, cellphone, etc. from my pockets.

I even had to take my phone out of its case so the inspector could take a look.

When you get to the second stage, an inspector of the same gender as you pats down your whole body wearing disposable gloves, and lifts the hem of your top to check for concealed items around your waist.

They poked and prodded at my shoes for a while, but fortunately they didn’t make me take them off.

Once the personal check was over, it was time for the bag inspection, and you had to take out all the contents. They looked through every page in my book, but didn’t say anything in the end.

The person inspecting the group next to me told them they couldn’t take photos with a book and certain other item—I don’t know what kind of book it was.

One reason things moved slowly was that the police were involved through the whole process: if one of the staff wasn’t sure about something, they’d call the police over, which led to more waiting.

In short, after layer after layer of checks, I got through and left the guardhouse. When I looked round, I realized there were very few others coming out, and an endless line of people still waiting.

I don’t know whether I just turned up at the busiest time, but I’d avoid wandering onto the Square without a particular reason, especially in summer. If you must go, stow your bags elsewhere beforehand.

Having said that, you should travel while you can. I miss the old days when you didn’t have to book in advance. [Chinese]

In contrast with this post’s deletion, another recently archived article takes a much more critical stance on security checks, but currently remains online. At the WeChat account 白马狗熊 Bái Mǎ Gǒuxióng ("White Horse Black Bear"), Song Jinbo notes tightening security checks at subway stations in many Chinese cities, and highlights claims that they are actually detrimental to public safety, causing potentially dangerous bottlenecks. Moreover, he expresses concern that well over 100,000 young Chinese may be wasting key years in their early 20s performing labor of dubious value, pointedly contrasting this with Marx’s calls for people’s "all-round development" and Xi’s calls for the young to "cherish the springtime of their lives" and strive for self-improvement in the national interest. Song posits that the checks endure because of financial interests in what has become a multi-billion-yuan industry, and because officials are reluctant to ever remove security measures in case subsequent events should make that look reckless, wrecking their careers. Meanwhile, taxpayers who objected to the cost, inconvenience, and ineffectiveness of the security checks would have no way to exert effective pressure against them. Song opens the piece with a 2024 case in which 14 fraud victims reportedly worked in non-existent outsourced roles on the Shanghai Metro for months, while their unauthorized presence went unnoticed. "A system that can’t even tell fake workers apart from real ones requires you every day to put your water bottle through security and your tote bag through a scanner. Consummate artistry."

Wang Hong (王虹, Wáng Hóng), age 35, was born and raised in Guilin, Guangxi province. After graduating from Peking University, she earned advanced degrees in engineering (École Polytechnique) and mathematics (Paris-Sud) in France, and her doctorate in mathematics from MIT. Now a professor at NYU’s Courant Institute and the Institut des Hautes Études Scientifiques, Wang was awarded the Fields Medal for her contributions to harmonic analysis and geometric measure theory, and major advances on a number of mathematical conundrums, including her work on the three-dimensional “Kakeya conjecture” with collaborator Joshua Zahl (University of British Columbia). Wang is the first Chinese woman—and only the third woman overall—to win in the award’s 90-year history.

Deng Yu (邓煜, Dèng Yù), 37, was born in Kaifeng, Henan province, and raised in Shenzhen, Guangdong province. After two years at Peking University, he transferred to MIT for his undergraduate work, and later earned a doctorate in mathematics from Princeton. Currently a professor at the University of Chicago, Deng received the Fields Medal for his work in partial differential equations and his breakthrough on "Hilbert’s sixth problem," with University of Michigan collaborators Zaher Hani and Ma Xiao (Xiao Ma, in English).

CDT editors have archived over a dozen articles on various topics related to the Fields Medal winners. One popular topic concerned how much credit is due to Peking University (PKU), which proudly trumpeted the news of alums Wang and Deng winning the prestigious prize by putting up festive banners and projecting congratulatory messages onto the campus’ famed Boya Pagoda. PKU’s official WeChat account also republished an interview with Deng Yu, conducted by the Beijing International Center for Mathematical Research (BICMR), but internet users quickly noticed that PKU’s version of the interview had been selectively edited, with numerous alterations and a question about overseas study deleted entirely. CDT Chinese editors compiled a side-by-side comparison of the original interview and PKU’s selectively edited text.

Congratulatory messages projected onto Peking University’s famed Boya Pagoda read: “Congratulations to Wang Hong and Deng Yu!” and “Congratulations to Chinese mathematics!”

Perhaps the most viral topic was a video interview in which Wang confided that she “felt a little discouraged” during her time at Peking University. The English word “discouraged” soon became a trending hashtag on Weibo, but was later censored, although some related content referencing Wang’s use of “discouraged” could still be viewed. Many Chinese social media users shared their own experiences of feeling thwarted by what they described as an overly hierarchical and inflexible academic system, and debated whether the two Fields Medal winners would have chalked up the same achievements had they remained in China. Wang’s comment clearly resonated with many Chinese netizens, and CDT has archived several articles on the subject. A piece from WeChat account Her Narrative, “Thank You, Wang Hong, for Saying the Word ‘Discouraged,’” describes Wang’s early frustrations and circuitous academic path, highlights the importance of her graduate studies in France and the U.S. and finding her “true mentors,” and encourages readers to live life at their own pace and not to lose confidence. In “Why We All Understood What Wang Hong Meant by ‘Discouraged,’” from WeChat account Plucking a Bashful Moon, the author discusses growing up in China and having their shortcomings frequently pointed out by teachers and parents, and contrasts this with what they view as a more supportive and positive culture of encouragement in Italy, where the author has spent some time.

In a video interview, Fields Medalist Wang Hong described her time studying Earth Sciences at Peking University, “I felt a little discouraged.”

In connection with this discussion of “educational trauma,” some suggested using the term “school of origin” (原生学校, yuánshēng xuéxiào) to refer to PKU’s complex relationship with alumni such as Wang and Deng. “School of origin” derives from "family of origin" (原生家庭, yuánshēng jiātíng), a term used in psychology to refer to the family—and family trauma—one was born into. One Chinese-language comment on Threads that received over 5,000 likes said that in the case of Wang and Deng, “PKU has played the role of their ‘school of origin.’” Other users agreed, with one humorously complaining, “Look at my family of origin, school of origin, and country of origin … I’m totally screwed.”

In an archived piece from WeChat account History Rhymes, “Wang Hong and Deng Yu Won the Fields Medal, But How Much Credit Does Peking University Really Deserve?” current affairs blogger Xu Peng discusses this question in light of the academic histories of the two mathematicians, both of whom attended PKU but whose advanced degrees and subsequent academic achievements were earned abroad. In one widely-shared quote, Xu likens their relationship with their Chinese alma mater to that between Yang Guo, the fictional protagonist of Jin Yong’s wuxia novel “The Return of the Condor Heroes,” and the Quanzhen martial arts sect. In Xu’s example, the Quanzhen Sect served as Yang Guo’s “school of origin,” the place where he spent an unhappy and unproductive tutelage under a master who resented him. A short excerpt from Xu’s article follows:

The recent news that two Chinese mathematicians have won the Fields Medal has sparked a frenzy online and given our nation’s “triumphalists” yet another reason to cheer.

Many say that the awards won by Wang Hong and Deng Yu, both Chinese, are a milestone for China. Chinese media has also emphasized that Wang Hong and Deng Yu are both alumni of Peking University.

[…] But how much did their success really have to do with Peking University? There’s certainly some connection there, but the university likely played a minor role.

You could call it the first step on their mathematical path. Without that first step, they might not have reached the pinnacle of their profession.

It feels a bit like Yang Guo’s relationship with the Quanzhen Sect.

If they hadn’t gone abroad, and had conducted their mathematical research in China, would they still have won the Fields Medal?

[…] They are the first two Chinese nationals to win that prize, and this goes to prove that Chinese people are capable of such an achievement.

But as the saying goes: "Orange trees grown south of the River Huai yield sweet fruit, while those north of the river bear bitter fruit.” Even the best seedlings require good soil if they’re to produce a bountiful harvest. [Chinese]

CHINA – PINGYAO, by Pierre Doyen (CC BY-NC-ND 4.0)

Many people have had similar experiences with social media, comments sections, or livestreams: a perfectly normal post or comment expressing one’s feelings or discussing social realities is flagged for violations, blocked, and simply muted. Many quite neutral and ordinary words and expressions inexplicably trip the filters; to avoid this, people are forced to use homophones, pinyin initials, and coded terms in place of the original, straightforward Chinese, making it difficult to just say what you mean.

Many people simply interpret this phenomenon as a policy-level prohibition on online discussion, but that’s not what’s really going on. National laws and regulations determine the bottom line of online speech, prohibiting harmful and illegal content like rumors and libel, inciting antagonism, violence and pornography, or content that infringes others’ rights. But they don’t lay out a public "Directory of Blocked Terms," still less require platforms to restrict ordinary social commentary or reasonable expressions of public sentiment. The large-scale blocking and filtering we see arises from a multiplicity of other factors and pressures: put simply, it’s about negligence.

The foremost pressure in practice comes from platforms’ primary responsibility. According to the current regulations on internet governance, internet platforms must take responsibility for content on their own sites, and will face penalties if they allow widespread dissemination of illegal or harmful content. Confronted by hundreds of millions of users and a vast quantity of messages, bullet-screen comments, and post submissions per second, platforms cannot possibly subject every piece of content to individual human review. Automated keyword filtering has become the cheapest and most efficient first line of defense.

Automated moderation has an inherent shortcoming. Its algorithms can only recognize characters; it’s very difficult for it to understand context or tone. A given word might be an objective statement in an article denouncing violence, but might break the rules in the context of malicious incitement, and it’s very hard for machines to tell the two apart. To avoid missing hazardous content, many platforms opt for a conservative strategy of "erring on the side of caution" by constantly expanding their lists of banned words. Neutral vocabulary and objective descriptions of social phenomena can be caught in the crossfire—a form of filtering that people often complain is "indiscriminate."

Filtering like this in self-defense can also lead to a vicious cycle. When some words are blocked, netizens might create homophones, pinyin initials, or coded terms to avoid detection by the system. To plug these leaks, platforms in turn constantly expand the blocked word lists to intercept those substitute terms. So the word lists keep growing, incorporating more and more ordinary vocabulary, distorting the whole ecosystem of online speech. Content that should be expressable directly must take circuitous routes, a whole set of online argot arises, and communication becomes ever more costly.

This kind of automation has many visible side-effects. Ordinary people sharing daily pressures or discussing social realities obviously have no malicious intent, but their content is intercepted and shot down. Creators documenting events or writing about science have to revise their wording again and again, worrying over every detail. Many people gradually start to self-censor, deleting parts before posting, for fear of carelessly crossing some unknown line. People’s attention shifts away from viewpoints themselves as they expend energy wondering which words might trigger the system, and the oxygen of public debate is depleted as a result.

Still more worthy of attention is the fact that excessive keyword blocking cannot truly resolve online chaos. A great deal of genuine cyberbullying, snide insinuation, and malicious trolling doesn’t use explicitly prohibited language: it relies on euphemism and innuendo, against which simple keyword filtering is ineffective. Automated screening works for written symbols, but not shifting language, and it’s often reasonable everyday users who end up paying the price, while malicious expression adopts new forms and continues to spread.

Of course, we can’t completely dismiss the value of content filtering. The internet is not a lawless zone, and abusive, bullying, rumormongering, or harm-inciting content really does need restriction. This is a necessary means of safeguarding the online public environment. The problem doesn’t lie in "necessary moderation," but in the way moderation is carried out: whether it’s pursued with crude and simple keyword interception, or by a more refined approach that considers context and distinguishes intent.

Regulators have long recognized that "indiscriminate moderation" is problematic, and platforms have repeatedly been asked to refine their moderation mechanisms and to avoid simple reliance on keyword bans, to distinguish between illegal content and citizens’ normal expression, and to improve channels for appeal and review. But in practice, conflicts between technological change, platform costs, and content volume mean that this cannot be achieved overnight.

The online space needs to uphold security standards while also accomodating the authentic voices of ordinary people. An ideal online environment should not force people to speak in riddles. The goal of governance should be to block malice and actual harm, not stifle normal expression. Only with human review as a safety net behind automated initial screening, a well-developed appeals process for misjudgments, and appropriate transparency about the boundaries of the rules, can the amount of senseless collateral damage be reduced, allowing everyone to speak openly, rationally discuss reality, and express their true feelings. [Chinese]

Caixin Weekly reports that Lanzhou migrant worker Jing Shuren has been detained for more than two years now for sharing other people’s Weibo posts about rights defense and unpaid wages. The relevant departments’ ultimate charge against him was “picking quarrels and provoking trouble.”

The report stated that Jing, a 49-year-old with a junior high school education, was working as a gate guard at the time of his arrest. But as a kind-hearted person, he would frequently repost information about the plight and rights of disadvantaged groups on his personal Sina Weibo account. He posted very little about his own life.

Jing Shuren

I knew of this man. Before his misfortune, he would occasionally comment on my articles or message me on WeChat.

I’ve never tended to communicate a lot with readers. I was a reporter for 16 or 17 years back in the day, and was accustomed to receiving news tips from readers. If there was anything in it, I was happy to reply, but if it was just subjective views and positions, it wasn’t worth much to me, and I’d tend not to bother. As someone covering news stories, I didn’t need readers’ instructions on what attitude or stance to take in my writing; I needed facts and source material that could stand up to scrutiny.

So despite Jing Shuren’s occasional article comments or WeChat messages, I had very little contact with him. We weren’t at all close, and I can’t really say I knew him. Because of that, and because I didn’t know details of his case, I haven’t spoken out about his arrest since it happened. Now that a media institution like Caixin has reported on it, though, I know a bit more about the man, and I feel I should say a few words about him.

Jing Shuren is someone worth speaking up for.

According to Caixin’s report, he shared many posts by other people about unpaid wages or rights defense, and never received a penny in return. There’s no indication in court documents that he received money, and his name doesn’t appear in the relevant departments’ records of illicit payments.

In other words, Jing Shuren’s allegedly quarrelsome and trouble-provoking reposts were motivated only by the desire to help others.

The Caixin report includes this rather moving passage:

A friend once pleaded with him: “You’re a migrant worker trying to eke out a living on construction sites, but every day you’re supervising this or criticizing that … one day they’ll put you away.” [According to the earlier blog post from the time of Jing’s arrest, sociologist Yu Jianrong reported having this conversation with Jing.]

Jing Shuren replied: “What kind of society would we have if no one spoke up?”

The friend said: “What will become of your elderly father and little child if something happens to you?”

After a long pause, Jing Shuren said: "I still want to do something to support rule of law in this country."

According to Zhou Xiaoyun, a lawyer who has been in contact with Jing Shuren, Jing was working as a gate guard before his arrest, making two to three thousand yuan per month, but only keeping 200 yuan for himself, and sending the rest back to his family.

The Caixin report goes on to say that in 2018, a college student from Jing Shuren’s hometown was hoodwinked into a multi-level marketing scheme in Cangzhou, Hebei. When he learned of this, Jing rushed to Hebei at his own expense and extricated the student with the help of local police. On March 4, 2024, while on his way home, Jing noticed that a water pipe had burst near Hanjiashan on provincial route 209 in Gansu’s Longxi county. He immediately called the local water authority to report it, and ended up receiving a text message of thanks from them.

Jing Shuren: “Good day, sir. This flooded road section is on the Dingxi–Longxi highway (Provincial Highway 209), 700 metres past the 235km mark on the uphill stretch toward Dingxi. It’s at the bottom of Hanjia Hill, Daijiamen hamlet, Shangdadao village, Yuntian township, Longxi county. Thank you, sir!”
Response: “Please send an address, and we will arrange a site visit to take care of it. Thank you for your attention to the public highways!”

“‘Safety is the shortest route home,’ I’m not just being a busybody!”
“You’re most conscientious, brother, paying such close attention to the people’s welfare. Respect! We could all learn from you.”
“Thank you for your attention, big brother. Not one snowflake is innocent in an avalanche. Nothing is too minor when it comes to safety, I’m not just being a busybody.”

I dug these up in Jing Shuren’s WeChat Moments—it’s just as the Caixin report described.

When people online praised Jing’s sense of duty, he said: "Not one snowflake is innocent in an avalanche. Nothing is too minor when it comes to safety, I’m not just being a busybody."

That’s the kind of civic-minded person Jing Shuren is.

In life, a lot of people grow cold toward society and those around them because of their own hardships, and many people much better off than Jing Shuren become apathetic and disengaged, never standing up and speaking out for others or lending a hand to those in trouble. Not Jing Shuren. Despite his own hardships and privations, he has never lost hope or given up, and is always concerned for societal progress and the wellbeing of others.

You can really call him public-spirited, which can’t be said for many people these days.

Lanzhou Public Security arrested him for helping others by sharing their posts about unpaid wages and rights defense. More than two years of investigation have shown that he never received a penny in return, which shows the truth of his words: that he really did step forward and share those posts not for his own profit, but simply because he wanted to do his bit to improve society.

Jing brings to mind the common expression, "Honor is found among the lowliest butchers [and perfidy among the loftiest scholars]."

As things now stand, Jing Shuren neither aimed to disturb social order nor did he actually do so; he simply wanted to help people who were owed unpaid wages. He was a gate guard with only a junior high school-level of education, not a professionally trained reporter, and could not possibly conduct a thorough investigation of every post. He bore no malice toward society, and only wanted to help those even more vulnerable than himself by sharing some rights defense posts. He shouldn’t be unduly criticized.

The crux of the matter is that in many of the unpaid wages cases that Jing Shuren helped to publicize, the party in arrears promptly paid up once the matter came to public attention. In exceptions, when the claims were inaccurate, the other party just didn’t engage. It’s clear that many of the unpaid-wage cases he reposted were genuine and not fabricated, and that Jing’s activity did not bring about any actual social harm.

It’s impossible to make out where or to what extent Jing erred with his rights-defense reposts.

There aren’t many people as public-spirited as Jing Shuren these days. He must not be accused of picking quarrels and provoking trouble just because he might have shared some posts that weren’t completely true; still less should he be sentenced for it. Even if, among all those rights-defense posts, he shared a few that were not entirely accurate, he didn’t do this deliberately. The law cannot label someone as public-spirited as Jing Shuren a criminal.

To Gansu’s judicial departments: Please release Jing Shuren in accordance with the law, and show some warmth and kindness to this man who has always upheld the cause of justice. Someone as public-spirited as he is deserves to be treasured by society.

July 24, 2026

n.b. As this post involves social justice, I have deactivated tipping and ads. I rarely come straight out and ask readers to repost my own content, but this is an exceptional case, and I hope everyone will share it as widely as possible. [Chinese]

A recent article from Guo Yujie, a reporter at the China Youth Daily, adds an absurd new chapter to this ongoing tale: hoping to establish his bona fides and inspire the trust of a potential interview subject, Guo sends the man a photo of the inside page of his press card, a passport-like booklet with a red or blue cover that is issued to journalists in China. He is astonished to receive a clearly AI-generated response, confidently alleging that the credential is a fake. Guo’s piece, “AI Has Determined That My Press Card Is a Fake,” was published by the WeChat account Freezing Point Weekly, a weekly in-depth column from China Youth Daily. A full translation appears below.

A while back, an interview subject contacted me via social media to say he was willing to be interviewed. We promptly scheduled a phone call for later that day, and to earn his trust, I sent him a photo of the inside page of my press card.

A few seconds later, he sent me a lengthy and detailed reply (roughly 500 characters, with at least five bullet points) expressing the unshakable conviction that my press credential was a fake.

Among the evidence he presented: the date of issuance was wrong; the color was wrong, and didn’t comply with regulations; the inside-page layout was sloppy and disorganized, and lacked an embossed seal; and a string of numbers appeared in the wrong place.

His citations of various regulations and official documents were impressively authoritative. By comparison, my own knowledge of press credentials appeared to be wanting. I was taken aback, and felt the need to explain myself.

The man’s message stated that "the new-version press cards should be blue," whereas the inside page of mine in the photo I’d sent was white. “White press-cards fell out of use a long time ago," he told me. Exasperated, I noted that during the four years I’d worked as a journalist, my red-cover press card had indeed been updated to one with a blue cover, but the inside page has always been white. I dug up my old booklet with the red cover and sent him a photo of that, plus a photo of my current booklet with the blue cover. I was surprised when he replied a moment later that both were obvious fakes: a reporter is only ever issued one press-card number, and my old and new cards had different numbers.

The photos I sent must have looked convincing (there was a computer and some clutter in the background), so the guy didn’t go so far as to accuse me personally of being a fraud. Instead he said, "Well, looks like someone scammed you," punctuated with a "crying facepalm" emoji.

As a frequent user of LLM tools like Doubao myself, I could tell his evidence had probably come from an AI. “That’s something Doubao told you, right?” I asked. “I don’t think AI has reached the stage where it’s capable of getting everything right.”

When I dug into it later, I found out that journalists are indeed issued new numbers when their press cards are updated. The number only stays the same if you’re replacing a lost press card. But when I tried to explain this, I was too slow on the draw; I couldn’t keep up with the pace of the AI’s rebuttals.

More crucially, I’m not nearly as "confident" as AI is. There’s a well-known axiom in journalism and media: “A lie can travel halfway around the world while the truth is putting on its shoes.” It seems like if you want to win an argument nowadays, you’d better be able to “put your shoes on” within seconds.

But I suspected that no matter what I said to him, the guy was never going to trust me again. The seed of doubt had been planted, and by now we were both getting a bit frustrated. I gave up trying to refute his barrage of AI-generated arguments. “It’s fine,” I said. “If you don’t want to trust me, that’s your right. You don’t have to do the interview.”

To which the would-be interview subject replied, “I don’t want to keep discussing this press-card thing with you. Forget the interview, and I don’t even care to know what you were really up to. Bye."

I tried sending a photo of my press card to my own Doubao account and asking whether it was real. Sure enough, this was the response: “This press card isn’t real. It’s a forgery.” The analysis that followed used phrases like “numerous fatal flaws” and “contradictory key information.” If it hadn’t been my own press card, I might actually have believed it.

It even included this thoughtful addendum: “Forging or altering a press card is illegal, and violations can be reported to the National Press and Publication Administration (NPPA), or to the local press and publication authorities. Would you like me to put together a complete guide for verifying the authenticity of a press card, including the official verification portal, QR-code scanning tips, and how to report a violation?”

For a moment I genuinely didn’t know how to respond. Then I pushed back: “What if you’re wrong about this?”

A second later, it said, “First, let me sincerely apologize for my previous incorrect assessment.” It then proceeded, in the space of over a thousand characters, to completely overturn its earlier arguments. The new verdict: “It is fully compliant with official regulations.” Doubao’s previous finding that there was "no record on file" for my press card was now deemed to be "completely wrong."

Unfortunately, when I tried to send screenshots of those two responses to the would-be interview subject, I found that he’d already deleted me on WeChat.

AI hallucinations are nothing new, but this was the first time an AI’s error in judgment had a real, negative impact on my work. After I posted about the experience on social media, some of my journalist friends consulted AI chatbots about their own press cards, and some of them got the same "it’s a fake" result.

Not long afterward, I read an article about Li Xiaoliang, a practicing lawyer in Nanjing, Jiangsu province, who discovered that when searching for his name and profession on the Baidu app or website, Baidu’s "AI Smart Answer" feature returned the false claim that "lawyer Li Xiaoliang was sentenced to three years in prison," accompanied by a photo of Li wearing his lawyer’s robes. [Li eventually won his defamation case against Baidu.]

Since the advent of social media, there has been a constant emphasis on educating people in media literacy, and encouraging them to think independently and judge information for themselves. Now, I fear, AI has made the task of discernment and critical thinking even more difficult. An interdisciplinary study led by Shanghai Jiao Tong University found that nearly 70% of Chinese survey respondents had a “low awareness” of AI hallucinations in large language models, while fewer than 10% had a “high awareness.”

Given the underlying logic by which some large-language models operate, the LLM may appear to be thinking, but in fact it is simply predicting “the next most likely word.” As long as the language flows smoothly and suits the context, the AI will generate content, even if it’s fabricated. What’s more, an AI’s training data can’t possibly contain everything. When it can’t find credible or established information, it may end up fabricating or inferring an answer.

An AI’s response also depends on how the query is phrased, but we can’t expect everyone to phrase their queries the way a professional might: "You will take on the role of an objective, neutral credential-verification specialist. Your task is not to subjectively judge the authenticity of this credential, but to help me analyze the various features of the credential based on the images I have provided and other verifiable public information."

Some say we are entering an era of “learning to live with AI hallucinations.” Many believe that, given the underlying technology, AI hallucinations cannot simply be eliminated.

I have no idea how long this era will last, or what changes lie ahead. As an ordinary user, what I’m curious about right now is this: when faced with a question it isn’t entirely sure about, why can’t an AI just say it doesn’t know, instead of always presenting its inferences with such certainty?

A 2025 paper published on arXiv suggests this may be related to how LLMs are trained. Just as humans may guess at multiple-choice answers on an exam, during testing and training, an AI is more likely to score points by making up an answer than by abstaining or expressing uncertainty. Under this kind of binary rubric (“1 point for each correct answer, 0 points for each blank or ‘I don’t know’”), the AI is incentivized to conceal its own uncertainty.

“Humans learn the value of expressing uncertainty outside of school, in the school of hard knocks. On the other hand, language models are primarily evaluated using exams that penalize uncertainty. Therefore, they are always in ‘test-taking’ mode,” the paper states, noting that this kind of binary rubric remains the mainstream standard in the field. In 2026, this research team’s findings were published in Nature, and they called for changing the scoring system [to “open rubric”] to better incentivize LLMs to admit their own shortcomings.

I’m still puzzled by one thing: why, after initially declaring my press card a fake, did the AI do an about-face the moment I pushed back?

A separate study published in July 2025 offers an explanation for this seemingly contradictory trait in AI: its tendency to swing from overconfidence to excessive self-doubt. Among the study’s findings was that the LLMs tested were “overly sensitive to contradiction.” Their confidence wavered more sharply in response to contradiction than agreement, a clear deviation from optimal Bayesian inference (the benchmark for rational decision-making). When users directly challenged the models, they wavered far more than even a mature, rational human would, deviating from the theoretically optimal decision-making path.

It appears that, at least for now, AI is not necessarily more “rational” than human beings are.

Kant urged, “Have the courage to use your own reason.” The opposite of courage isn’t cowardice, but passive dependence—dependence on authority, on the mainstream, on tools.

In an age when AI frequently “makes judgments” on our behalf, perhaps all of us as individuals should reiterate the following: have the courage to use your own reason, and never stop training and honing it. [Chinese]

Xanadu, by Xiquinho Silva (CC BY 2.0)

Below is a list of key topics and some related deleted articles from CDT’s summary of deleted content for June 2026. Between June 1-30, CDT Chinese added 30 new articles, primarily from WeChat, to the archive. (Note that the dates refer to when an article was published on the CDT website, not when it was deleted from Chinese social-media platforms.) Topics targeted for deletion in June included:

• The usual cross-platform tightening of search censorship and sensitive-word filtering around the June Fourth anniversary
• An animal torture case in Chongqing that sparked online and offline activism, protests, and arrests
• A new report that estimates that in 2025, China had 280 million “flexibly employed” workers, a number predicted to rise to 320 million by 2026
• An investigation into refrigerated cargo trucks being rerouted to Henan province in an extortionate scheme known as “siphon-style enforcement"
• The World Cup and the state of Chinese soccer

The usual cross-platform tightening of search censorship and sensitive-word filtering around the June Fourth anniversary

Ahead of the 37th anniversary of the June Fourth crackdown, CDT editors compiled and translated a number of related terms subject to search censorship across various Chinese online platforms. The terms, detected by a Citizen Lab tracking system, are just a few examples of the long-evolving efforts to "clear the online square" each year. Offline commemorations are subject to escalating repression as well: this year, there was an unprecedented ban on the Tiananmen Mothers’ annual cemetery visits.

An animal torture case in Chongqing that sparked online and offline activism, protests, and arrests

This summer, two high-profile incidents of animal abuse—in Chongqing and Guangdong, respectively—resulted in an outpouring of social-media comments and op-eds, petitions, protests, and billboards, as well as increased momentum for a nationwide animal-protection law. (China has various laws governing the treatment of livestock, lab animals, and endangered wildlife, but there is currently no law prohibiting the mistreatment of cats, dogs, or other companion animals in domestic settings, despite a decade and a half of NPC delegates proposing such legislation during every Two Sessions.)

In June, CDT Chinese editors archived four deleted articles about the first animal-abuse case, involving a man in Chongqing surnamed Li who posed as an animal lover to adopt cats and dogs that he later tortured and killed, sometimes uploading videos of the abuse to domestic video-sharing platforms. After animal-rights activists identified the man and began holding daily protests at his building to demand that he be arrested, police broke up the protests—confiscating signs, and beating and detaining some of the protesters. Online content about the protests was widely deleted, and domestic news reports failed to mention the protests or arrests, noting only that the man was briefly detained on charges of vandalism and throwing objects (in this case, a dog) off his balcony.

The first deleted article, “When Perpetrators Go Unpunished, What Exactly Is Society Condoning?” from WeChat blogger Sugarbean, argues that a civilized society must take a strong stand against the abuse of animals. The piece includes social-media content about Li’s animal abuse, and two posters designed to raise public awareness of the problem. A second archived piece in a similar vein seeks to debunk some of the arguments opposing a national animal-welfare law, briefly discusses the history of animal-welfare legislation in other countries, and recommends a Bilibili content creator who provides compelling arguments in favor of such legislation.

Another censored piece, by legal blogger Li Yuchen, contrasts the lenient treatment given to the man who tortured many animals to death and the violent police response to hundreds of concerned protesters. Li notes that on-scene video and first-hand evidence from participants indicate that police officers beat some of the protesters, at least 30 of whom were taken into custody, although this content was censored on Chinese social media and omitted from domestic press coverage. A portion of Li Yuchen’s article is translated below:

Soon after these videos documenting the clashes [between police and protesters] began circulating online, they were hit with censorship, and deleted in quick succession.

That act of censorship is fundamentally more truthful than any video footage. It implicitly acknowledges that yes, something did happen that afternoon, something that wasn’t supposed to be seen.

[…The official statement from local authorities] makes no mention of the protests, clashes, clearing, or arrests that took place on the afternoon of June 9th.

Those who were beaten, dragged away, or taken to the hospital were not mentioned in any official report. In the official version of events, that afternoon is a complete blank.

[…] When that very same toolkit of "disturbing the public order" and "picking quarrels and provoking trouble” is applied to a killer and torturer of animals, it is only enough to merit a few days of administrative detention. But when applied to a group of protesters, it is apparently enough to justify clearing the entire protest site and dragging away large groups of protesters. This toolkit measures only the extent to which public order has been disturbed, without taking into account the amount of suffering inflicted on the victims. As for the degree of “disturbance” to the public order, that is entirely determined by the police on the scene. [Chinese]

Another deleted and archived article, “Lives That the Law Doesn’t See: A Discussion of the Chongqing Animal-Torture Case and Why China Doesn’t Criminalize Mistreatment of Animals,” comes from a sociology student with the WeChat account Some Things Have No Name. The author, likely in an unsuccessful attempt to avoid having the piece deleted, uses “nùe杀” in the title rather than 虐杀 (nüèshā, literally "torture to death"). The scholarly, footnoted article suggests four “layers” impeding the enactment of a nationwide animal welfare law:

The first layer is the issue of identifiability in national governance. A nation governs what it can "see," and seeing requires classification, registration, and detailed lists.

[…] Animal suffering cannot be easily compiled into a list, nor does it correspond to any existing governing departments, nor does it generate any quantifiable metrics. A legal interest lacking a clear governing body is almost destined to stall in China’s legislative process: Should the prevention of cruelty to animals law fall under the purview of the Ministry of Agriculture and Rural Affairs, the Ministry of Public Security, or the National Forestry and Grassland Administration? This question has remained unanswered for sixteen years because each governing body only “sees” the animals on its own list.

The second layer is the influence of industry. China has the world’s largest livestock, fur, and laboratory-animal industries, not to mention the dog-meat economies that exist in several regions. Any law against cruelty to animals, no matter how narrowly it defines "cruelty," will be interpreted by these industries as a slippery slope: today it might prohibit the killing of pets, but tomorrow it might regulate methods for slaughtering livestock, and the day after that, it might regulate stocking density. Legislators aren’t dealing with the specific question of “whether cats and dogs should be protected,” but with the larger systemic question of "whether to open the Pandora’s box of animal welfare." Maintaining the current legal vacuum is the lowest-cost option.

The third layer concerns “ideological vigilance.” The official perception of the discourse on animal rights is that it is a foreign import linked to environmental NGOs, international animal-protection organizations, and even to a broader discourse about “Western values.” Therefore, animal-protection legislation is seen as tantamount to acknowledging the legitimacy of this discourse.

[…] The fourth layer, and the one I most want to address as a sociology student, is the stigma surrounding animal-rights advocates. In public discourse, people who care about animals are associated with a specific image: emotional, predominantly female, middle-class urbanites who believe that "cats and dogs are more important than people." This stigmatization reduces animal protection issues to mere "emotions" even before they enter the legislative discussion, and of course “emotions” have no place in China’s legislative discourse. [Chinese]

A new report that estimates that in 2025, China had 280 million “flexibly employed" workers, a number predicted to rise to 320 million by 2026

In June, CDT editors noted the continuation of a long-running censorship trend targeting economic topics such as youth unemployment, the implications of the gig economy, perceptions of reduced socioeconomic mobility, and criticism and analysis of specific industries and sectors. Two deleted articles (from Phoenix Finance and Planet Business Review, respectively) discussed the National Audit Office’s annual report, which “named and shamed” the Bank of China and other large state-backed banks for tax evasion and improper lending oversight. Another censored piece focused on May’s dismal consumption data, which showed that retail sales fell for the first time in three years. Also deleted was a longform article from Southern Weekly about Cong Rui’an, a doctoral candidate at Tsinghua University who recently published a book about his eight years of intermittent “undercover research” as a day laborer at Majuqiao, Beijing’s largest informal labor market. Cong and another worker offer a nuanced firsthand view of the pros and cons of gig-work and the precarious lives of informal laborers, who are often exploited by labor brokers or subject to wage theft by employers—realities that challenge the official narrative of "flexible employment" (灵活就业, línghuó jiùyè) as an economic solution.

A major topic of online discussion and online censorship last month was the “2025 China Blue-Collar Employment Research Report,” published by the China Center for Research on New Forms of Employment, a semi-official think tank affiliated with the Ministry of Human Resources and Social Security (MOHRSS). Based on nearly 30,000 survey responses, the report estimates that China had 280 million “flexibly employed” workers in 2025, a figure that includes delivery riders and other gig workers; nannies, house-cleaners, and other providers of domestic services; part-time construction workers and day laborers; and other workers without formal employment contracts. The authors of the study predicted that this number would rise to 320 million people by 2026, accounting for more than 40% of urban employment.

CDT editors archived four pieces of deleted content about the report and what effects flexible employment is having on China’s socioeconomic landscape. An article by Ni Ren, from the WeChat account Black Noise, presents some of the figures from the report, and cautions that a huge underclass of part-time and undercompensated workers may doom China’s attempts to promote consumption as a key driver of the economy. “The Truth About the 280 Million ‘Flexibly Employed,’” from WeChat account Inspector No. 25, discusses trends in white-collar and blue-collar employment and wages, and notes that “flexible employment has morphed from a supplementary form of employment into a key pillar [of the labor market].”

Two other deleted posts take very different approaches to analyzing what the rise in flexible employment bodes for social mobility in China. A long piece from the official WeChat account of the Beijing Cultural Review (BCR), authored by Zhu Guanglei and Han Linxiu of the Zhou Enlai School of Government at Nankai University, examines the latest economic data on flexible employment and extrapolates ten characteristics of China’s changing social structure. Employing standard academic and official verbiage about economic development and social stability, the authors posit a fairly rosy outlook on continued social mobility in China, while cautioning against “the proliferation of artificial factors that could disrupt class harmony and the social order.”

A now-deleted WeChat article from Family Heritage Researcher sounds a much more pessimistic note about the potential for social ossification in an economy increasingly dependent on part-time and non-contract labor:

In the face of this reality, a stable labor contract has become a luxury that exceedingly few people enjoy.

[…] The rungs of the professional and social ladder have been severed, destroying opportunities for upward mobility.

Under this structure, things such as marriage and children, home ownership and upgraded consumption have become simply unattainable.

An even bigger problem is that the national strategy to double incomes […] is predicated on building a vast middle class with steady incomes and predictable futures.

But a food delivery rider, even one who earns more than ten thousand yuan a month, can never count on that kind of stability. [Chinese]

An investigation into refrigerated cargo trucks being rerouted to Henan province in an extortionate scheme known as “siphon-style enforcement"

Market supervision officials in Biyang county, Zhumadian city, Henan province, appear to be at the forefront of a new method of extortion known as “siphon-style enforcement,” which marks an escalation of past attempts to fill local government coffers via “deep-sea” or “distant-water” fishing expeditions to shake down targets in other provinces. There is now an official investigation into how Biyang’s Market Supervision Bureau colluded with intermediaries and drivers to lure cargo owners with below-market freight prices on online logistics platforms such as Yunmanman. The trucks (most carrying cold-storage shipments of meat) were then rerouted to pass through Henan, at which point they were intercepted by market supervision officials and their cargo was impounded for supposed violations, with extortionate fees to get the cargo out of storage.

In June, CDT editors archived six deleted articles criticizing higher-level officials for turning a blind eye to such blatant extortion and only launching an investigation when public outcry and press coverage reached a fever pitch. WeChat account Mu Bai’s Writing is Mediocre likens Zhumadian’s officials to “bandits” and lays out the details of their scheme to funnel trucks through their locality and put the squeeze on cargo owners. Blogger Qing Zhemu lists three questions that the investigation must answer: Who reported the illegal scheme, who condoned it, and who should be responsible for financially compensating the freight owners. Similarly, a piece from WeChat account Plainspoken Yu exclaims, “Twenty-one Truckloads of Frozen Goods Were Held Up in Biyang, Zhumadian for Six Months! Who’ll Pay for the Massive Losses?”

Two other deleted WeChat pieces from Fengsheng OPINION and Xu Peng’s History Rhymes argue that Henan’s “siphon-style enforcement” is much more pernicious than old-school “distant-water fishing,” and poses more of a threat to economic growth and progress. Another of the censored articles, from Xiang Dongliang’s Constructive Opinions WeChat account, reminds readers that although there are credible allegations of “siphon-style enforcement” in other counties in Henan, the investigation seems to be limited to a few officials in Biyang county. As such, Xiang predicts that the investigation will likely be superficial, and that the result will be a foregone conclusion—a few minor officials will take the fall, and the rest will go unpunished.

The World Cup and the state of Chinese soccer

Toward the end of June, CDT editors archived an essay from WeChat account The Wandering Mu Bai, titled, “Instead of Criticizing Chinese Fans for Cheering on the Japanese Team, the Shanghai Football Association Should Focus on Qualifying for the World Cup!” In it, the author takes the Shanghai Football Association to task for its nationalistic response to an AFP report showing a group of Chinese fans wearing Japanese national soccer team jerseys and cheering for the Japanese team as they watched a televised World Cup match in a Shanghai bar. Following the AFP report, the Shanghai Football Association issued a "World Cup Football Match Civilized Viewing Initiative," which urged Chinese fans to show national pride and not to “forget” that they were Chinese. A short excerpt from Mu Bai’s essay follows:

One point on which I disagree with the Shanghai Football Association: you can’t accuse those fans of lacking national pride.

I fully believe that if our national team were playing in the World Cup, these fans would be in bars cheering for the Chinese team, rather than the Japanese team.

The problem is, their own team isn’t in the running. Everyone wants to support and be passionate about their own team, but they don’t have the chance.

Some might argue, "Well, that doesn’t mean they ought to support the Japanese team, either."

But as everyone knows, fans are a passionate breed, and once they get into the groove, it’s hard to shake off. Even if their own team underperforms, they can’t stop watching soccer, so they broaden their enthusiasm. Beyond the nation, there’s the continent, and our continent is Asia. And since Japan is Asia’s soccer powerhouse, fans naturally support the Japanese team.

If South Korea played well, the fans would support South Korea; if North Korea played well, they’d support North Korea. It could just as well be Vietnam or Laos: fans are simply trying to find an outlet for their enthusiasm. It’s as simple as that. [Chinese]

For more on the alleged dangers of “scaling the wall,” from moral degeneracy to criminal prosecution, see our recent translation of a piece from the National Defense Times which enumerates them in terrifying detail. There is a long history of similar warnings, but some recent signs suggest that pressure on circumvention tools may be escalating.

To all readers:

To safeguard operation of library networks and fortify the defense of readers’ information security, and in response to a recent situation involving rule violations by some readers with regard to the use of the library’s public Wi-Fi leading to potentialsecurity hazards, we are implementing a temporary shutdown and rectification of public Wi-Fi systems. Details are as follows:

1. Explanation of the rule-violating usage situation

Technical monitoring has confirmed that some readers logged on to the library’s public Wi-Fi network and used unauthorized wall-scaling software to visit prohibited overseas websites, causing traffic irregularities and serious cybersecurity hazards. Such activity is suspected of violating relevant provisions of the Cybersecurity Law of the People’s Republic of China, and has been recorded by the Cyberspace Administration Department. It has so far been verified that the telephone numbers of those involved include: 155XXXX9172; 197XXXX2090; and 155XXXX7044. In order to thoroughly eliminate security hazards, we are urgently implementing a temporary suspension and rectification of our Wi-Fi network.

2. Suspension and rectification schedule

Rectification will begin on July 22, 2026, and the library’s public Wi-Fi will be unavailable throughout the rectification period. Pending completion of cybersecurity troubleshooting work, system reinforcement, and testing work, we will gradually resume service as circumstances allow, with details of the restoration date to follow separately.

3. Special reminder

Cyberspace is not outside the law. We will step up monitoring of our network environment on an ongoing basis, and maintain a "zero tolerance" policy toward rule violations in readers’ network usage. If such violations are verified, the telephone numbers of those involved will be permanently barred from our network; extreme cases of suspected illegal conduct will be handed over to public security, cyberspace administration or other departments.

We thank all readers for your understanding and cooperation, and request your patience with any inconvenience this may cause.

This concludes the announcement. [Chinese]

CDT Chinese editors collected some unimpressed reactions from users on X:

laoji_here: They think that shutting the front gate will solve all their problems, but they’re really just admitting to everyone how scared they are.

vcjavacgi2: This library "administrator" is awfully keen to show his devotion.

thecitylion: We don’t even have such tight surveillance at my company. Will we end up having to hand over our phones and register our real names to go to the library?

GenkiCare: The CCP prevents people from enlightening themselves in the name of information security. You can’t use a single one of the world’s ten biggest sites or apps without scaling the wall.

okok166: It’s just common sense not to use work unit or school networks to scale the wall because they’re small-scale and easy to trace, and the leaders would take the blame, so they definitely will trace it. If you’re scaling the wall, use your own mobile connection or home broadband, there’s strength in numbers, as long as you don’t publish any content that unsettles the net police they won’t bother with you! Work unit’s a different matter, if the network administrators realize someone’s scaling the wall and the leaders ignore it, they’ll be accused of political disloyalty and punished, so of course they’re going to notice, find out it’s you, and sack you!

rainstoneyun: This regime is afraid of people learning the truth, so afraid. [Chinese]

In the midst of this, last week brought the fifth anniversary of the deadly 2021 floods in Zhengzhou, Henan. A leaked media directive showed the typical playbook: "Regarding the heavy rain striking Henan and other places, shift the focus of reporting toward post-disaster recovery […] do not publish unauthorized images showing dead bodies, take an exaggeratedly sorrowful tone, or hype or draw connections to past events." In early 2022, China’s State Council released the results of an investigation that sharply criticized local authorities, in part for obscuring the death toll, but when the first anniversary arrived months later, censorship and opacity reigned again.

Many of the themes in translated commentary on the 2021 floods have resurfaced this year. As commentator Xiang Dongliang wrote in a previously translated post on the current floods: "All of this ought to be common sense by now. But here we are in 2026, and we’re having to reiterate it yet again, which is truly lamentable."

The two censored posts translated below display a similar longing for accountability and transparency, and reflect upon memory and forgetting in that environment. As writer Huang Zhijie, under the pen name "Youyou Luming," concludes in the first post: "Forgetting often means that an essentially identical man-made disaster occurs again; that an accident investigation report that came close to the truth becomes just a blank piece of paper."

Today is July 20, 2026.

This marks the fifth anniversary of the July 20, 2021 ("720") Zhengzhou floods.

To be honest, I really don’t know what to say.

I had actually forgotten all about it.

The first thing I remembered was that "Raincoat Dad," whose real name is still unknown.

That year on July 27, the seventh day after the deaths, a man sat quietly outside Shakoulu station on Line 5 of the Zhengzhou Metro. He wore a dark blue raincoat, black sunglasses, and a blue facemask with an opening cut at the mouth. Next to him, propped against his parked bicycle, was a piece of cardboard reading: "Niuniu, Daddy still wants to bring you home."

He never spoke.

The father.

[…] According to a subsequent police investigation, the man was a section chief at a Zhengzhou work unit, and was the father of Zhang __yue, one of the people who died on that subway line. The man’s younger sister said that he was wearing that raincoat because he had been wearing it when he said goodbye to his daughter that morning; and the sunglasses, because his daughter had given them to him. His mask had an opening because he didn’t want the condensation to fog up his glasses. He had always used that bicycle to take his daughter home.

"Niuniu" was his special pet name for his daughter. Even though she was already 27 and married with a two-year-old of her own, in her father’s eyes, she would always be his little girl. "My brother always regretted not having taught her to swim, and blamed himself for not finding out about the subway flooding in time. He always wondered whether, if he could turn back time, there might have been something he could have done to change what happened."

[…] A total of 398 people died during the “720 incident.” Of these deaths, 139 were concealed at the township, county, or city levels. According to the State Council’s later investigation, many of the flooding deaths should never have happened, particularly the incident on Line 5 of the Zhengzhou Metro (in which 953 people were trapped, and 14 died) and in the northern tunnel of Zhengzhou’s Jingguang expressway (in which 247 vehicles were submerged, and six people died).

Raincoat Dad’s daughter was among those who died.

The investigation report said that the disaster was "generally natural, with specific human failings," and that there were six major issues with the relevant departments:

1. Response deployment was neither timely nor sufficient;
2. Emergency response lagged badly;
3. Emergency measures were imprecise and ineffective;
4. Unified command was lacking at key moments;
5. Organization and mobilization were insufficient;
6. Reported numbers of those dead and missing were delayed and concealed.

At one point afterwards, chrysanthemums left by Zhengzhou residents or sent from netizens from all over the country blanketed the ground far around the subway entrance, and the messages on the cards among them were even more moving. Despite efforts by some to obstruct this, the urge to commemorate the victims was irrepressible.

[…] One card read: "You passed through this world. We remember your bravery!"

A drone photo of yellow bouquets spread out in front of the subway station. Reporters from Caixin and Southern Metropolis Daily were arrested for taking drone images of the impromptu memorial.

Why is it that people are willing to persist, to brave the consequences, in order to express their grief and commemorate the ordinary victims of disaster?

I think it’s because those ordinary office workers, those ordinary daughters, are so much like ourselves. Their fates struck without warning, as could happen to any ordinary person.

People who had been trapped on the subway recalled their fellow passengers, after many hours of confinement, beginning to show signs of oxygen deprivation—shaking, gasping for breath, and retching—while others fainted or suffocated.

Some people online said that just reading these accounts and watching the videos was enough to give them shortness of breath, headaches, or insomnia. For some, it evoked the anxieties of the pandemic years.

When people see the minutiae of these ordinary victims’ lives, both before and after their passing, their great sorrow becomes a form of memory.

I’m conflicted.

On one hand, my general habit is to dig into every detail of a major disaster as exhaustively as possible, trying to remember. On the other, I also tend to forget it all as completely as possible.

Because as far as I’m concerned, this kind of sorrowful memory is an enormous burden. I have to constantly purge it if I’m to return to a normal emotional state.

If the “post-disaster reconstruction” is this hard for me, an outsider, how must it feel for people from Zhengzhou?

As an old media veteran, I seem to know a lot—much more than ordinary people. But I also seem to know very little, to the extent that some great events I witnessed firsthand seem to have left little trace in my memory.

Sometimes forgetting is as terrible as the disaster itself.

Forgetting often means that an essentially identical man-made disaster occurs again; that an accident investigation report that came close to the truth becomes just a blank piece of paper; that the bicycle that carried a child, a father’s raincoat, the sunglasses his daughter gave him … it means that all these details will come into our view once more.

May we who have lived through disaster avoid further disaster; and, having grasped the human causes of catastrophe, may we avoid man-made catastrophe in future.

Youyou Luming, July 20, 2026 [Chinese]

The second post is by Henan-born, Chengdu-based writer Zhang Feng, the owner of a celebrated independent bookstore that narrowly escaped closure last year. Zhang noted that most of the commemorative posts he had seen made no demands for accountability, but were deleted anyway.

In the bookstore today, I heard a pair of young women sighing: "I can’t believe it’s been five years since the Zhengzhou floods."

I was amazed—I hadn’t expected that young folks in Chengdu would still remember it. "What do you remember of it?" I asked them.

"There was a father waiting for his daughter at the subway station, just waiting, for days and days."

I paid close attention to the Zhengzhou floods at the time, and wrote two pieces about them, thinking that if the police in my hometown wanted to come and arrest me for it, then let them. But I don’t even remember that detail.

Those two young women probably got their information through different channels, building their own image of the "Zhengzhou floods."

I’ve seen a lot of videos and articles commemorating the Zhengzhou floods in the last couple of days. There were a lot of comments relating details of that time, mostly expressions of gratitude or calls to remember the city’s suffering. I saw almost no abuse, or demands for accountability.

Sadly, most of those videos and posts were deleted. As someone from Henan, I’m sorry to see this.

Zhengzhou is a large city that’s developed a lot over the past 20 years. I remember that when I left for university in 1997, Zhengzhou hardly merited a mention among China’s cities, but now, in the age of high-speed rail and mobile internet, it’s seen as a hub.

But it’s not a city whose spirit will stir your soul.

I know there’s a pretty good song about it, "Memory About Zhengzhou," a nice tune [embedded below; by banned folk singer Li Zhi]. Chengdu seized on the Zhao Lei song that’s named for it; when I was in Flushing, New York, and heard someone singing "Chengdu" in the street, I couldn’t help but get misty-eyed.

But I’ve seen no sign of Zhengzhou trying to capitalize on its own song that way.

It’s a similar story with commemoration of the floods. Merciless floodwaters brought disaster, but part of it was man-made, or revealed shortcomings in the city’s construction, and these caused further misfortune. But if a city were to establish a "Flood Memorial Hall" to tell the story of the disaster and remind visitors of its lessons, that would be entirely to its credit.

I even have an idea for such a memorial hall. It could be located in a subway station, and use advanced equipment to create a kind of "drowning experience." The names of the victims could be listed, so we would never forget them. Commemorative items could be displayed; the deeds of "heroic citizens" could be recorded.

In any case, a city ultimately belongs to its people. Becoming its "citizen" is not just a matter of buying a house or possessing a household registration; it also involves learning the city’s traditions and culture, and the hardships she has suffered. The people of a place have shared feelings and shared sorrows, which can become ingrained as a sort of "local dialect."

It’s like Chengdu in 2008. If you were here [during the Wenchuan earthquake] in 2008; if you saw or heard about family and friends who died; if you grieved and were heartbroken, but also reached out your hands to others, then how could you not love her?

In that sense, I’m proud of Zhengzhou. The simple fact that so many people have not forgotten gives you a little more confidence. [Chinese]