This course begins where PHIL 145C ends. That course treats a now-famous piece of internet writing — a woman’s reply to a man who told her he would not pay for sex work because it was risky, expensive, embarrassing, and socially dangerous — as a problem in argument analysis. It asks what the woman’s piece presupposes, where it reaches too quickly from one person’s behaviour to an entire gender, and how the accompanying comment threads convert private speech into moral diagnosis. That analysis is real work, and it stays useful in this course too. But it also leaves a question unanswered: why do exchanges like this one appear so predictably on Chinese platforms in the late 2010s and 2020s, why do they travel so quickly once posted, and why do so many readers feel the argument land in their own chest rather than in the abstract?

To answer that, we need to stop treating the exchange as a pair of arguments and start treating it as a social event. The woman on the other end of that conversation was not typing in a vacuum. Her vocabulary for naming the encounter — the distinction between a reason of prudence and a reason of principle, the phrase “嫖娼” as moral category rather than neutral description, the suspicion that a man who starts with risk-management is a man whose ethics are downstream of consequence — already existed in the culture. The man’s defenders and detractors reached for labels already in circulation: 普信男 (ordinary but arrogant man), 女拳 (feminist fist, used as slur), 舔狗 (simp), 厌女 (misogyny). The comment algorithms surfaced certain kinds of replies and buried others. None of this is decided by the two participants. The sociological question is how that surrounding apparatus works, and why it keeps producing conflict that feels new each time while running on machinery that is by now quite stable.

The course pursues that question through a China-led comparison. The Chinese case is not pathological, and it is not uniquely combustible. But the way platformisation, marriage-market pressure, housing anxiety, shifting family expectations, and state-managed speech interact in mainland Chinese digital publics produces a density of online gender antagonism that makes many of the underlying mechanisms visible. Comparable dynamics run in the United States (the manosphere, call-out feminism, the tradwife economy), in South Korea (the Ilbe / Megalia cycle and its political afterlife), in Japan (herbivore discourse, the low-fertility panic), and in parts of Europe. Each case will appear in this course, mostly in Chapter 10. But the home ground is Chinese, because it is the setting where the student for whom this course is built lives, reads, and argues.

Three working assumptions organise the rest of the text, and it is worth stating them at the start because the course will not re-justify them every chapter. First, online gender conflict is not reducible to bad argument or bad personality. It is tied to the institutions of family, labour, media, and law, and if those are stable the arguments will keep returning even when the specific participants change. Second, the conflict is not purely symbolic. Material insecurities about housing, marriage-timing, care work, and physical safety are continually translated into discourse, which is why a fight that looks like it is about one word is often about rent. Third, no honest analysis can proceed by treating “men” or “women” as internally uniform blocs. The strongest sociological explanations preserve internal differentiation — by class, region, education, hukou, sexuality, and age — while still identifying the patterned inequality that connects them.

These assumptions are necessary because digital discourse constantly invites their opposites. A post that circulates successfully usually does so by narrowing explanation, not by widening it: one salient injury, one obvious villain, one emotionally satisfying frame. Sociology is less efficient in that sense. It refuses to reach closure on the schedule of a comment thread. That refusal is part of what the course is trying to train.

Sources and References

Primary books and major scholarly syntheses

• Arlie Russell Hochschild, The Managed Heart: Commercialization of Human Feeling (1983); The Second Shift (1989).
• Erving Goffman, The Presentation of Self in Everyday Life (1959); Frame Analysis (1974); Interaction Ritual (1967).
• R.W. Connell, Gender and Power (1987); Masculinities (1995); R.W. Connell and James Messerschmidt, “Hegemonic Masculinity: Rethinking the Concept,” Gender & Society (2005).
• Judith Butler, Gender Trouble (1990); Bodies That Matter (1993).
• Kimberlé Crenshaw, “Demarginalizing the Intersection of Race and Sex” (1989) and related writings on intersectionality.
• Patricia Hill Collins, Black Feminist Thought (1990).
• Sara Ahmed, The Cultural Politics of Emotion (2004).
• Eva Illouz, Cold Intimacies (2007); Why Love Hurts (2012).
• Nancy Fraser, writings on recognition, redistribution, and feminist politics; with Rahel Jaeggi, Capitalism: A Conversation in Critical Theory (2018).
• Michael Kimmel, Angry White Men (2013).
• Wendy Brown, States of Injury (1995).
• Zeynep Tufekci, Twitter and Tear Gas (2017).
• Tarleton Gillespie, Custodians of the Internet (2018).
• Whitney Phillips and Ryan Milner, The Ambivalent Internet (2017); You Are Here (2021).
• danah boyd, It’s Complicated (2014).
• Safiya Umoja Noble, Algorithms of Oppression (2018).
• Martha Nussbaum, Sex and Social Justice (1999).
• Catharine MacKinnon, writings on pornography and sex equality.
• Ronald Weitzer, research on sex work and regulation.
• Leta Hong Fincher, Leftover Women: The Resurgence of Gender Inequality in China (2014).
• Yunxiang Yan, The Individualization of Chinese Society (2009); Private Life Under Socialism (2003).
• James Farrer, Opening Up: Youth Sex Culture and Market Reform in Shanghai (2002).
• Harriet Evans, Women and Sexuality in China (1997).

Peer-reviewed research and journal literature

• Gender & Society, Signs, Feminist Media Studies, New Media & Society, Information, Communication & Society, Social Media + Society, Chinese Journal of Communication, Theory, Culture & Society, Sexualities, American Sociological Review, American Journal of Sociology, British Journal of Sociology, Journal of Sex Research.
• Research on online misogyny, incel cultures, gendered harassment, platform governance, moral outrage, identity-protective cognition, affective publics, and the ethnography of Chinese internet communities.

Official datasets and institutional reports

• China Internet Network Information Center (CNNIC), annual internet development reports.
• Chinese General Social Survey (CGSS), selected waves.
• World Values Survey; Pew Research Center reports on gender and internet use.
• Published materials from the Supreme People’s Court and Ministry of Public Security relevant to regulation, public order, and gender-based violence.

Chapter 1: From Argument to Social Event

The woman’s reply in the PHIL 145C case works argumentatively by isolating a presupposition. The man she is answering said, in effect, that his reluctance to pay for sex was grounded in risk and cost. She points out that if his only reasons are prudential, then in the absence of risk and cost he would presumably do it; which in turn implies that there is no ethical difference for him between this transaction and any other consumer decision. That is a clean move in argument analysis. It identifies a hidden premise, exposes what the premise commits the speaker to, and refuses to let the speaker retreat into a position he did not originally hold.

But the reply did not go viral because it was a clean move. It went viral because thousands of women recognised themselves in it and thousands of men recognised themselves on the wrong side of it, and because comment sections across Weibo, Zhihu, Xiaohongshu, and Douban spent two weeks turning one private exchange into a generalised diagnosis of Chinese men, Chinese women, the collapse of respect, the hypocrisy of the 小粉红 generation, and whatever else the reader happened to be angry about that week. None of that reception is controlled by the quality of the original argument. It is controlled by something else.

That something else is what the sociologist Charles Tilly, in another context, called the available repertoire of contention (抗争剧目): the finite set of recognisable moves, genres, and accusations that a given public already knows how to perform. When a Chinese internet user encounters a private exchange about sex, consent, and money, she does not interpret it from scratch. She interprets it through a vocabulary that has been settling for the better part of a decade. The post is a 厌女 post or a 女拳 post. The commenter is a 普信男 or a 雌竞 (female-on-female competition) case. The situation is, or is not, an instance of 荡妇羞辱 (slut-shaming). These labels carry compressed causal stories about who is wronged, who is responsible, and what response is appropriate. Platforms did not invent them, but platforms gave them enormous circulation advantage, because labels are cheap to type and expensive to dispute, and comment threads reward what is cheap.

This is what we mean by a public discourse (公共话语). A public discourse is not the sum of what people say on a given day. It is a patterned field in which some ways of speaking become normal, legible, and amplifiable, while others become marked, unintuitive, or slow. When we say that Chinese online gender discourse is a public discourse rather than a collection of private opinions, we mean that the moves available to a participant are already structured before she opens the app, and her choice is not whether to use the available moves but which of them to deploy and how.

A central methodological consequence follows. It is never enough, in sociology, to ask is this user’s argument good? That is the PHIL 145C question, and it is a legitimate one. The sociology question is why does this kind of argument keep appearing here, among these people, now? The answers will involve institutions, platforms, identities, and histories; they will rarely involve the ethical character of individual posters.

Micro, Meso, Macro

One of the durable methodological habits of sociology is insisting that a single event has to be read at more than one level of scale. The PHIL 145C exchange illustrates the point. At the micro level (微观互动), we have two people in a text conversation, using particular words, managing a particular interactional situation, performing particular identities to one another. Both parties are also performing to an imagined audience: the friends each will tell about this, the screenshotters who may later surface it, the version of themselves they want the other to remember. Goffman’s language of impression management (印象管理) fits here. The conversation is not only information transfer; it is identity work under observation.

At the meso level, we have the platform environment: which app this happened on, what that app rewards, what gets promoted to the For-You feed, what the comment ranking algorithm does, what moderation is applied to which words, which subcommunities pick it up first. The argument would have landed differently on Douban than on Weibo, and differently on Weibo than in a WeChat group, even with identical text.

At the macro level, we have the gender order (性别秩序) in which this conversation is intelligible at all: a housing market that makes marriage expensive, a marriage-market discourse that makes female age into a wasting asset, a sex-work regime that is formally illegal but practically present, a wage structure that still pays men more on average, a family system that still asks adult children to absorb parental expectation, and a media economy that profits from making all of this into spectacle.

No one of these levels is sufficient. A purely micro reading makes the dispute look like two people’s personalities. A purely meso reading makes it look like algorithms alone. A purely macro reading makes it look like there are no individuals at all, only structures moving through them. The course repeatedly moves between the three, because the interesting questions usually live at the seams where levels do not line up cleanly.

Issue Conflict and Identity Conflict

One more distinction, which will recur, is between issue conflict (议题冲突) and identity conflict (身份冲突). An issue conflict is a disagreement about a claim, a policy, or a case — whether sex work should be decriminalised, whether a given divorce settlement was just, whether a piece of legislation will reduce harassment. An identity conflict is a disagreement in which the disagreement itself becomes proof of who a person is. Once the latter shift occurs, a discussion about the legal regulation of prostitution becomes a test of moral belonging: are you the kind of person who takes women’s autonomy seriously, or are you the kind of person who doesn’t?

This distinction matters because most of the discourse we are going to read in this course is, technically, issue-framed — about dating norms, sex-work policy, bride price, workplace harassment, parental leave, online moderation. But emotionally, most of it is identity-framed. Participants behave as though what is at stake is not the policy but the moral standing of everyone in the conversation. The slide from issue to identity is one of the most reliable features of contemporary online gender discourse, and chapters 7 and 8 will spend real time on it.

A Note on Speed

One more feature of the PHIL 145C case is worth flagging before we move on, because it will recur throughout the course. The case became what it became within about seventy-two hours. By Monday morning, versions of the screenshot were on Weibo. By Tuesday evening, critical reply videos were on Douyin. By Thursday, there were Xiaohongshu posts in both directions, commentary from several mid-tier public-account writers on WeChat, and the first wave of secondary responses attacking not the original parties but the people who had commented on them. By the following Monday, a new case had displaced it, and by two weeks later, anyone who tried to cite it had to include a summary because the reader no longer remembered the specifics.

This timescale is not a detail. It is a structural feature. Public discourse that runs at this speed has no time for the kinds of clarification, retraction, and correction that slower discourse permits. Claims circulate in the versions in which they are most portable; inconvenient complications are shed in transit. The resulting public — what some media scholars call a velocitised public — is not simply the old public but faster. It is a different public, with different cognitive demands on its participants and different tolerances for ambiguity. A careful reader has to train herself to read at a different speed than the discourse runs, which is one of the implicit disciplines this course is trying to teach.

By the End of This Opening Chapter

We already have a rough map. Online gender discourse in China, and comparable publics elsewhere, is conducted through a repertoire of already-established labels and genres. Its participants are visible to each other and to audiences at multiple scales. Its arguments register as issue-framed but run as identity-framed. Its outcomes are not decided only by individual speakers but by an apparatus of circulation that treats some moves as cheap and others as expensive. And its speed forces all of the above into an interpretive environment where careful reading is itself a non-default practice. The rest of the course is an attempt to understand that apparatus, the speeds at which it operates, and the forms of attention that can be sustained inside it.

Chapter 2: Gender Order, Patriarchy, Hegemonic Masculinity

The oldest term in this toolkit is patriarchy (父权制), and it is also the most abused. In casual online use “patriarchy” serves as a total explanation for any objectionable male behaviour and, on the other side, as a straw position to be ridiculed as ideological excess. Neither of those uses is what the term means in sociology.

Patriarchy, as sociologists use it, names a set of historically variable institutional arrangements in which men as a group are structurally advantaged in authority, status, resources, and control over institutions. Two things are important in that formulation. The first is “as a group” — which is a claim about distributions and institutional access, not about whether any specific man in any specific room is in charge. A migrant construction worker, an unemployed graduate, and a venture-capital founder do not have the same relationship to power, and sociology cannot afford to pretend they do. The second is “historically variable.” Patriarchy in a Qing-dynasty gentry household, in a Mao-era work unit, in a Shenzhen tech firm, and in a Shanghai marriage-market discourse are not the same system. The word names a family of arrangements, not a single eternal hierarchy.

This is where R.W. Connell’s concept of hegemonic masculinity (霸权男性气质) earns its keep. Connell’s claim is not that most men embody some dominant ideal type. It is that every field of practice tends to produce a culturally authoritative model of masculinity against which other masculinities are measured and from which women are excluded. In contemporary urban China, the hegemonic model fuses several elements that do not always sit easily together: economic provision (house, car, stable income), sexual competence (a woman who can be shown to others), emotional control (不矫情, not fragile or whiny), family-proper conduct (respect toward parents, especially mother), and increasingly a certain kind of cosmopolitan polish. Few men hit all of these. Most men fall short on at least one and feel it. The hegemonic model is not a description of reality; it is a scale that organises male self-evaluation.

That scale interacts badly with labour markets and housing. A man in his late twenties working a good but unspectacular white-collar job in Chengdu, paying rent that consumes a third of his income, facing a property market in which a one-bedroom requires either family assistance or a two-decade mortgage, is living an ordinary contemporary Chinese life. But every signal in his environment is telling him that ordinariness is failure. Marriage-market discourse frames him as 普信男 if he imagines himself desirable on his current terms; his parents ask every New Year when he plans to buy a house; viral essays explain that women his own age are out of reach because he is not “excellent” enough; friends who did buy property earlier now look prescient rather than lucky. His grievance, when it arrives, is not a discovery that the world is unjust. It is the recognition that the hegemonic masculinity script requires him to succeed at a game whose entry ticket is outside his labour-market reach.

This is also why a purely moral vocabulary fails. Telling such a man that his feelings are “entitlement” is not wrong in all contexts, but it does not describe what has happened to him. He is not insisting that he deserves a spouse. He is noticing that the cultural script he was trained on — produce, provide, protect, and in return be chosen — is being run with an increasingly unreachable threshold. Online grievance discourse then offers him ready-made explanations: women have become materialistic, feminism has ruined them, the state has favoured them, other men have stolen the field. These explanations are analytically weak, but they are narratively legible, and they attach his private confusion to a public enemy. That attachment is what we will later call an affective investment (情感投入) in identity-framed discourse.

Emphasised Femininity and Its Scripts

Connell’s original framework also included the concept of emphasised femininity (强调的女性气质), which names the culturally authoritative model of femininity that is produced in relation to hegemonic masculinity — not opposite to it, but structured as its complement. The emphasised-feminine ideal is the femininity that a hegemonic masculinity treats as desirable, and it is typically organised around attractiveness, emotional attentiveness, supportive competence, and a specific relation to motherhood and household.

Connell was careful to note that emphasised femininity is not the only femininity; it is simply the culturally authoritative one, and most actual women negotiate with it rather than embody it. But because it is authoritative, it organises the symbolic economy in which other femininities are judged. In contemporary urban China, the emphasised-feminine ideal fuses several components: slim, pale, well-dressed in a particular cosmopolitan style; emotionally warm and supportive without being clingy; professionally competent without being threatening; ready for marriage and motherhood at a specific age; proficient at the domestic labour that marriage is presumed to require.

Just as few men hit all of the hegemonic-masculine components, few women hit all of the emphasised-feminine ones. The differences between ideal and actual are managed by the performers themselves — through self-presentation, consumption, social-media curation, and, increasingly, appearance-modifying interventions — and the management is itself the labour that the course has earlier called emotional labour and relational labour. It is a full-time job to keep one’s publicly-visible femininity close enough to the authoritative template that it does not disqualify.

The online consequence is that Chinese female influencer content on lifestyle platforms is, in substantial part, the performance of emphasised femininity at commercial scale, and its audience is substantially other women learning how to perform it. This is neither cynical nor simple. The performers often hold ambivalent relationships to the ideal they perform, and the audience often holds ambivalent relationships to the ideal they are consuming. But the ambivalence does not make the performance-consumption cycle stop. It continues, and it shapes what the gender order looks like from the inside.

Intersectionality and Its Limits Online

Feminist sociology counters any single-axis account by insisting on intersectionality (交叉性): gender is always lived through class, region, race, sexuality, age, disability, and other stratifications, and a discourse that speaks only of “women” and “men” as two totalised blocs is usually about to make an analytical error. Kimberlé Crenshaw’s original argument was that the anti-discrimination law of the late 1980s United States could not protect a Black woman who was discriminated against because she was both Black and a woman, because the legal categories recognised each identity separately. Patricia Hill Collins extended the argument into a broader critique of the reproduction of knowledge itself: whose experiences become treated as general, and whose as merely particular.

The Chinese analogue is less about race and more about hukou (户口), region, and rural/urban position. When Weibo marriage-market discourse says “现在的女生都只要有房有车的男生,” the “女生” in question is almost always an educated urban woman in a first- or second-tier city. The 留守妇女 (woman left behind in a rural village by a migrant husband), the 打工妹 (female factory worker) in Dongguan, the graduate student in Hubei, and the recently laid-off tech worker’s wife are different women in different labour markets, with different risk profiles and different marriage markets, and lumping them together erases more than it clarifies. The same is true on the other side: the migrant construction worker, the 985/211-graduate coder, the second-generation business owner, and the rural teacher inherit very different versions of the masculine script, and the grievance each is prone to is different.

Intersectionality online is therefore not optional. It is a guard against a specific recurring error, which is the move from “this story about urban middle-class dating” to “this is what men/women are like.” Platforms encourage that error because the most visible users are disproportionately urban and middle-class and because their concerns generalise badly.

Social Reproduction

A last concept for this chapter: social reproduction (社会再生产). In feminist political economy, the term names the labour and institutions that sustain daily life and the next generation — childcare, eldercare, household work, emotional support, the transmission of norms and skills. This labour is historically female, historically unpaid, and historically invisible to wage statistics, which is why feminist economists since the 1970s have insisted on naming it.

Most of what gets fought about online in Chinese gender discourse — bride price (彩礼), dowry (嫁妆), who pays for the wedding, who buys the house, who names the children, who cares for aging parents, who takes parental leave, whose career adapts to whose, who goes to 家长会 (parent-teacher meetings), who manages the 补习班 (tutoring centre) schedule — is, underneath the surface, an argument about the distribution of reproductive labour. The arguments look interpersonal; they are institutional. A couple arguing over whether he will quit his job to let her career run is not doing a personal negotiation in the way a mid-century advice column imagines. They are inheriting a generational pattern in which his career was presumed dominant, and they are trying to negotiate it without any social infrastructure (public childcare at scale, subsidised eldercare, generous paternity leave) that would reduce the stakes of the decision.

Every time you read a viral Chinese post about “现在的男人/女人想要什么,” ask whether the underlying dispute is a distributional question about social reproduction that has been personalised into a character argument. It usually is.

A Concrete Example: The 彩礼 Debate

To see this operating in one coherent example, consider the recurring online argument about bride price (彩礼). The stylised online version of the debate has two positions. The first says that bride price is a structural problem because it converts women into priced objects, damages young men economically, and reveals a materialistic turn in Chinese marriage. The second says that bride price is defensible because it compensates the bride’s family for the loss of a daughter, functions as insurance against divorce, and signals the seriousness of the groom’s commitment. These two positions fight repeatedly on Weibo, usually with one side’s supporters calling the other side’s supporters misogynistic, materialistic, or disconnected from “real” Chinese life.

Now read it sociologically. Bride price in contemporary mainland China is highly regionally variable — ethnographic and survey work has documented rates from essentially zero in some urban areas to sums exceeding a million yuan in parts of Jiangxi, Fujian, Henan, and Gansu. The high-bride-price regions are disproportionately areas with imbalanced sex ratios (a product of sex-selective abortion during the one-child policy, whose demographic effects are now working through marriage markets), relatively weak local labour markets, and declining agricultural income. In those regions, bride price functions partly as intergenerational wealth transfer (from the groom’s family to the bride’s, compensating the bride’s family for having raised a daughter whose labour will now be performed for another household) and partly as risk-pooling (the bride’s family retains something if the marriage fails, because legal protection after divorce is weak in many respects).

None of that is a defence of bride price. It is a description of the institutional work it does. The online debate, conducted largely between first-tier-city young adults who have little direct experience with high-bride-price contexts, frames the practice either as simple patriarchal commodification or as simple female gold-digging. Both frames miss the function. The dispute is actually a dispute about the distribution of risk and labour across two households at a moment when neither the state, the labour market, nor the family-law system provides what participants would need to make marriage safe on other terms. Personalising it into “Chinese women want X” or “Chinese men deserve Y” converts an institutional question into a character question, and the character question can never be resolved, because it is the wrong question.

This is the characteristic move of online gender discourse. Chapter 7 will return to it when we examine how labels like 扶弟魔 and 接盘侠 perform the same personalisation work on slightly different underlying disputes. For now, the point is that the entire grammar of the debate — its accusatory tone, its totalising generalisations, its demand that each participant declare a moral position on “what women want” — is downstream of a simpler institutional fact. Social reproduction in contemporary urban China has to be negotiated between two households under conditions that make the negotiation hard, and the hardness of the negotiation gets pushed into the only place it can be safely expressed, which is the rhetorical field.

Chapter 3: The Platform as Actor

A platform is not a container. It is an institution with its own incentives, and the incentives are not neutral between types of speech. A sociology of online gender discourse that takes the platform seriously treats it as one of the actors in the conflict, not as the setting.

Consider what a typical Chinese short-video platform does to a piece of speech about gender. An account posts a sixty-second video — say, a woman narrating a frustrating encounter with her boyfriend’s parents. The platform’s recommendation system immediately has to decide what to do with it. The signal it watches most closely is early engagement: comments, shares, rewatches, time-on-video, emotional reactions. Certain kinds of content reliably generate that engagement: content that produces a strong identification response in one audience and a strong rejection response in another. A video that says “my boyfriend’s mother believes X, and I think X is outrageous” will do this. Half the audience will feel vindicated; half will feel provoked; both will comment; both will share it to others who they expect to react the other way. The video spreads. The platform has not “chosen” misogyny or misandry. It has chosen engagement, and engagement has selected for antagonism.

This is what we mean by algorithmic amplification (算法放大). The algorithm is not neutral, but it is also not ideologically directed in the way conspiracy theories imagine. It is instrumentally directed — at the extractable value of attention — and antagonistic gender content is attention-rich. Over time, this selection pressure changes what gets made. Creators learn which formats travel. They make more of those. Audiences habituate to the resulting rhythm: big opening claim, a clip of provocation, a punchline that demands a stance. The rhythm is not, in itself, about gender. But it selects for speech that produces factional reaction, and gender supplies an endless stock of such reaction.

Platform Governance and Uneven Moderation

Platform governance (平台治理) is the set of rules, design decisions, keyword filters, appeal procedures, and moderation practices through which the platform manages what gets said and what gets suppressed. Governance is never total, never neutral, and never applied evenly. A Chinese platform will remove certain forms of explicit political speech reliably, remove pornography with some reliability, remove certain insults with less reliability, and almost never remove coded slurs that do the same work as the explicit ones. A misogynistic or misandrist post that avoids the top-layer banned words will usually survive. This is not a bug; it is the mathematical consequence of keyword filtering plus an enormous volume of posts.

The uneven pattern has sociological consequences. When governance removes the overt forms of a grievance but leaves the covert forms, the covert forms take over. Users learn what is sayable. Moderation thus does not eliminate antagonism; it restyles it. Chinese internet users, who have been operating under this pressure for longer and in more varied ways than users in many other publics, are accordingly extraordinarily inventive with coded speech. One of the durable fascinations of Chinese online gender discourse for comparative sociology is that so many of its recognisable labels — 普信男, 舔狗, 妈宝男, 接盘侠 (man who marries a woman with a previous relationship), 伏弟魔 (daughter who impoverishes herself for her brother), 扶弟魔 (same gist, slightly different coding) — evolved quickly and stayed flexible precisely because moderation forces discourse into coded form.

Affective Publics

The media theorist Zizi Papacharissi’s term affective publics (情感公众) helps name what happens when users gather around a circulating feeling more than around a circulating position. An affective public does not need coherent doctrine. What holds it together is the shared feeling that one has been dismissed, exploited, or made ridiculous by the other side. People who disagree with each other on specifics — should sex work be legal, should bride price be banned, should mothers-in-law have a say in the marriage — can still share the affective bond of resentment toward a common enemy, and that bond will organise their speech more strongly than their professed positions.

This is one reason gender discourse communities are so resistant to argument. The community is not held together by positions; it is held together by feeling. Refuting a position does nothing to the feeling. This also explains why moderation cycles rarely reduce such communities. Suspending accounts removes individual speakers; it does not dissolve the affective attachment that makes replacement speakers appear.

Metrics as Evidence

Finally, platforms quantify everything they touch. A post acquires a visible like count, share count, comment count, a visible percentage agreement among reactions. These metrics become epistemic objects in their own right. A post with 200,000 likes feels representative of public opinion. It is not. It is representative of platform engagement on a given day inside a recommender system optimised for a specific purpose. But the feeling of representativeness is extremely hard to shake. A user who is repeatedly shown high-engagement posts from one side of a gender argument will come to believe that side is winning, or losing, or larger, or smaller — and will calibrate their own speech accordingly.

This metric-as-evidence dynamic is particularly dangerous in gender discourse because participants are primed to be looking for proof of where the culture is going. A like count serves that hunger. Whether the like count reflects anything real about the Chinese population in 2026 is a separate question that the metric itself cannot answer.

Platform Differences Matter

One of the reliable errors in casual talk about “Chinese online gender discourse” is to treat the various platforms as interchangeable. They are not. The platforms have distinct demographics, distinct moderation cultures, distinct interaction rhythms, and distinct economic models, and the same content behaves very differently across them.

Weibo (微博) is text-first, short-form, high-churn, and skews toward immediate reaction. Its discourse culture is argumentative, performative, and heavily influenced by celebrity accounts and paid opinion amplifiers. Weibo is where a gender controversy becomes “the current thing” because Weibo has the fastest news cycle. It is also where moderation is most visible, in the sense that suspended accounts and deleted threads are part of the everyday texture of the platform.

Zhihu (知乎) is long-form Q&A, originally oriented toward high-education users and technical topics, gradually shifted toward a broader and more commercially-oriented audience through the 2010s. Gender discourse on Zhihu tends to take the form of extended answers with the pretence of neutrality, citation, and analytical rigour. A characteristic Zhihu genre is the “experienced commenter” answer that frames a gendered claim as derived from data, life observation, and cross-cultural comparison. The actual epistemic quality of these answers varies enormously. Zhihu’s discourse is slower than Weibo’s and more likely to produce the kind of coded, semi-respectable anti-feminism that escapes simple keyword moderation.

Douban (豆瓣), with its group-based structure, has historically been one of the main incubators of Chinese internet feminism, particularly on the 女权之声 and successor groups. Douban’s moderation has been periodically aggressive toward feminist content — several prominent groups have been closed over the past decade — which produces a discourse habit of coded speech, quick migration to successor spaces, and sharpened community boundaries. The users who stay are disproportionately committed.

Xiaohongshu (小红书), originally a lifestyle and e-commerce platform, acquired a large female-skewed user base and became, in the late 2010s and 2020s, a major site of gender-coded lifestyle content: beauty, fashion, relationship advice, dating strategy, marriage analysis. Its gender discourse is usually wrapped in lifestyle register and rarely confrontational on the surface, but the underlying assumptions about what counts as a good relationship, a good partner, a good wife, or a good mother are highly contested.

Douyin (抖音) and Kuaishou (快手) are short-video platforms with different demographics — Douyin skews more urban and first-tier, Kuaishou more non-coastal and lower-tier, though the difference has narrowed. Gender content on short-video platforms operates through a distinctive format: a provocation video, a reaction video, a reaction-to-reaction video, a compilation. The algorithmic promotion of this chain is one of the clearest cases of platform-driven antagonism generation.

WeChat public accounts (微信公众号) host the longer-form essays that circulate around a controversy: the 5000-word Xiaohongshu influencer post, the 3000-word Douban-feminist analysis, the 8000-word masculinity-studies piece from an academic adjacent to the discussion. These essays shape the eventual narrative of a controversy even if their real-time engagement is modest, because they provide the shared texts that commentators on other platforms quote.

Each of these platforms has a different business model, a different advertising ecosystem, a different relationship to state regulators, and a different engineer’s-eye view of what it is optimising for. A careful analyst of Chinese online gender discourse has to know which platform a piece of content lives on, because the platform is doing interpretive work on the content before any human reader sees it.

Moderation as Selection Pressure

A deeper point about moderation. The naive model of moderation is that it removes bad content and leaves good content. The actual model is that moderation changes what kinds of bad content can be produced, which changes what production incentives creators face, which changes the overall texture of discourse.

Consider the Chinese moderation environment specifically. Overt political speech about party leadership, about specific sensitive historical events, or about specific contentious policies is removed fast and reliably. Overt pornography is removed fast and reliably. Specific prohibited words are filtered automatically. But coded versions of most content categories — speech that uses puns, homophones, image-based circumvention, or simple rephrasing — survive much longer. Moderation therefore shifts discourse toward coded forms, which has several downstream effects.

First, the coded forms are often funnier, more ironic, and more in-group-legible than the uncoded forms would have been. This makes the moderated discourse paradoxically more attractive to retain readers. Second, the coded forms are harder for outside observers — including researchers, non-native speakers, and state agencies themselves — to track, which accelerates the dialectic between moderation and circumvention. Third, the coded forms develop a community ownership dimension that the uncoded forms would not have had: the community that can parse the code is a community in a way that a set of uncoded posters would not have been.

Gender discourse illustrates all of this. The sharpest Chinese feminist writing in 2026 is not written in the register of 2016 feminist writing, not because the writers changed their minds but because the moderation environment changed, and the new writing has to be legible as feminism to feminists and illegible as organised feminism to moderators. The resulting prose is more allusive, more layered, more dependent on shared reference. Some of it is very good. All of it is a specific product of the institutional environment in which it is produced, not a transparent window onto what its authors would have written under other conditions.

This is why, for a serious student of Chinese gender discourse, reading the platform environment is not separable from reading the texts. The text is a move in a game whose rules the platform sets. The platform is an actor.

Chapter 4: Emotional Labour, Face-Work, and Self-Presentation

When two people argue about sex, dating, money, and moral character in public, they are almost never only exchanging positions. They are also performing selves to an audience, managing the impression those selves make, and doing what Hochschild called emotional labour — the work of producing the appropriate feelings (and appropriate displays of feelings) given the interactional rules in force.

Arlie Hochschild’s original formulation of emotional labour (情感劳动) was about paid service work: flight attendants who must produce friendliness, debt collectors who must produce controlled menace. But the concept has wider reach, and nowhere wider than in public gender discourse. Women online are frequently expected to perform a specific affective mix: sympathetic but not naive, firm but not aggressive, sexual but not disreputable, emotionally literate but not hysterical, political but not militant. Men are expected to perform a different mix: confident but not arrogant, competent but not boastful, emotionally self-managed but not cold, sexually interested but not creepy, protective but not paternalistic. The windows are narrow, and the punishment for stepping outside them is quick.

A single interaction sits inside these rules. Consider the original PHIL 145C exchange. The man’s answer (“it’s risky, expensive, embarrassing”) is not a moral statement; it is a performance of realist masculinity, calibrated for male peer audiences in which earnest moralism is a face-losing move. The woman’s reply, which reframes his answer as ethical poverty, is a performance calibrated for a female-coded audience in which moral seriousness is a face-saving move. Neither utterance is a clean window onto the speaker’s soul. Both are moves in a game whose rules differ by gender-coded audience, and both succeed or fail partly by reference to those rules.

This is why internet arguments about “what did he really mean” tend to get nowhere. The utterance is not a record of a belief; it is a situated move. A careful sociologist does not ask what the speaker truly believes. She asks what audience this move is pitched to, what face it is designed to preserve, what face it is designed to damage, and what moves are available in response.

Face-Work and the Anticipation of Humiliation

Goffman’s face-work (面子工作) supplies the second half of this story. Interactants work to avoid humiliation — their own and, when social norms demand, others’. In public digital discourse, face threats accumulate faster than they resolve. To be labelled naive, predatory, manipulative, bitter, prudish, cheap, or hypocritical is to suffer symbolic injury, and the injury is permanent in the sense that screenshots do not decay. As a result, users adopt defensive rhetorical positions before substantive argument. They pre-emptively frame themselves as rational, trauma-aware, structurally informed, experienced, or simply tired of idiots. These pre-framings are not random verbal mannerisms. They are face-strategies in an environment that treats face as a resource under constant attack.

One of the most useful exercises in reading Chinese gender discourse is to ignore the surface position of a post and read only the face-work. A man who opens with “不是我说,” or a woman who opens with “我不是那种女拳,” is not primarily stating a position. She is pre-empting the accusation she expects. Once you see this pattern, the thread is often more legible as a sequence of face-manoeuvres than as a sequence of claims.

Relational Labour and Its Gendered Distribution

A further wrinkle: the discursive labour of explaining, teaching, pacifying, and making-legible is itself gendered in its expected distribution. Women online often report a distinctive exhaustion at being expected to explain, patiently and repeatedly, what harassment is, what catcalling does, why a given remark is not a compliment, why a given law matters. Men online often report a mirror-image exhaustion at being expected to publicly perform moral refinement in a register that conflicts with the realist-masculine register rewarded among male peers. These experiences are not symmetrical in structure, stakes, or consequence — women’s labour here is tied to real safety costs and real professional costs in a way that men’s is usually not — but both are genuine instances of relational labour (关系劳动) that the discussion itself demands, and both help explain why serious participation in these discourses is so draining.

The accumulated cost of this labour has a quiet effect on who stays and who leaves. Those with the most analytically careful positions often exit the discourse, because the cost of maintaining those positions under constant face-threat is higher than the cost of maintaining simpler, more totalising positions. What remains, over time, is a skewed field in which the loudest and most identity-locked participants are over-represented. This is one of the mechanisms by which online discourse grows more polarised without any individual participant becoming more polarised: the moderates just stop posting.

The Second Shift Online

Hochschild’s 1989 book The Second Shift documented that American women in dual-earner couples performed the majority of household and care labour even when they earned comparable incomes to their partners, and that the ensuing time poverty and emotional exhaustion were not equally shared. The Chinese analogue has been documented by Yan Yunxiang, Zheng Tiantian, Ji Yingchun, and others: educated urban Chinese women in dual-earner marriages still perform the majority of domestic and care labour, even as they approach or exceed their husbands’ earnings. The pattern has been labelled 双重负担 (double burden) in Chinese feminist writing.

The online discursive version of this is what we might call the second shift online. A Chinese woman who posts about harassment, discrimination, or relational inequality is often subjected to a wave of counter-posting that asks her to justify her position, defend her tone, provide more evidence, concede partial points, and reassure male readers that her analysis is not anti-male. Each of these demands is a labour request. The cumulative time spent on these demands is measurable: some Chinese feminist writers have documented spending hours per day on comment moderation and reply labour on their own posts. None of this labour is paid, and none of it is equally shared with the male interlocutors who generate the demand.

The gendered distribution of this labour has a second-order effect on the production of analysis. A woman who spends two hours a day moderating her replies has two fewer hours to read, write, and think. Over a career, this is a substantial redistribution of intellectual energy. The male interlocutor who makes the demand has no equivalent burden. The long-run effect is that Chinese (and non-Chinese) feminist analysis is produced at a systemic disadvantage relative to the discourse it responds to, and a portion of that disadvantage is mechanically an effect of the discourse environment itself.

Emotional Labour and the Influencer Economy

A further layer is that the influencer economy has commercialised emotional labour in gendered ways. A female dating-advice influencer on Xiaohongshu is not only performing a persona; she is delivering a service whose value is partly the emotional labour of holding her audience’s attention, soothing their anxieties, and validating their concerns. The male financial-advice influencer on Douyin is performing a different emotional labour: projecting competence, reducing anxiety about money, modelling a kind of stable adulthood.

What makes this sociologically interesting is that the commercial register of the labour is different by gender, and the consequences when the register is broken are different. A female influencer whose content becomes too critical, too angry, or too explicitly political often loses followers. A male influencer whose content becomes too soft, too vulnerable, or too emotionally open often loses followers. The economic incentive is therefore to stay within the gendered register. Over time, this produces a cohort of high-visibility influencers whose publicly available selves are exaggerated gender performances of the register their audience rewards.

Now fold this back into gender discourse. When a dispute happens between a female and a male influencer, the dispute takes place between two pre-calibrated performances, neither of which is the private self of the performer. The audience, which has been trained to read these performances as authentic, takes the dispute as a genuine encounter between “a woman” and “a man.” It is not. It is a dispute between two commercial avatars. The authenticity of the dispute is a function of the commercial system, not of the interpersonal reality. This is one of the most under-studied features of contemporary gender discourse, and it is also one of the least visible, because the commercial scaffolding is designed to disappear into the content.

Codes of Fragility and Codes of Toughness

A specific variation of face-work worth naming is the contrast between what we might call fragility codes and toughness codes in Chinese online gender performance. Fragility codes — softness of tone, self-deprecating anecdote, explicit vulnerability, willingness to cry on camera, emphasis on being “被X伤害过” — are rewarded in certain audience segments (particularly on Xiaohongshu and in female-majority WeChat reading audiences) and penalised in others (particularly on Weibo and in certain male-majority subcultures, where fragility is decoded as “玻璃心”). Toughness codes — ironic distance, blunt declarative statements, public mockery of the other side, performative rationality — are rewarded on Zhihu-style long-form platforms and in certain Weibo subcultures, and penalised on Xiaohongshu.

The same speaker can perform both, but not simultaneously, and not to the same audience. A Chinese female feminist writer who performs fragility on Xiaohongshu and toughness on Zhihu is not being inconsistent; she is calibrating for her audience. A male participant who performs toughness on Weibo and vulnerability in a private WeChat group is doing the same. Cross-platform readers occasionally catch these performers in the mismatch and accuse them of hypocrisy, but the “hypocrisy” is simply the visible edge of a performer whose repertoire is wider than a single platform’s register.

Understanding this is one of the small methodological disciplines the student of this course should pick up. Never evaluate a performer’s positions from a single platform. Always check adjacent platforms to see how the same person adjusts. The adjustments reveal the rule set the performer is working under, and the rule set is more sociologically interesting than the performer’s “true” view (which is often not well-defined).

The Private-Public Seam

One further complication. Chinese users often maintain multiple online identities across overlapping networks, ranging from the fully public (Weibo account under a known name) through the semi-public (Xiaohongshu lifestyle account under a pseudonym) to the private (WeChat朋友圈 visible only to a few hundred contacts). The same person’s speech on a gender topic may be radically different across these registers, not because she is hiding her views from the public register but because the private register is subject to different face-work, different audience expectations, and different possible consequences.

What users say on their 朋友圈 to friends and family is often more moderate, more ambivalent, more idiosyncratic than what they say publicly. The public register demands a performed coherence; the private register tolerates real-time processing. A student who has done fieldwork in Chinese online communities will have encountered this pattern repeatedly: the “extreme” public voice and the “moderate” private voice coexisting in the same person without either being insincere.

This has a subtle implication for how to read Chinese gender discourse. The visible discourse is the portion of Chinese gender thinking that has passed through the face-work and audience-calibration processes of public performance. The actual distribution of views — what Chinese adults, in thousands of unobserved conversations with close friends and family, are actually working out — is considerably wider, quieter, and less factional than the visible surface suggests. There is no method for accessing it cleanly, but the careful analyst knows that it is there, and knows that the surface is not the full population.

Why Hypocrisy Is the Master Accusation

A final observation about the emotional-labour structure of gender discourse: accusations of hypocrisy are disproportionately common, and disproportionately effective. The reason is that public actors in this discourse are judged not only on their claims but on whether their performed self matches the moral role they claim. A commentator who speaks in therapeutic language but appears cruel, or in realist language but appears evasive, becomes rhetorically destroyable. This makes consistency performance a full-time job for anyone who stays long enough to build an audience, which is why influencers in this space tend to speak in increasingly narrow, branded, pre-rehearsed registers. The alternative is being caught out, and being caught out is socially terminal.

Chapter 5: Sexual Scripts, Respectability, and the Moral Regulation of Intimacy

Sexual conduct is never governed by law alone. It is also governed by culturally specific scripts that tell people what a situation means, what roles are available, what risks are thinkable, and what justifications can be offered. Gagnon and Simon’s sexual script theory distinguishes three levels of organisation: the cultural scenarios (文化脚本) available in a society, the interpersonal scripts (人际脚本) two actors work out between them, and the intrapsychic scripts (内心脚本) an individual uses to narrate her own desire and behaviour. The three levels talk to each other, but they are not identical, and public gender conflict frequently occurs exactly at the points where they fail to align.

A useful way into this is respectability. Respectability (体面 / 可敬性) is the demand that legitimacy be earned through disciplined conformity to dominant norms. It is the claim, issued by institutions and enforced by peers, that certain conduct marks you as a serious person deserving of protection, advancement, or marriage, and other conduct marks you as self-disqualifying. Respectability is always gendered, and it is always stratified.

In contemporary urban China, female respectability is disproportionately loaded onto the management of sexual history, marriage timing, and public conduct. A woman whose sexual history is perceived as too extensive, or too visible, or too early, pays a marriage-market penalty that is not imposed on male counterparts. A woman who marries late pays a different penalty. A woman who does not marry at all pays a social penalty through her parents, even when she is materially self-sufficient. Male respectability, by contrast, is disproportionately loaded onto provision and status: income, housing, family contribution, professional trajectory. A man without those is marked as unserious in a way that no amount of personal conduct repairs.

The Sex Work Debate as Symbolic Battlefield

Public debate about sex work (性工作) in Chinese discourse is a particularly revealing case because it is almost never, on the surface, about sex workers. It is about the participants’ underlying positions on autonomy, male entitlement, female vulnerability, state authority, and the legitimacy of markets.

Officially, commercial sex is prohibited in mainland China. In practice, enforcement is uneven, geographically variable, and entangled with urban labour markets, migrant labour flows, neighbourhood policing priorities, and entertainment-industry politics. A sociology of Chinese sex work must therefore distinguish several layers: the criminal-law regime, the administrative-punishment regime (劳动教养 was formally abolished in 2013, but administrative detention of sex workers continues under other mechanisms), the local policing pattern, the social-stigma regime, and the migration-economy regime in which many sex workers operate.

These layers do not line up with the moral positions in online debate. Abolitionist-feminist arguments emphasise patriarchal domination, male entitlement, and the market’s capacity to normalise exploitation. Sex-positive and labour-oriented arguments emphasise worker agency, the harm done by stigma, and the evidence that criminalisation increases rather than decreases worker vulnerability. Both positions have serious scholarly and empirical support. Online, both positions are routinely flattened into caricature.

The sociological move is to notice that the intensity of the debate is not proportionate to the participants’ empirical engagement with sex workers. Most participants in these debates have never spoken with one. The debate is doing something else: it is a stand-in for underlying positions on gender, markets, state, and morality. That is why the sex-work argument is so often the argument that breaks a Chinese gender discussion — it forces underlying positions into the open in a form too stark for the usual face-work to contain.

Moral Panic, Properly Used

The concept of moral panic (道德恐慌), introduced by Stanley Cohen in his study of 1960s British responses to the Mods and Rockers, names a specific social pattern: a condition, episode, person, or group is defined as a threat to societal values; the threat is stylised in the media; experts, politicians, and respectable people pronounce diagnoses and solutions; the condition disappears, is dealt with, or becomes normalised; and the panic recedes. Cohen’s point is that the scale of the social response is disproportionate to the empirical threat.

The term is widely used, and widely misused, in contemporary gender discourse. Not every intense reaction to a gender issue is a panic. Widespread attention to workplace harassment after a well-publicised case is not a panic; it is a response to a real pattern whose scale had previously been underestimated. A wave of outrage after a specific incident of violence is not, by itself, a panic; it may be accurate proportional response. The diagnostic question for moral panic is whether the response scales with the empirical threat or with the symbolic mobilisation.

A useful contemporary Chinese case is the “男人集体性失格” discourse that periodically surfaces online — the claim that Chinese men have collectively become unmarriageable, undisciplined, unreliable, etc. The diagnostic question is whether this maps to any consistent empirical pattern in male conduct, or whether it scales with the affective requirements of a particular public. The scholarly answer is usually the second. That does not mean the grievances fuelling the discourse are unreal; it means the discourse itself is doing work that exceeds the grievances.

Historical Scripts Still in the Room

Sexual scripts are rarely invented from scratch. They inherit, layer, and mutate. Contemporary Chinese scripts about female chastity and male honour did not appear with platform capitalism. They are stratified sediments of late-imperial household Confucianism, Republican-era reformist feminism, Maoist gender-neutralisation rhetoric ("妇女能顶半边天"), post-Mao retraditionalisation, and post-reform consumer-era re-gendering. Each layer left residues. A script that says a respectable woman does not initiate sexual contact is not simply “tradition”; it is a particular arrangement that was disrupted by the Maoist gender-flattening period and re-established, in modified form, during the 1990s and 2000s, with assistance from a commercialised wedding industry and a commercialised marriage-market discourse.

This layered history is why contemporary Chinese sexual scripts can feel simultaneously “old” and “new.” They are in fact both. The demand that a bride be sexually inexperienced is genealogically old; the demand that a bride demonstrate an Instagrammable wedding aesthetic is genealogically new; the two demands run together in the same marriage-market discussion because the system that generates them is a composite inheritance. A reader looking for the “authentic Chinese tradition” under the noise is looking for something that does not exist as a clean object. There is only the composite.

The same layering applies on the male side. The script for respectable masculinity in mainland Chinese urban contexts is partly inherited from Confucian household responsibility, partly from the socialist-era cadre ideal, partly from post-reform entrepreneurial masculinity, and partly from the globalised professional-managerial model that the 1990s-2010s educational system cultivated. No contemporary urban Chinese man is only one of these. Every viable masculine performance is a negotiation among them. When online discourse talks about “what Chinese men are like,” it is almost always picking out one layer of this composite and treating it as the whole. Which layer is chosen depends on the rhetorical need, not the empirical population.

Respectability and the Sexual Double Standard

Cross-culturally, sexual double standards are remarkably persistent. The specific content varies, but the structural pattern — that the same sexual act has different respectability consequences for a man and a woman, usually higher costs for the woman — recurs in almost every documented society, and it has proved surprisingly resistant to the economic and educational changes of the last century.

In contemporary mainland Chinese urban contexts, the double standard is partly softened and partly relocated. It is softened in the sense that explicit public condemnation of unmarried women’s sexual activity is much rarer than a generation ago. It is relocated in the sense that the condemnation now operates through marriage-market signalling (a woman with a widely known previous serious relationship is framed as having “降低价值”) and through coded online speech (“前任多” as implicit disqualifier, “纯” as explicit preference). The double standard is quieter but has not disappeared; it has moved to the register where it can be maintained without being explicitly defended.

This is where 荡妇羞辱 (slut-shaming) as a category enters the discourse. The feminist import of the term has been to name a specific social mechanism — the differential sanctioning of female sexual conduct — so that it can be identified and challenged. The counter-import, on the other side, has been to denounce the feminist naming as its own form of virtue-signalling. The online argument typically proceeds through each side accusing the other of failing to understand what the term refers to, and both being partly right. What the student of the course should hold onto is that the phenomenon exists — differential sanctioning is empirically documented — and that the dispute over the label is a second-order politics about who gets to name it. Both the phenomenon and the naming-dispute are sociological objects. The naming-dispute does not settle whether the phenomenon exists.

Marriage, Divorce, and the Moral Regulation of Exit

A related site of moral regulation is divorce. Mainland Chinese divorce law underwent significant change in 2021 with the introduction of the 离婚冷静期 (divorce cooling-off period), a thirty-day mandatory waiting period between filing for an uncontested divorce and the divorce becoming final. The policy was explicitly framed in terms of reducing “impulsive divorce” and supporting family stability. Its effects have been contested: feminist critics argue that it disproportionately harms women in abusive relationships for whom exit speed is safety-relevant; supporters argue that it addresses a genuine social concern about the fragility of marriage.

Whatever one’s view of the specific policy, it is an instructive instance of moral regulation because it codifies a position on whether exit from marriage should be easy or hard, and whose interests count in that decision. The online debate about the 离婚冷静期 is therefore not really about thirty days. It is about whose relationship to the institution of marriage is treated as presumptively legitimate — the person who wants to stay, the person who wants to leave, or the state’s interest in marriage durability. All three are plausible candidates; the law can only prioritise one.

This is a recurring feature of moral regulation in the Chinese gender context: the state articulates a position, the position codifies a priority, and the discourse around the position becomes a proxy fight over which priorities the institutional apparatus should treat as default. The dispute is not over whether morality should be regulated; it is over whose morality should be regulated into default.

Chapter 6: Moral Psychology and the Limits of Argument

Although this is a sociology course, it cannot avoid moral psychology. Public antagonism is not sustained by incentives alone. It is sustained by how people process threat, belonging, and moral meaning, and those psychological mechanisms operate socially — they are shaped by group membership, media exposure, and institutional trust.

Jonathan Haidt’s social intuitionist model (社会直觉主义模型) is the simplest entry point. Haidt argues that moral judgement is intuitive first and justificatory second: people feel that something is wrong before they can say why, and then recruit reasons. This is not a claim that reasoning is irrelevant. It is a claim about the direction of flow. Reasons are often post-hoc narrations of a judgement already made.

In online gender discourse, the consequence is that a weak argument will remain persuasive to its own side if the underlying intuition is strong, and a strong argument will fail to move the other side if their underlying intuition is opposed. What reasons do is not change minds — they rarely do — but consolidate in-group coherence, provide ammunition, and demonstrate seriousness. A well-constructed post on misogyny or misandry is not trying to persuade the other side. It is equipping its own side.

Identity-Protective Cognition

Dan Kahan’s research on identity-protective cognition (身份保护性认知) pushes this further. People do not evaluate evidence in a vacuum. They interpret it in ways that protect identities they value. When a controversy becomes identity-linked, additional information often increases rather than decreases disagreement, because the better-informed partisans on each side become more skilled at defending their identity-congruent conclusions.

This is one of the most reliably depressing findings in contemporary social science, and it matters for gender discourse because gender discourse is unusually identity-saturated. An argument about whether a specific workplace policy is effective looks like a factual question; it is also a question about who the arguer is, what she thinks of feminism, what she thinks of men, and what she thinks of the legitimacy of state or corporate authority. More information does not settle it; it armours the disputants.

Kahan’s conclusion is not nihilism — he has written extensively on conditions under which identity-protective cognition relaxes — but the conditions are demanding: high perceived trustworthiness of the source, low salience of identity stakes, and presentation of evidence in identity-neutral form. Online gender discourse meets none of these conditions and is structurally designed to undo them.

Affective Polarisation

The last piece is affective polarisation (情感极化). Researchers distinguish between ideological polarisation — disagreement about policy — and affective polarisation — dislike, distrust, and moral contempt for the other side. The striking empirical finding in many Western contexts is that ideological polarisation has grown modestly while affective polarisation has grown sharply. People’s specific policy disagreements have not increased as much as their willingness to believe the other side is not only wrong but morally compromised.

The Chinese context is harder to measure cleanly, but the everyday experience of Chinese online gender discourse matches this pattern: participants differ on specific policies less than one would expect, but treat each other as belonging to morally opposed tribes. The affective gap is larger than the policy gap. This is not unique to China or to gender; it is a feature of polarised attention economies generally. But it is unusually legible in gender discourse, because gender discourse has been one of the earliest arenas in which the affective-polarisation mechanism could be observed clearly.

Social Identity Theory and Its Reach

The psychologist Henri Tajfel’s social identity theory (社会身份理论), developed in the 1970s, showed experimentally that people allocated to groups on essentially arbitrary criteria rapidly develop in-group preference, out-group suspicion, and a willingness to forgo personal advantage to produce advantage for the in-group over the out-group. The experiments — the original ones using colour preferences and abstract-art judgements as group-assignment criteria — demonstrated that the group-identification mechanism is activated by quite thin cues.

Online gender discourse provides thick rather than thin cues: sustained platform membership, shared vocabulary, shared grievances, shared rituals of posting and reposting. The group-identification mechanism is therefore not weakly activated but strongly so. The consequence, as Tajfel’s followers demonstrated in later work, is that out-group members are encountered less as diverse individuals and more as instances of a symbolic enemy. Platforms accelerate this by making out-group exemplars extraordinarily portable — a single screenshot of “what men are really like” or “what feminists are really like” can circulate through hundreds of thousands of in-group users in a day.

Once the mechanism is operating at scale, even otherwise careful thinkers have difficulty resisting its pull. It is easier to see a single out-group member as representative than to hold in mind the distribution of out-group views. The cognitive cost of proportional representation is high; the cognitive reward of symbolic enemy is immediate. This is not a failure of individual character; it is the cost structure of categorical thinking under adversarial informational conditions.

The Exit Problem

Putting these three pieces together — intuition-first judgement, identity-protective cognition, affective polarisation — gives us what we might call the exit problem. An individual participant who has internalised the in-group’s framing cannot cheaply leave it. Exit requires admitting that one’s previous vehemence was at least partly wrong, which is a face-cost in the in-group sense. It also requires accepting some of the out-group’s framing as partly right, which is a face-cost in the self-concept sense. And it usually means losing social connections built on shared contempt. The cost of exit is high enough that most participants do not pay it. They continue to perform a position they themselves have begun to doubt, because the performance is easier than the exit.

This is a useful thing to hold onto when reading public gender discourse. Many of the most vehement participants are not the most convinced. They are the most invested. The two are different. Convinced participants can update; invested participants cannot, because updating has a cost that convinced participants do not face.

Moral Outrage as Signal and Reward

A growing research literature on the microeconomics of online moral outrage — in particular, work by Molly Crockett, William Brady, and colleagues — has shown that expressions of moral outrage on social platforms generate reliable social rewards (likes, reshares, follower growth) and that users adjust their behaviour in response. This is an unusually clean empirical demonstration of a point that social theorists had argued on other grounds: that platform environments are not neutral with respect to the forms of speech they incubate, and that moral outrage specifically is favoured.

The mechanism is straightforward. Outrage expressions signal in-group alignment cheaply and unambiguously. They invite co-outrage (which is also rewarded). They are easy to perform even without much information. And they feel productive to the performer — they feel like doing something, even when they are doing only themselves.

For gender discourse, the consequence is that a platform environment with a moral-outrage reward gradient will over-produce outraged gender speech relative to the level of actual grievance. A community in which every grievance is amplifiable will appear to contain more grievance than it does. This is not a claim that the underlying grievances are fake. It is a claim about calibration: the aggregate scream of the discourse is louder than the aggregate grievance it expresses, because the medium boosts the loud signal without proportionally boosting the quiet one.

One of the practically useful exercises in this course is to ask, of a given gender controversy: is the public scream proportional to what the empirical record suggests is happening? Sometimes yes — and when it is, the scream is doing its democratic work. Sometimes no — and when it is not, the scream is doing commercial and affective work that is only metaphorically connected to the underlying reality. The two cases require different analyses and different responses.

Why Nuance Is Punished

A closing observation. One of the most reliable findings in experimental social psychology is that, in adversarial group settings, tentative or mixed-position speech is punished by both sides. The tentative speaker is read as weak, as not-fully-committed, as epistemically or morally suspect. This finding has been reproduced across domains — political debate, scientific controversy, religious argument — and it appears again in gender discourse with particular force.

The consequence is a visible supply problem. Nuance is not merely hard; it is socially costly. A participant who tries to say “well, both sides have a point, and the truth is probably X in this case and Y in that case” is often attacked by both sides more harshly than either side attacks the other. Over time, the supply of nuanced public speech falls, not because there are fewer people capable of nuance but because the ones capable of nuance have learned that nuance is expensive.

This is one of the most consequential and least discussed features of modern gender discourse. The visible positions are not a census of available views; they are the survivors of a selection process that punishes moderation. Reading the visible positions as “what Chinese people think about gender” is like reading the volunteer army as “who wants to fight” — it confuses willingness to be visible with representativeness of the population.

Chapter 7: Labels, Frames, and the Meme Economy

Online gender discourse does not begin as an open field of possible interpretations. It begins with available labels. A label is a compressed sociology: it names a type of person, attaches a causal story to the type, delivers a moral evaluation, and tells the reader what affect is appropriate. A reader encountering “普信男” does not need to be told what argument to bring. The word carries the argument.

Chinese internet vocabulary for gender conflict is unusually rich, partly because moderation pressure rewards code-switching and partly because the speed of platform circulation rewards short labels. A partial (and dated the moment it is written) lexicon includes:

• 普信男 — an ordinary man with unwarranted self-confidence about his desirability or importance.
• 舔狗 — a man who performs extreme subservience for a woman’s attention, often without reciprocation.
• 妈宝男 — a man emotionally dominated by his mother, unable to act independently in a marriage.
• 接盘侠 — a man who marries a woman with a previous serious relationship, framed as absorbing another man’s discard.
• 扶弟魔 / 伏弟魔 — a woman who impoverishes herself and/or her marriage to support her brother.
• 女拳 — “feminist fist,” used as a slur for any feminism perceived as aggressive.
• 田园女权 — “rural feminism,” used to accuse Chinese feminists of cherry-picking rights without duties.
• 男权 — “men’s rights,” used both by anti-feminists seriously and by feminists as critique.
• 雌竞 — female-on-female competition for male attention, framed as self-defeating.
• 慕强 — admiration of the strong, used both neutrally and as accusation.
• 海王 / 海后 — “sea king / sea queen,” a person who juggles many romantic partners.
• 直男癌 — “straight-man cancer,” a cluster of stereotyped chauvinistic attitudes.
• 厌女 / 厌男 — misogyny / misandry, used as both description and weapon.

Each of these started as diagnostic vocabulary for a specific pattern. Each has, over time, inflated. A term that originally picked out “a specific kind of condescending male self-regard” becomes a generalised insult applied to any man the speaker dislikes. This inflation is the master pattern of online label evolution, and the sociologically interesting question is why it happens so reliably.

The answer is that labels serve multiple functions that trade off against each other. A precise label is a good diagnostic tool. A general label is a good weapon. Precision and weaponisability are in tension. In low-stakes environments, precision wins; in high-stakes identity conflict, weaponisability wins. Over time, any label that is weaponisable will drift toward generality, and any label that stays precise will fall out of circulation because it is not weaponisable.

Framing, and Why Frames Are Exhaustive

A frame is a selection principle. It picks out some dimensions of a situation and suppresses others. Erving Goffman’s frame analysis and, in a more media-oriented form, Robert Entman’s work on framing both emphasise that the contest over an event is very often a contest over which frame is brought to bear.

Take a viral Chinese case about a woman who refused to marry her long-term partner unless his family bought a house in her name. The available frames include:

• structural-patriarchy frame: she is rationally responding to a system in which marital property law and social norms leave her vulnerable; the house in her name is insurance.
• female-opportunism frame: she is exploiting her partner’s commitment to extract wealth that his family worked for.
• housing-market frame: the dispute is an effect of property prices that no individual couple can solve; the gendered form is incidental.
• family-politics frame: the dispute is actually between her and his parents, mediated through him.
• platform-sensationalism frame: the original post was optimised for virality and misrepresents a more mundane real situation.
• generational frame: she and her partner grew up in a period of rapid marriage-market change; neither has a stable script.

Each frame captures something real. Each suppresses something real. Online escalation occurs when users treat one frame as exhaustive and any invocation of the others as bad faith. This is not a pathology of specific users; it is a structural feature of fast discourse. Nuance is expensive.

Meme Logic and Discursive Typification

The meme — understood broadly, including short-form video, captioned screenshots, recurring jokes, and quote-unquote templates — extends the label economy. A meme is not just a joke. It is a form of distributed classification. When a meme format circulates (say, “我男朋友说X, 我的反应:”), every use of it sorts the referent into a type. The particular boyfriend becomes instance-of a type; the particular reaction becomes instance-of a type; the particular grievance becomes instance-of a type.

Over time, a population exposed to these formats encounters its own intimate life through the typological grid they provide. The effect is discursive typification (话语类型化): people come to experience their own relationships and others’ relationships as instances of recognisable online types rather than as particular cases. When a woman thinks “oh, that’s a 妈宝男 thing he’s doing,” she is simultaneously diagnosing behaviour and placing it in a typified frame that comes with pre-scripted responses. The frame does work — it gives her a name for something she had experienced as confusing — but it also flattens. A label that classifies also forecloses.

The meme economy’s long-run effect on relationships is genuinely sociologically open. Some scholars argue it gives people — particularly women — vocabulary for patterns that used to be unnamed and therefore unaddressable. Others argue it over-typifies and makes particular people harder to see. Both are probably right depending on the case.

Asymmetric Evidence Politics

A final pattern worth flagging: asymmetric evidence politics (不对称证据政治). Online gender discourse runs on anecdote, but anecdote is valued asymmetrically. Stories from one’s own side become evidence of structural pattern; stories from the other side become exceptional, anomalous, fabricated, or staged. This pattern is not unique to Chinese publics, but it is unusually visible there because the volume of personal-confession content is so high. Every weekend brings a new batch of screenshotted chat logs. Which ones become “representative” is not randomly distributed.

A good diagnostic question for a reader of online gender discourse is: for each anecdote the post cites approvingly, could I construct a parallel anecdote from the other direction that would be dismissed? If yes — and the answer is almost always yes — then the post is doing asymmetric-evidence work.

The Genealogy of 普信男

To see how labels evolve, it is worth following one of them carefully. The term 普信男 entered mainstream circulation through a variety show called 脱口秀大会 in late 2020, when the stand-up comedian Yang Li delivered the line “他为什么看起来那么普通, 却可以那么自信?” about men’s self-perception. Within days the line had been screenshotted, turned into captions, and compressed into the three-character label that spread across Weibo, Xiaohongshu, and WeChat.

In its first month of wide circulation, the term was used in a relatively narrow sense: men whose self-regard exceeded what their observable characteristics would seem to warrant, particularly in romantic contexts. Within three months, it had expanded to cover any man a speaker wanted to dismiss as arrogant. Within a year, it had become a general-purpose gendered insult detached from its original diagnostic content. By 2022, some Chinese feminist writers had publicly distanced themselves from the term, noting that its generalisation had cost it whatever analytical value it had possessed.

This genealogy is instructive for several reasons. First, it shows how a single media moment can seed a lexicon item that persists long after the moment. Second, it shows the inflation trajectory: every viable label drifts toward generality because generality is weaponisable. Third, it shows the political consequences of inflation: the original analytical intent — naming a specific social pattern — is eroded, and the term becomes a rhetorical counter rather than a diagnostic tool. Fourth, it shows how quickly a label acquires its own politics: users of 普信男 become a faction; opponents of the term become another faction; the original question of whether there is a pattern worth naming gets lost in the factional dispute.

The Counter-Label Economy

Every label economy is also a counter-label economy. Online Chinese gender discourse is a sustained generation of labels and counter-labels. 女拳 answers feminism; 田园女权 answers feminism more specifically; 妇女之友 is ironic praise for a man who panders to women; 舔狗 is derogation for a man who subordinates himself for female attention; 接盘侠 is derogation for a man marrying a non-first-romantic-partner woman; 小仙女 is derogation for a woman framed as entitled. Each generates defences, reappropriations, and counter-deployments.

One general rule is that labels used by an out-group against an in-group are more susceptible to ironic reclamation than labels used by an in-group against itself. 女拳, used against Chinese feminists, has been partially reclaimed by some feminist users as a badge. 普信男, used against men, is less reclaimable because its derogatory edge is harder to invert. This asymmetry is not moral; it is structural, and it has to do with the different rhetorical topographies of being the insulter versus being the insulted.

A second general rule is that a label economy with many active labels is, in one sense, a sign of discursive richness — participants have distinctions available to them — and in another sense, a sign of discursive combat, because the labels exist because the fight exists. A discourse with no labels is a discourse nobody is fighting; a discourse with many labels is one with a sustained and varied fight. The multiplication of Chinese gender labels in the 2010s-2020s is accordingly a measure of how intensely the gender question has been contested in that period.

Chapter 8: The Chinese Case in Detail

Chapters 2 through 7 developed a set of conceptual tools. This chapter applies them to the Chinese case as a connected story, because the pieces only make sense when you see how they lock together.

The first thing to notice is that contemporary Chinese gender antagonism is not an ancient problem with a digital surface. Many of its sharpest forms date from the 2010s. The “直男癌” label popularised in the early 2010s. The “剩女” (“leftover woman”) discourse that the state and state-adjacent media promoted from the mid-2000s onward, which Leta Hong Fincher documents in detail. The first wave of Chinese online MeToo-adjacent disclosures in 2018. The post-2018 politicisation of gender discourse on Weibo. The rise of Xiaohongshu as a female-skewed platform whose cross-gender posts travel disproportionately. The Douyin/Kuaishou short-video boom after 2017. The hardening of marriage-market discourse as housing prices rose through the early 2020s and then became uncertain. Each of these is a specific event that changed the shape of the discourse, and the cumulative effect is the recognisable pattern of 2026.

Marriage Market as Master Arena

The single most determinative institutional setting in Chinese online gender discourse is the marriage market. Almost every conflict you read online is either literally about it or analogically about it. “婚恋市场” is not a metaphor; it is an actual infrastructure of matchmaking events, matchmaking corners in city parks, matchmaking apps, 相亲 (blind-date) introductions, parental networks, and increasingly algorithmic recommendation systems.

This market has several features that make it structurally combustible. First, it is age-stratified in extremely gendered ways: women are considered to “age out” faster than men, and the resulting time pressure falls asymmetrically. Second, it is closely tied to housing and assets, so financial capacity is not separable from romantic candidacy. Third, parental involvement is normative rather than anomalous: for most urban Chinese adults in their twenties and thirties, marriage is co-owned with parents in a way it generally is not in contemporary North American or European contexts. Fourth, the market is geographically variable: the ratio of marriage-age men to women, the regional income gradient, and the hukou politics mean that the “same” marriage decision is not the same decision in Beijing, in Chengdu, and in a county-level city in Henan.

The online discourse that treats this market is therefore not ideological overlay on a neutral institution. It is speech about a material system whose terms are actually imposed. When a woman online complains that “现在的男生只想白嫖,” she is speaking inside a system in which certain forms of pre-marital commitment extraction are widely practiced precisely because the post-marital legal environment gives her less protection than she would like. When a man online complains that “现在的女生只看房子,” he is speaking inside a system in which property has been the defining store of middle-class wealth for twenty years and is the most readable signal of long-term stability. Both are right about the system. They are wrong to think the system’s pathologies are the other side’s fault.

The Feminist Turn and Its Backlash

The 2010s also saw a distinct feminist turn in Chinese online discourse. The turn had several strands: state-feminist vocabulary from 全国妇联 (All-China Women’s Federation) and the legal system, which mobilised around domestic-violence legislation (2016) and gendered workplace law; academic feminist vocabulary imported and translated from North American and European feminist scholarship; and grassroots feminist vocabulary developed on Weibo and Douban, much of it sharper, more political, and more impatient than the academic version.

The backlash to this turn has been multiple. Part of it is organic: large numbers of male users reacting to perceived threat, assembling rough versions of familiar anti-feminist arguments (evolutionary-psychology-flavored claims, traditionalist claims, grievance claims). Part of it is state-adjacent: periodic crackdowns on prominent feminist accounts, particularly those that have crossed into political terrain that the censors are more sensitive about. Part of it is commercial: anti-feminist content reliably produces male-engagement numbers, and platforms and creators respond to that economic incentive independent of any political motivation.

The result is a public in which feminist positions are simultaneously more visible, more contested, and more strategically coded than they were a decade ago. A careful Chinese feminist writer in 2026 uses a different vocabulary than she did in 2016, not because her position has changed but because the coded-speech environment has.

State and Family

The state is a less mechanical actor in Chinese gender discourse than outside observers often imagine, and a less absent one than sophisticated Chinese users sometimes pretend. The state does suppress certain discussions: large-scale organised feminist mobilisation, discussions that connect gender to broader political structure, accounts that accumulate too large a female following and start resembling political organisation. It promotes other discussions: the “three children” fertility policy after 2021 spawned a wave of pro-natal state-media content, some of which reads as backlash to 1990s-2000s family-planning norms and some of which is simply economic-policy response to demographic anxiety. It tolerates interpersonal gender antagonism in many forms, because such antagonism is politically containable in a way that feminist organisation is not.

The family is in some ways more consequential than the state. The central reason is that most of the material disputes at the core of Chinese gender discourse — marriage, housing, care work, childbirth timing, eldercare — run through families. A young adult’s freedom to make a decision against parental wishes is real but costly. A decision made against parental wishes tends to produce consequences in the form of withdrawn support, inherited apartments not being bought on the children’s behalf, sustained conflict at family gatherings, and — in the long term — fraught eldercare arrangements. The discourse that imagines young Chinese adults as fully autonomous romantic agents does not describe the institutional reality they live in.

The Compressed Timeline

A specific feature of Chinese gender discourse is its compressed timeline. American gender discourse has unfolded, in its current form, over several decades: 1960s-1970s second-wave feminism, 1980s-1990s culture wars, 2000s online, 2010s social media, 2017 MeToo, post-MeToo fragmentation. Chinese gender discourse has unfolded largely in the 2010s and 2020s. The equivalent transitions have been compressed into about fifteen years, and the institutional context — a one-party state, explosive platform growth, the second-largest economy in the world simultaneously rising and slowing — has no parallel elsewhere.

One consequence is that Chinese gender discourse picks up from, reacts to, and remixes patterns from older Western gender discourse in compressed time. This is not evidence of derivativeness; it is evidence of network-speed cultural transmission. Another consequence is that Chinese participants often deploy analytical frames (intersectionality, masculinity studies, sex-positive feminism, radical feminism) that in Western discourse are distinguished by decades of internal argument, and they encounter them nearly simultaneously, which produces internal contradictions that Western discourse had time to process sequentially.

The Short History of the Last Fifteen Years

A short chronology helps anchor the claims in this chapter. The following timeline is selective and Chinese-internet-specific; a fuller historiography would need to include state-policy dates, demographic data, and international events that are omitted here.

2007. The All-China Women’s Federation publishes material using the term 剩女 applied to unmarried urban women over 27. The term subsequently circulates widely through commercial media in the late 2000s and early 2010s.

2010–2013. Rapid growth of Weibo as the dominant Chinese microblog platform. Early feminist accounts emerge and gain visibility. Simultaneous growth of commercial dating-advice content that codifies marriage-market logic.

2013–2015. Sina Weibo undergoes the first major wave of account suspensions and platform restructuring. The 女权之声 WeChat public account becomes a leading platform for Chinese feminist writing; it is repeatedly suspended and restored through the decade.

2014. Leta Hong Fincher’s Leftover Women is published in English. The book’s critique becomes a reference point for anglophone and (in translation) Chinese feminist analysis of state-media marriage discourse.

2015. Five Chinese feminist activists — later known as the 女权五姐妹 (Feminist Five) — are detained after planning a campaign against sexual harassment on public transport. The detention and international response are a high-visibility moment for Chinese feminist politics.

2015–2017. Rapid growth of Xiaohongshu from a niche overseas-shopping platform into a major lifestyle content platform, shifting toward a female-skewed user base and developing its distinctive lifestyle-feminism discourse.

2016. Passage of China’s Anti-Domestic Violence Law, a significant piece of gender legislation that establishes protection orders and makes domestic violence explicit grounds for state intervention. The law’s actual implementation has been widely critiqued as uneven.

2017. Launch of Douyin as a major short-video platform. Its algorithmic promotion system begins to shape the rhythm of viral content, including gender-coded content.

2018. The first major wave of Chinese MeToo disclosures, starting with academic cases at Beihang and Peking University, expanding into media, NGO, and tech sectors. The volume of disclosures is unprecedented; the moderation response is uneven.

2018, late. The 脱口秀大会 (Rock & Roast) stand-up programme broadcasts Yang Li’s 普信男 line; the term spreads rapidly.

2019–2020. Continued MeToo cases. Platform moderation tightens around specific feminist discussions, particularly those linking gender to broader political topics. Several prominent feminist accounts are suspended.

2021. The 三孩政策 (three-child policy) is announced in response to demographic concerns; state media launches a wave of pro-natal content that significantly reshapes the Chinese gender discourse environment. The 离婚冷静期 enters force. Zhou Xiaoxuan loses her initial suit against Zhu Jun.

2022. Housing-market correction accelerates. The Tangshan restaurant attack, in which male assailants assaulted female patrons in a widely-viewed video, triggers a nationwide conversation about gender-based violence and police response.

2023–2025. Consolidation of platform-based gender content as an established content genre; continued contestation between feminist, anti-feminist, and traditionalist positions; ongoing but uneven state moderation; gradually sharpening economic pressure as the housing correction and broader economic slowdown affect marriage-market dynamics.

2026. The present. The discourse continues.

This short chronology is not a substitute for a careful historical study; each entry could be a chapter on its own. But it is enough to anchor the fact that contemporary Chinese gender discourse has a specific, traceable recent history, and that its 2026 texture is the cumulative product of specific events, institutional responses, platform changes, and policy shifts that could have gone otherwise. Nothing about the current state is inevitable.

The 剩女 Campaign

One of the defining episodes in the emergence of contemporary Chinese gender discourse was the 剩女 (“leftover women”) discourse, documented in detail by Leta Hong Fincher in her 2014 book Leftover Women. The term appeared in official state-media discourse in 2007 on the website of the All-China Women’s Federation, which explicitly applied it to unmarried urban women over the age of 27. For the following decade the term circulated widely through both state-backed and commercial media, framing late-marriage women as social problems in need of correction.

Fincher’s argument, grounded in interviews and policy analysis, was that the 剩女 campaign had several interlocking purposes. It pressured educated urban women into early marriage, which facilitated the acquisition of housing in the husband’s name at a moment when property was becoming the main store of middle-class wealth. It reduced the pool of unmarried women available for sustained individual economic autonomy. It supported a broader demographic-policy project concerned with the declining marriage and birth rates. And it produced a public narrative in which women’s own failures of selectivity were responsible for their marital situation, rather than demographic imbalance or housing markets.

The campaign’s measurable effects included a documented decline in women’s property ownership rates during the 2000s-2010s (as mortgages and titles concentrated in husbands’ names), a rise in reports of urban women feeling under time pressure to marry, and a wave of Chinese feminist analysis that named the campaign and its mechanisms. The same feminist analysis was then subjected to backlash: critics argued that the feminist reading was itself ideologically motivated, that state concern about marriage was legitimate demographic policy, and that the critics were importing Western feminist frames inappropriately.

Whatever one’s position on the specific policy questions, the 剩女 episode is a clean case of the state, commercial media, and public discourse coordinating — not in a conspiratorial sense but in an institutionally aligned sense — to produce a discourse object. The object was then taken up, contested, redirected, and partially inverted by feminist writing, but the original framing persists in popular usage. This kind of layered history, where a state-originated discourse becomes a site of sustained feminist contest, is characteristic of contemporary Chinese gender politics.

MeToo in China

The Chinese MeToo wave in 2018 provides a second defining episode. After the initial Western MeToo cascade in late 2017, Chinese universities became the first major site: a number of female students and former students went public with allegations against professors at prominent universities, including Beihang, Peking University, and Sun Yat-sen University. The cases produced institutional responses (professors were dismissed or placed under investigation), a public conversation about harassment in Chinese academia, and a general expansion of the genre of disclosure-posting across Chinese platforms.

Over the following years, the wave hit tech (Ali, Jingdong), media (a Caixin journalist), arts (a cluster of allegations against male writers, filmmakers, and public intellectuals), and law. Each cluster produced institutional consequences that varied by sector, platform, and the specific dynamics of the case. A notable feature of the Chinese MeToo discourse was that moderation and censorship pressures were uneven and sometimes inconsistent: some cases were allowed to trend and produce public discussion, others were removed quickly, and the reasons for the asymmetry were rarely transparent.

The 2021 弦子 case — in which television host Zhou Xiaoxuan sued China Central Television host Zhu Jun over alleged sexual harassment, lost at trial, and appealed — became a particular focal point. The case was covered widely before trial, discussed less widely during trial, and became harder to discuss after an unsuccessful appeal. The trajectory is instructive: the case started as a MeToo episode and became, in its trajectory, a case about the limits of what kind of gender-based public discourse was sustainable.

A 2020s consequence of the MeToo wave has been the hardening of a distinct Chinese feminist vocabulary around harassment, consent, and institutional accountability, and a parallel hardening of a backlash vocabulary that treats all such discourse as ideologically imported and institutionally destabilising. The two vocabularies face off on every new case. Neither has disappeared, and neither has dominated.

The Housing Market as Precipitant

A final element of the Chinese case is the housing market. From roughly 2000 through 2020, Chinese urban housing prices rose at a pace that made first-tier-city property increasingly unattainable on salary alone. During this period, marriage-linked property acquisition became a central feature of middle-class family strategy: parents contributed to down payments on children’s behalf; property was acquired around marriage and recorded in the husband’s name (or, increasingly, both names) before children arrived; and marriage market discourse came to treat housing as a near-prerequisite for serious romantic consideration.

Starting in 2021, the housing market entered a prolonged correction. The consequences for gender discourse were real though not mechanically predictable. On one hand, falling prices eased some of the acute economic pressure that had made housing a marriage-market hurdle. On the other, the correction undermined the middle-class wealth position of many families who had borrowed heavily to acquire property, and it created new grievances about who had made the decision, who bore the loss, and how the household’s financial future looked. Gender discourse around housing after 2021 has accordingly been quieter in some registers (the shrill “房子是第一标准” discourse of 2018 had softened by 2024) and more complicated in others (resentments about who paid for what and whose name appears on what deed became, if anything, more legally contested).

The point for the student of this course is that the material economics of Chinese urban life is not background to gender discourse. It is foreground. A serious analysis of 2026 Chinese gender discourse requires paying attention to the housing correction as one of its conditions of production. Nothing about what online participants feel or say is independent of what the housing market is doing.

Chapter 9: Sex Work, Law, and Harm Reduction

Few issues in contemporary gender discourse condense as much pressure as the regulation of sex work. The topic will not settle normatively in this course, and it should not — the serious disagreement among feminist scholars on the question is both real and sociologically productive. But a number of analytical points are stable enough to be worth stating.

First, the term itself. Sex work (性工作) emerged in English in the late 1970s, promoted by activists and researchers to shift attention from moral stigma to labour conditions. Critics of the term, including many abolitionist feminists, argue that it sanitises coercion and obscures that the conditions of entry are often unchosen. Supporters argue that it allows clearer analysis of working conditions, legal rights, and state violence, and that the label “prostitute” or “娼妓” is already loaded against the subject. The choice of term is not innocent. A careful writer notes which term she uses and why, rather than pretending to neutrality.

Second, the four broad policy regimes. Comparative scholarship identifies roughly four models:

• Full criminalisation (全面刑事化): both buying and selling sex are criminal; the mainland Chinese model is a version of this, though enforcement is uneven.
• Legalisation (合法化与管制): sex work is legal and regulated by the state, as in Germany or the Netherlands; licensing, taxation, and health regulation are central.
• Decriminalisation (非刑事化): selling sex is not criminal, and in some versions neither is buying; labour law rather than criminal law governs the sector; New Zealand since 2003 is the main example.
• Nordic / end-demand (北欧模式 / 需求端打击): selling sex is not criminal, but buying sex is; the premise is that sellers are victims and buyers perpetrators; Sweden since 1999 is the original model.

Each of these redistributes risk, power, and visibility differently. Full criminalisation tends to drive the labour underground and concentrate power in police and intermediaries. Legalisation creates two-tier systems in which workers compliant with regulation are better protected and those outside it are not. Decriminalisation requires welfare-state and anti-violence infrastructure to be effective. Nordic models seek to reduce demand while protecting sellers, but empirical evaluations are contested and depend on whether you measure reported prostitution, street visibility, client arrests, worker reports, or trafficking indicators.

Sociology evaluates these regimes not by moral label but by effects on violence, stigma, bargaining power, worker health, policing patterns, and the lived options available to workers. This is what is usually meant by a harm-reduction (减害) perspective. Even if one morally opposes commodified sex, one can still oppose legal arrangements that intensify worker vulnerability. These two positions are compatible; failing to distinguish them produces a lot of online confusion.

The Chinese Complication

Mainland Chinese sex work sits inside a particular configuration. Formally, the activity is prohibited and punishable. Practically, it is widely present in specific industries (KTV, certain hotels, certain online services), is entangled with migration flows (women moving from less-developed provinces to coastal cities), and is subject to periodic enforcement campaigns that concentrate harm on the workers rather than the clients. Administrative detention of sex workers — the “custody and education” (收容教育) system — existed until it was formally abolished in 2019 after years of critique. Post-2019 administrative handling continues under different mechanisms.

Public discourse about Chinese sex work operates largely independent of these institutional realities. The debate online is almost entirely conducted between non-sex-workers about what sex work means for them — for their understanding of female autonomy, male entitlement, state power, moral order. Actual sex workers are almost never present in the discussion, and when they are, they are typically treated either as abstract victims or as ideological counterexamples. This absence is not accidental. The combination of legal risk and social stigma makes sustained public participation costly, and the participants in a public debate are not a representative sample of the population the debate is about.

One useful corrective is to read ethnographic work. Tiantian Zheng’s Red Lights (2009) on KTV hostesses in Dalian, Elaine Jeffreys’ China, Sex and Prostitution (2004), and more recent work on online escort markets each supply concrete detail that recenters the worker. The point is not to settle normative disagreement but to ensure that the argument is about the people it claims to be about.

The Argument About the Argument

A recurring sociological observation about sex-work discourse is that it functions as a terminal question: when a gender discussion is going to break on a deep disagreement, it will usually break on this topic. The reason is that every underlying commitment — autonomy, structural power, state legitimacy, market morality, feminism’s internal debates — forces a position on sex work, and the positions are not easily compatible. A careful discussion can hold several of these commitments in partial tension. An online discussion rarely can.

This is why moderators in online gender spaces sometimes ban sex-work discussion entirely. It is not because the topic is unimportant. It is because it is the topic most likely to produce the affect-dominated, identity-locked conflict that other chapters of this course have described. Banning the topic does not resolve the underlying disagreement; it is simply a move to prevent a space from tearing itself apart.

The Worker, the Feminist, the Abolitionist, the Client

A useful analytic move is to separate four positions that any serious discussion of sex work has to distinguish, because flattening them produces the confusion that online debate runs on.

The sex worker position is empirically varied. Some sex workers identify the work as labour they would choose under a wider range of options, and want legal and social conditions that reduce violence and stigma. Others identify it as work they are doing under duress and want exit support, welfare protection, and criminal-justice interventions against traffickers and abusive clients. Still others occupy positions in between. Ethnographies of Chinese KTV hostesses, of street-level sex workers in major cities, of online escort providers, and of migrant workers in the commercial sex economy have documented all of these positions, often within the same person’s account over time. Any policy analysis that assumes a single sex-worker view is wrong about the empirical population.

The sex-positive feminist position emphasises worker agency, worker-led movements, and the observation that criminalisation disproportionately harms the vulnerable. Its analytical frame treats sex work as labour and its regulation as a labour-rights question. This position is strong on worker protection but is sometimes accused of underweighting the proportion of sex work that is produced by coercion, constrained choice, or limited exit options.

The abolitionist feminist position — associated with thinkers from Catharine MacKinnon through a wide contemporary literature — emphasises male entitlement, the structural continuity between purchased sex and other forms of male sexual access, and the argument that the commodification of women’s sexuality undermines gender equality broadly. Its analytical frame treats sex work as a harm produced by male demand and its regulation as a gender-equality question. This position is strong on demand-side analysis but is sometimes accused of overriding the testimony of sex workers who identify as labour actors.

The client position is the least studied and most morally charged. It includes men who buy sex, which is a substantial portion of the adult male population across most societies with available data, Chinese and otherwise. The client is morally legible in the abolitionist frame as the agent of the transaction. In the labour frame, the client is more ambiguous. A serious sociology of sex work eventually has to account for clients empirically: who they are, what they think they are doing, and how their behaviour responds to different regulatory regimes. The current online Chinese debate largely does not attempt this, which is one of its limitations.

Holding these four positions distinct allows a discussion to proceed without collapsing into factional identity conflict. A participant can adopt partial elements of each without being incoherent: one can, for instance, believe that worker-led decriminalisation is generally better than criminalisation, that the commercial sex economy is a gendered harm whose reduction would be good, and that neither of these positions is straightforwardly contradicted by the other. Most careful feminist writers in practice hold some such combination. Online debate, however, usually demands that one pick a team, which distorts what any individual writer would say in longer form.

The Licensing Question

One of the less-discussed facts about the regulation of sex work is that every regime licences something. Full criminalisation licenses police discretion, intermediary violence, and administrative detention. Legalisation licenses the state’s taxation of the activity and the creation of a two-tier system between licensed and unlicensed workers. Decriminalisation licenses labour-law frameworks and (where available) health-system access. Nordic models licence end-demand enforcement and, in their stronger versions, a therapeutic-apparatus framing of sellers. No regime licenses nothing; every regime licenses something.

The question is therefore not whether to regulate but what one is willing to let the regulatory apparatus do. Feminist and sociological debate about sex work becomes more productive when it is framed at this level, because the comparison is between concrete licensing packages rather than between idealised versions. The concrete packages have empirical costs and benefits that the idealised versions do not.

For Chinese policy analysis specifically, the interesting question is not “should we move to regime X” — such a move is not on any near-term policy table — but rather “which elements of which regimes are already present in Chinese practice, and which are plausible to strengthen or weaken administratively?” Harm-reduction elements (access to health services without arrest, decriminalisation of sex workers’ safety-protection activities, training of police against worker-directed violence) have been implemented in limited ways in some Chinese cities. Each such change moves the lived regime toward or away from specific regulatory packages. A policy conversation grounded in these particulars would look very different from the online debate.

Chapter 10: Comparative Cases

Comparison is not decoration. It is a control for the kind of civilisational explanation — “this happens because Chinese culture is X” — that online discourse is especially prone to. If similar patterns appear elsewhere, then the civilisational explanation is doing less work than the proponents believe. If the patterns differ, then the institutional conditions make a specifiable difference. Either way, comparison constrains speculation.

United States

The United States is the most visible comparison because its gender discourse is the most heavily exported. Three clusters are useful to keep in view.

The manosphere (男性圈) — a loose network of anti-feminist, men’s-rights, pick-up-artist, MGTOW, and red-pill communities, documented in scholarship since at least the early 2010s — arose in US internet cultures in the 2000s, escalated through the Gamergate moment in 2014, produced a number of specific recognisable figures (Jordan Peterson, Andrew Tate, various red-pill influencers), and by the 2020s had become a commercialised influencer economy with branded products, paid coaching, and social-media followings in the millions. Many of its tropes have circulated into Chinese discourse in translated form.

Call-out feminism, a looser category, names the online practice of public accountability-by-posting, particularly visible after 2014 and especially after 2017’s MeToo. It has produced real institutional consequences (ouster of individual abusers, policy changes at specific companies and universities) and real public ambivalence about its excesses (mistaken targets, disproportionate consequences for minor incidents, public-shaming dynamics that exceed the original complaint). The genre travels: a Chinese MeToo wave in 2018-2019 adopted many of its forms, though under a much more restrictive speech environment.

Tradwife discourse — women posting aspirational content about domestic femininity, often with explicit rejection of feminist frames — is a US-originating influencer phenomenon that has also found Chinese variants, though the Chinese versions are shaped more by housing and dual-career pressure than by the religious conservatism that organises parts of the US version.

South Korea

South Korea is, for many sociologists, the clearest case of gender discourse becoming explicit political identity. The Ilbe / Megalia cycle in the mid-2010s — male grievance forums and female counter-grievance forums generating reciprocal radicalisation — was followed by the entry of gender into partisan politics (particularly in the 2022 presidential election, in which the People Power Party explicitly courted young male anti-feminist voters and the Democratic Party was perceived as feminist-aligned). The youth demographic split along gender lines, which in comparative terms is unusually sharp, has been the subject of substantial scholarship.

The South Korean case is instructive because it shows the end-state of fully politicised gender discourse: gender does not stop being about intimate life, but it also becomes a core axis of electoral alignment. Whether Chinese discourse will evolve in this direction is an open question. The institutional context differs — electoral politics does not function as an outlet — but the affective patterns are recognisable.

Japan

Japan presents a different configuration. Rather than loud ideological confrontation, Japanese gender discourse has historically expressed strain through disinvestment: the 草食系男子 (herbivore men) discourse of the 2000s, the low fertility rate, the large and growing single population, and the documented decline in dating and sex among young people. The anxieties that produce shouting in China and South Korea produce quiet withdrawal in Japan. The explanation for this difference is not cultural essentialism; it has to do with labour-market structure, housing-market structure, and the specific history of Japanese gender politics since the bubble economy of the early 1990s.

The Japanese case is useful for a Chinese student because it shows one possible trajectory for a society in which marriage and family formation become expensive and uncertain: the politics gets quiet, because the underlying material institutions make the political fight feel pointless. Whether Chinese discourse moves toward the loud Korean trajectory or the quiet Japanese trajectory will depend partly on whether and how the underlying economic conditions resolve.

What Comparison Teaches

Across these cases, three recurring patterns are worth naming. First, where educational expansion outpaces stable mobility — particularly for young men — gender antagonism intensifies. Second, where housing and family formation are costly relative to income, the intimate-life politics heats up. Third, where platforms reward antagonism, the public face of gender discourse exceeds the actual prevalence of the positions it voices. These are not Chinese features. They are features of modern platformised economies under the specific conjunction of post-1990 globalisation and 2010s recommender systems. China is one instance, not the archetype.

Comparison also forecloses some explanations. A story that says “Chinese women have become X” or “Chinese men have become Y” does worse at explaining the phenomenon than a story that says “platformised gender discourse in any economy with these conditions produces X and Y patterns, and Chinese users are doing what users in comparable conditions also do.” The civilisational explanation is often a compliment to one’s own culture disguised as analysis. Comparative work removes that compliment.

A Closer Look at the Korean Case

The South Korean case deserves a more detailed reading because it illustrates, better than any other, what explicit gender politicisation looks like in a democratic context, and what the Chinese case might look like if the affective patterns continue to strengthen without finding electoral outlet.

The timeline goes roughly: in the mid-2010s, the anti-feminist forum Ilbe (일베) emerged as a high-intensity space for young male grievance, racism, and sexual politics. Its counterpart Megalia (메갈리아), and later Womad (워마드), mirrored its tactics in the opposite direction — radical feminist and, in the case of Womad, explicitly misandrist — and generated reciprocal radicalisation with Ilbe and adjacent male communities. By the late 2010s, the language of these forums had spread into mainstream South Korean discourse.

The 2022 presidential election crystallised the politicisation. People Power Party candidate Yoon Suk-yeol explicitly courted young male anti-feminist voters with a platform that included promises to abolish the Ministry of Gender Equality and Family. Exit-polling showed an unprecedented gender gap in the 20s cohort: young men swung heavily toward the PPP, young women heavily toward the Democratic Party. The election was narrowly won by Yoon. The subsequent political debate has continued to treat gender as a core partisan axis.

Scholars including Kim Jieun, Kim Bo-mi, and Jieun Kim have documented the institutional conditions: extraordinarily competitive labour markets for young men (military conscription is two years; young women are accordingly seen as having a head start in employment), a declining marriage and birth rate with intense national policy attention, and a commercial media environment that has profited enormously from gender-discourse content. The Korean case is not a special pathology; it is the result of these institutional conditions operating under democratic political competition.

For China, the Korean case is instructive for what it suggests about trajectory. The institutional conditions are partly parallel — competitive labour markets, expensive housing, a marriage-rate decline, a politicised media environment — but the democratic outlet is absent. This means that the affective energy that in Korea was channelled through electoral politics has, in China, nowhere to go except deeper into the discourse itself. Whether the absence of an electoral outlet makes the discourse worse or better is an open question.

A Closer Look at the US Case

The US case has been over-covered in some respects and under-covered in others. What is worth highlighting is the commercial infrastructure that has grown around gender discourse. Andrew Tate’s content empire before his arrest in late 2022 generated, by contemporary reporting, millions of dollars per month. The broader manosphere influencer economy is a multi-hundred-million-dollar annual market. The tradwife content economy generates comparable totals on the opposite side. Feminist influencer content, call-out accounts, and podcast networks operate at smaller but non-trivial scale.

This commercial infrastructure means that gender antagonism in the US is not simply an organic cultural phenomenon. It is a market with producers, distribution platforms, monetisation pipelines, and an ecosystem of ancillary services (coaching, courses, merchandise). The producers respond to commercial incentives in the same way any content producer does. A creator whose content becomes moderated becomes harder to monetise and moves to less-moderated platforms. The result is a distributed ecosystem across Twitter/X, YouTube, TikTok, Substack, Rumble, Telegram, and a long tail of smaller platforms, with creators moving across them in response to moderation and monetisation gradients.

China’s commercial infrastructure for gender content is smaller but growing, particularly on Xiaohongshu (for female-skewed content) and Douyin (for both). The economics differ from the US in that direct monetisation (advertising, subscriptions) is smaller, and platform-economy monetisation (product placement, live-streaming commerce) is larger. This means that Chinese gender-discourse creators have somewhat different incentives than their American counterparts, but the fundamental logic — that attention is salable and that antagonism produces attention — is the same.

What Recurring Patterns Tell Us

Across the US, Korean, Japanese, and Chinese cases, several invariants hold. First, where young men perceive their economic and social position to be declining relative to young women’s, high-intensity anti-feminist content finds an audience. Second, where commercial platforms reward engagement, engagement-generating gender content proliferates. Third, where the state cannot or will not intervene directly in discourse shaping, market logic dominates. Fourth, where any of these conditions is absent or altered, the texture of gender discourse changes in predictable ways — quieter under Japanese disinvestment, electorally channeled under Korean democracy, commercially saturated under US platform capitalism, coded and state-filtered under Chinese conditions.

The takeaway is not that Chinese gender discourse is just another case. It is that Chinese gender discourse is a specific articulation of patterns that are recognisably modern, platformised, and cross-nationally comparable. Understanding the articulation requires both attention to what is locally specific (the hukou system, the housing market, the state moderation apparatus, the particular historical layering of scripts) and attention to what is globally shared (attention-maximising platforms, post-industrial marriage-market anxieties, the economic disentanglement of male provider scripts from economic viability). A good analysis holds both in view.

Chapter 11: Methods for Studying This

If a student of this course were to write a research paper on some aspect of Chinese online gender discourse, how should she proceed? There is no single correct method. The object is multi-layered, and different methods capture different layers.

Content analysis (内容分析) codes recurring themes, labels, and narrative structures across large samples of posts. Its strength is scale. Its weakness is that it cannot tell you how users actually interpret what they read, or what the algorithmic exposure environment was. A content analysis of Weibo gender discourse that does not account for recommendation-system exposure is measuring what was posted, not what was received.

Digital ethnography (数字民族志) observes communities over time, tracks in-group slang, humour, sanctions, and identity performances. Its strength is interpretive depth. Its weaknesses are access, the researcher’s positionality (gender ethnography in particular is rarely done from a neutral standpoint), and the ethics of observing semi-public communities. Chinese digital ethnography has additional ethical complications around deanonymisation risk for participants.

Survey research (问卷调查) can measure declared attitudes, experiences of harassment, and exposure to specific content. Its strength is population-level measurement. Its weaknesses are self-report bias, question-wording effects, and the gap between declared attitude and observed interaction. In the Chinese context, survey access is additionally constrained by institutional gatekeeping and by the politicisation of certain questions.

In-depth interviews (深度访谈) can illuminate how users make sense of gender discourse in relation to biography, work, family, and intimacy. Their strength is depth of interpretation. Their weakness is that interviewees narrate rather than transparently reveal, and the narrative form itself is shaped by the cultural scripts under study.

Platform studies (平台研究) analyse the recommendation systems, moderation regimes, and interface designs themselves. Access to Chinese platform internals is limited for researchers outside the companies, but reverse-engineering approaches, policy-document analysis, and developer-interview work are productive. This is an under-developed method in Chinese gender scholarship, partly for access reasons.

Most strong studies combine methods. A typical design might pair content analysis of a corpus of posts with interviews of users who produced or engaged with that corpus. The pairing lets the researcher move between what was said, what was meant, and what was received.

The Researcher’s Position

One ethical point specific to this field: the researcher is never outside the discourse she studies. A Chinese female graduate student researching Chinese online misogyny is the subject of some of the discourse she is analysing; a male student researching online feminism occupies an asymmetric position in the communities he observes. Reflexivity is not decoration here. It is part of the methodological apparatus. Who you are affects what communities will speak to you, what you are able to hear, and what you will miss.

There is also a moral point about amplification. Some research on hateful discourse inadvertently amplifies the content it studies, either by training models on it or by quoting extensively from figures whose fame is partly a function of academic attention. The standard of care — what scholars sometimes call 研究伦理 — is to write about patterns rather than to platform particular bad actors unnecessarily, and to think explicitly about the effect of one’s own publication on the phenomenon.

A Worked Example: Studying a Single Controversy

To make these methodological points concrete, consider how a researcher might study a single online gender controversy — say, the weeklong trajectory of a viral Xiaohongshu post about a marriage proposal.

A content-analysis approach would collect the original post, the first wave of comments, the second-wave responses on other platforms, and perhaps a sample of the tertiary commentary. She would code each post for frame, tone, label usage, emotional register, and positional alignment. She might produce a frequency table of labels deployed, a network graph of how commenters engaged each other, and an analysis of which responses travelled.

A digital ethnography approach would complement this by observing the communities that engaged with the post. The researcher might spend weeks following specific Xiaohongshu accounts, Weibo subcommunities, and WeChat groups, noting how the controversy was discussed internally, what vocabulary was used, and how in-group norms shaped interpretation. This work cannot be done retrospectively in the same depth; it has to happen in real time.

An interview approach would recruit participants who produced or engaged with the controversy and ask them to reconstruct their reasoning. What did they think when they saw the post? Why did they comment? What did they mean by specific terms they used? Did they change their mind over the week, and why or why not? The interviews would usually reveal a much more complex reasoning process than the posts suggest. Users who look factional in public often reason more ambiguously in private.

A platform-studies approach would attempt to reconstruct the recommendation trajectory: which users saw the post first, how the algorithm routed it, what engagement signals pushed it forward or suppressed it. Access to this data is limited for outside researchers, but inferential approaches — examining when views spiked, which adjacent content was being promoted, and what moderation interventions visibly happened — can recover a partial picture.

None of these approaches alone captures the controversy. Together, they produce a multi-dimensional account: what was said, how the communities read it, what the reasoning under the saying was, and what the platform machinery did with it. This is the kind of account the field aspires to, and that a careful student of the course should be able to recognise and evaluate when she encounters published work.

Quantitative Methods and Their Limits

A separate methodological remark concerns quantitative methods. Large-scale natural-language-processing studies of gender discourse — sentiment analysis, topic modelling, network-influence analysis — have become more feasible with contemporary tools. They produce impressive aggregate pictures and can identify patterns invisible at the scale of individual reading.

But they have specific limitations for this domain. Sentiment analysis does badly on irony, coded speech, and in-group humour, all of which are characteristic of Chinese gender discourse. Topic modelling groups content by surface lexical features and often misses the structural similarities that the content shares with content using different vocabulary. Network analysis produces influence maps that measure retransmission but not interpretation. Each of these methods is useful for what it does and misleading when taken as a final answer.

The careful use of quantitative methods is therefore diagnostic rather than conclusive. They can flag patterns that require interpretive attention. They rarely, on their own, settle a sociological question. A study that reports that “women’s posts have 20% higher emotional-arousal scores than men’s posts” is reporting a fact about the specific corpus, the specific NLP model, and the specific operationalisation of “emotional arousal.” Whether it says anything about gender depends on a great deal of interpretive work that the number by itself cannot do.

Chapter 12: Public Reason and the Limits of Optimism

The course does not end with a solution. Online gender conflict reflects real injuries, structural inequalities, and incompatible interests in some domains. Sociology cannot convert all disagreement into misunderstanding, and it should not try.

But the course also does not end with resignation. Conflict is socially organised, and its form and intensity are therefore socially alterable. The question is not whether gender disagreement can be eliminated — it cannot — but what institutional arrangements could reduce the specific kinds of cruelty, injustice, and symbolic escalation that current arrangements produce.

Several levers are visible. Platforms could reduce frictionless amplification of humiliation content (specifically, small changes to repost mechanics and to ranking of pile-on replies have measurable effects). Educational institutions could teach media literacy that includes frame analysis and not just fact-checking. Legal institutions could address harassment, workplace discrimination, and domestic violence more consistently, which would reduce some of the material grievances that fuel discourse antagonism. Welfare and housing policies could reduce the material desperation that translates, through the marriage market, into online gender blame. Family policies could redistribute care work — public childcare, real paternity leave, eldercare infrastructure — which would reduce the intimate-life disputes that become public.

None of these alone resolves anything. All of them, seriously implemented, would change the substrate on which the discourse runs. This is what it means to say that sociology cannot solve the problem but can specify what would reduce it. The specification is not a programme; it is a map of levers.

The Remaining Hard Question

There is one question the course has circled but not answered directly, and it is worth stating clearly at the end. It is the question of whether serious public reasoning about gender is possible on platformised media at all.

Public reason (公共理性), as used by political philosophers, does not mean bloodless neutrality. It means argument that distinguishes empirical claims from symbolic insults, structural analysis from totalised blame, and policy evaluation from identity warfare. On its most demanding formulations, it requires participants to offer reasons that other reasonable participants could, in principle, accept.

Nothing in the current architecture of attention-maximising platforms supports this. Everything in that architecture — the virality metrics, the affective reward structure, the identity-sorting recommender systems, the face-threat density — actively opposes it. A participant who tries to reason publicly in this sense is, under current conditions, selecting herself out of visibility. The participants who remain visible are the ones whose style is compatible with the medium, and the compatible style is not the reasoning style.

This is not a counsel of despair. Public reasoning about gender continues to happen — in classrooms, in long-form writing, in small-scale communities, in book clubs, in slow podcasts, in serious journalism. It happens offline more often than online. What platformisation has done is not eliminate public reason but move it out of the venues platformisation dominates. Recognising that is part of reading the current situation honestly.

The course began with the PHIL 145C exchange. It has insisted, through twelve chapters, that the exchange is not best read as an argument. But the reader of these notes is now doing something closer to reading it as an argument — slowly, away from the thread, with commentary, in a format that allows revision. The course is, in that sense, already an instance of the thing it has been circling around. The sociology of online gender conflict is not only an object of study; it is also an implicit argument for the forms of reading and writing in which that study is possible. Those forms are rarer than they used to be. They are not obsolete. They are worth the trouble.

That is the final claim of the course, and the one the reader is invited to disagree with. It is certainly not universally held. But it is what motivates every chapter here: the belief that understanding why a discourse makes honest thought difficult is, itself, one of the things that makes honest thought possible.

What a Better Discourse Would Actually Look Like

It is worth, at the close of the course, saying concretely what a better discourse would look like, because abstract invocation of “public reason” is easy and concrete specification is hard.

A better Chinese online gender discourse would begin from the recognition that the marriage market is a structural problem, not a moral one. Housing, care infrastructure, and labour-market conditions would be the primary objects of discussion, not personal character. Bride-price arguments would be arguments about regional development and inheritance law, not about whether Chinese women are mercenary. Masculinity-crisis arguments would be arguments about labour-market restructuring and educational sorting, not about whether Chinese men have gone soft.

A better discourse would distinguish between empirically testable claims and identity-positioning claims. “Young Chinese men are experiencing economic decline relative to young women” is an empirical claim with a specifiable truth-value, and the current evidence on it is mixed and regional. “Young Chinese men have become entitled” is an identity-positioning claim with no truth-value in the same sense. Treating the two as equivalent is the kind of error a better discourse would not make.

A better discourse would treat labels as tools rather than as weapons. A label like 妈宝男 might return to its original diagnostic function — naming a specific relational pattern — rather than serving as a generalised insult. A label like 田园女权 might be rejected entirely by users who notice that its work is almost entirely rhetorical. A discourse in which labels were used carefully would be slower, less exciting, and considerably more useful.

A better discourse would treat ideological opponents as potential sources of partial correction. The radical-feminist critique of commercialised sex work catches something real; the labour-feminist critique of abolitionist approaches catches something real; either position alone is worse than a careful synthesis. Discourse environments that punish synthesis as “centrism” or “bothsidesism” are not doing intellectual work; they are doing political work whose cost is intellectual erosion.

A better discourse would be slower. It would accept that the urgent rhythms of platform circulation are incompatible with the kinds of reading and thinking that gender questions demand. It would therefore produce its best work in venues — long essays, books, sustained classroom discussion, serious podcasts — that are structurally slower than the platforms. It would treat the platforms as distribution surfaces rather than as primary venues.

What to Do When the Discourse Is Bad

Given that the current discourse is, in many respects, worse than this specification, what should a serious student do?

First, protect time. The hours spent reading online gender discourse do not translate proportionally into understanding. Most participants of the discourse are not gaining understanding; they are maintaining affective alignment. A student of the subject should budget her engagement accordingly and read widely outside the discourse.

Second, protect registers. Maintain the capacity to think in registers other than the ones the discourse incentivises. This is genuinely harder than it sounds. The discourse trains its participants in a specific idiom; sustained immersion in the idiom changes how one thinks. Regularly reading in other idioms — academic work, non-Chinese sources on comparable topics, literary or historical treatments of gender — is a deliberate cognitive hygiene measure.

Third, protect relationships. Gender discourse has a notable capacity to poison specific interpersonal relationships — with family members, with partners, with friends. Some of this is productive: it forces difficult conversations that needed to happen. But much of it is not: it imports public positions into private relationships where the positions do not fit the actual history and texture of the relationship. A discipline of keeping public-discourse idioms out of specific relationships — not avoiding the topics, but refusing to import the rhetorical moves — is often a worthwhile practice.

Fourth, accept that the discourse is not going to improve through individual effort. The structural conditions that produce its current form are not going to change because a few people behave better. The appropriate response to this is neither despair nor withdrawal but calibration: invest in the forms of attention and conversation that actually produce understanding, and treat the platform discourse as the weather rather than as the conversation.

The Student at the End of the Course

A student who has read this course to the end has done something that the ambient discourse does not reward: she has sustained a long reading on a topic that is usually read in fragments. She has encountered arguments she may disagree with, framed in a register that demands she understand them before she dismiss them. She has spent time on mechanisms and institutions rather than on individuals and characters. She has acquired a vocabulary — 性别秩序, 霸权男性气质, 算法放大, 情感公众, 话语类型化, 身份保护性认知 — that is strange to the ambient discourse and useful for thinking with.

None of this makes her smarter than the discourse participants who did not take the course. It makes her equipped to read the discourse with a different kind of attention. Whether she uses that equipment to continue engaging the discourse, to withdraw from it selectively, or to write about it in her own way is her choice. The course has no view on that. It has a view only on what the equipment is for.

The equipment is for holding together things the discourse holds apart: individual biography and institutional pattern, immediate affect and long historical process, local specificity and cross-national comparison, serious disagreement and serious engagement. These are not fashionable skills in any period, and they are particularly unfashionable in platformised ones. They are worth acquiring anyway, because the alternative is to be read by the discourse rather than to read it.

That is the last claim. The rest is up to the reader.