Suffer If You Don't Work? From CSSA to UBI, Whose Principle Is “More Work, More Pay”?
@Mythogen Engine · Oct 20, 2026
“Snatching Rice from the Beggar's Bowl”, Part 1 | Previous: Prologue | Next: “Global Top, Local Bottom”
Not resenting the rich, not hating the poor: struggle is not competition, and in the end it reaches every floor.
2036: a self-service kiosk
One morning in 2036, Mr Chan is queuing at a self-service kiosk in Tuen Mun town centre.
He is fifty-two and spent twenty years as a logistics clerk. Three years ago his company installed an AI scheduling system, and his department went from twelve people to one. The one was not him.
It is his turn. He presses his palm to the sensor and the screen lights up: “Survival points balance: 1,284. Redeemable this week: rice, canned food, basic clothing, utility allowance.”
He wants to exchange points for a second-hand tablet for his daughter's online classes.
A line appears on the screen: “This item is not a basic necessity. The design principle of this scheme is to secure survival, not to provide enjoyment of life.”
Mr Chan freezes. He knows that line. Ten years ago he left a comment under a video about welfare: “Hand out canned food, that's enough. Don't let them live too comfortably.” The comment got more than three hundred likes.
At the time, he assumed he would always be the one doing the liking.
1. Four months, three kinds of unemployment
Mr Chan is fictional. But every rule he meets at the kiosk was already proposed by someone in 2026.
In 2026, Sun Channel's Lorey Flash Talk released three videos within four months, all about unemployment, yet they gave unemployment three different moralities.
The June episode was about a 42-year-old middle-class Hong Kong father: out of work for five months, his monthly pay cut from HK$50,000–60,000 to HK$20,000, and no one would hire him even for a junior role. The host said he too was 42 and a father, and found it heartbreaking. He called such people “structural NEETs”, noted that the new gains from AI had been harvested by the top elite of tech giants, which also exploit loopholes in global tax rules, and praised Germany for using a strong state to rein in capital and put workers first.
The first October episode took up the case of a girl carrying beer in Fu Tai Estate, Tuen Mun, and ended with welfare reform: cash assistance (CSSA, Hong Kong's main welfare payment) for able-bodied adults should have a time limit, after which it is frozen and replaced with food, clothing, utility allowances and vouchers — a deliberately designed subsidy with “no enjoyment of life whatsoever”.
A week later, the second episode, on AI and universal basic income (UBI), warned that future UBI would no longer be paid in cash but in “survival points” exchangeable for food, healthcare and electricity, which the issuer could cut according to a score, or stop altogether. The video called it tech giants' “most perfect electronic shackles”.
Rations in kind instead of cash, eligibility decided by the issuer, revocable at any time. The two October episodes describe the same design: the first calls it reform, the second calls it techno-feudalism.
Put the three together: middle-class unemployment is a structural problem deserving sympathy; working-class unemployment is a character problem requiring a tighter grip; and once AI may put everyone out of work, the same tightening design becomes techno-feudalism. Unemployment itself has not changed; what changes is which class the unemployed belong to.
Mr Chan at the start could be that father ten years on: pitied in June, squeezed in October, and by 2036 standing in front of a kiosk.
To avoid confusion, “the first episode” and “the second episode” below refer to the two October videos; the June video is called “the June episode”.
This essay questions the premise that makes this shift possible: that more work, more pay equals worth, and that people who don't work should live worse. Let us start with the first episode.
2. One girl, two minutes
A ten-year-old girl in Fu Tai Estate, Tuen Mun, was filmed by a passer-by carrying two large bags of beer weighing about 12 kg. Her stepfather later said it was to “train her large and small muscles”. After the clip spread, police first classified the case as a “misunderstanding”, then passed it to the child abuse investigation unit, and arrested the couple on suspicion of child abuse or neglect; both were released on bail pending investigation, and the three siblings were admitted to hospital for assessment by the Social Welfare Department. The case remains unresolved.
Children helping carry things is perfectly normal. Plenty of parents have asked a ten-year-old to carry a pack or two of soft drinks or beer; sometimes the child obliges, sometimes they sulk and refuse to carry anything. Had the bags held soft drinks, the clip might never have been filmed, and even if it had, it would hardly have become a welfare case.
What made it a welfare case was the beer. The beer was for the mother, and the family was on welfare — put those two facts side by side, and without another word of comment, viewers conclude “welfare pays for drinking”. Between a child-abuse case and a welfare case there was only a beer can.
That does not mean nothing was wrong in the clip. The problem is not the beer but the weight, and whether she could refuse. The girl later said she had carried six four-packs of 500 ml cans that day, 24 cans, about 12 kg, the most she had ever carried, and admitted it was hard; before, the most had been four packs. A child that age typically weighs a little over 30 kg, so she was carrying nearly a third of her body weight, while the Department of Health recommends a schoolbag weigh no more than a tenth. When passers-by offered help she refused, saying she was afraid her father would scold her, and that she often helped carry beer. Her biological father said the elder daughter was timid and dared not resist, so the burden of chores fell on her; her younger sister was bolder and would refuse.
A child who sulks has room to refuse. The real question is whether this girl had any.
Then came two commentary videos: the first episode of Lorey Flash Talk, and Chiu Sin-hin's Chiu's Reading Life, with more than 170,000 views between them. They share one thing — the girl takes up only the first minute or two. The next ten-plus minutes are about welfare, “lying flat”, taxpayers and welfare reform.
A suspected child-abuse case became a welfare case within two minutes. The girl became an entrance: viewers came in through her, then discussed why their own lives were so hard.
The first task of this essay is to step back through that entrance and separate the problems that were melted into one.
3. What the two videos said
Lorey's argument rests on a principle from 1834. The “principle of less eligibility” in England's New Poor Law required that the able-bodied who depended on relief must live below the standard of the lowest-paid independent labourer.
On this basis he proposed that able-bodied adults could receive cash welfare for only a set number of years in a lifetime; after that, if their pension and tax records were still zero, the cash would be frozen at once and replaced with vouchers for food, clothing, utilities and low-speed phone cards. In his words, the subsidy should be designed to have “no enjoyment of life whatsoever”. The sick, the disabled, elderly people living alone and single-parent carers would be exempt from the limit.
Chiu Sin-hin borrowed the “Samaritan's dilemma” proposed by the economist James Buchanan in the 1970s: helping once is a good deed, but if recipients expect help to continue, their behaviour changes and the safety net becomes a way of life.
He went further, arguing that social workers and medical staff, out of sympathy and professional interest, tend to help people apply for benefits. He therefore proposed that a neutral third party assess disability claims, that unemployment assistance last only three months to a year, and that rent subsidies be set five to ten per cent below average rents.
The details of the two plans differ, but the premise is the same: if a person is able to work and does not, their life should be worse than that of people who work. That premise is what this essay sets out to examine.
4. The facts are thinner than the anger
The figures cited in the videos and comment sections contradict each other. On maintenance payments, Chiu said HK$2,000 a month; commenters said HK$6,000, and some, quoting a TV programme, said HK$20,000. Under a court order, the biological father must pay HK$2,000 a month for his two daughters combined, but he admitted to the media that he was more than HK$30,000 in arrears (HK01). The “double income” the family was accused of had not even been fully received.
As for the HK$18,000 a month, Chiu said it was welfare plus disability allowance; but commenters had already pointed out that the two cannot be claimed together, and that disabled people simply receive a higher standard rate within CSSA. The Disneyland gold card and several trips a year came from “netizens' revelations”; Chiu himself said the family never said where they went, and it might just have been the Greater Bay Area.
More importantly, CSSA is means-tested by household, and maintenance counts as income and is deducted from the assistance. If this family really were receiving substantial maintenance while drawing full CSSA, that would be a problem of declaration and checking, which is a different matter from whether the system is generous. The first needs investigation, the second needs data, and neither can be deduced from a street clip.
As for the family itself, the mother told reporters she has a chronic illness and is quitting drinking, and the children have special educational needs (HK01); when the stepfather left the police station, he was holding a summary of an Education Bureau educational psychology assessment (HK01). These claims are unverified. But the videos' portrait of the family as “able-bodied and choosing to lie flat” is equally unverified — and the videos did not even mention this side.
On the broader picture of “welfare feeding the lazy”, statistics can say only so much. A 2018 LegCo Research Office summary showed elderly cases making up about 62% of all CSSA cases, with single-parent and ill-health cases about 10% each (LegCo Research Office, 2018); in 2015, unemployment cases were only about 7% (LegCo Research Office, 2016). Even when unemployment cases surged past 18,000 during the 2020 pandemic, elderly cases still numbered more than 135,000 (HK01). These figures show that CSSA mainly serves the elderly, but they measure case categories, not whether declarations are honest.
An approved case is, statistically, an eligible case. Undeclared income that goes undetected appears in no number. Many Hong Kong people know families like this: property on the mainland, cash income, yet still drawing CSSA in Hong Kong. This lived impression does not contradict the official statistics — what statistics cannot see is exactly where checks cannot reach.
CSSA's means test rests on self-declaration, and the Social Welfare Department has difficulty independently and fully verifying assets and income outside Hong Kong. Ironically, the government itself uses the cross-border gap in living costs: the Guangdong and Fujian Scheme lets eligible elderly CSSA recipients move to the two provinces and keep receiving assistance. Chiu's suggestion that moving welfare households to the Greater Bay Area could save half has already been done officially for the elderly; while some families have run the arbitrage in the opposite direction on their own — assets and income kept on the mainland, assistance drawn in Hong Kong.
So concealment is a real problem, but it is a gap in the checking system, not evidence that “welfare is too generous”, and it certainly cannot be applied to everyone on CSSA. Lumping it together with “able-bodied and choosing to lie flat” only drowns the cross-border checking problem that really needs solving in anger at an entire group.
This also means at least four things must be handled separately: whether the child was abused, whether the family concealed income, whether welfare weakens the will to work, and whether means-testing has failed. The two videos melted the four into one.
5. Who the 1834 principle was really for
The principle of less eligibility is often presented as fairness: those who work should not live worse than those who don't. But its real function in 1834 was as a wage mechanism.
The New Poor Law largely abolished outdoor relief: the able-bodied who wanted help had to enter the workhouse. Workhouses were deliberately made more miserable than the worst job: families separated, monotonous food, hard labour. The aim was plain — to make the poor accept any wage the market offered rather than go in. The Andover workhouse scandal of 1845 revealed inmates so hungry they fought over the marrow on bones, which shows where the principle's logic ends. A hundred and eighty years later, society is more civilised; there is no need to make people gnaw bones, so vouchers are handed out instead.
In other words, the principle of less eligibility protects not workers but the fact that workers have no way out. If not working means misery, a person has no standing to refuse a bad job. The tighter survival is tied to work, the greater employers' bargaining power.
Lorey's subsidy with “no enjoyment of life whatsoever” is exactly a modern workhouse: no walls needed, just make life dull and joyless. A week later, the same design reappeared in the second episode under a new name: electronic shackles. The “more work, more pay” the two hosts speak of looks like fairness on the surface; in effect it builds a safeguard for low wages.
6. Hong Kong has no unemployment insurance
The same suggestion keeps appearing in the comments under both videos: US unemployment benefits last only six months, and Hong Kong should follow.
The comparison mixes up its objects. US unemployment benefits are insurance, funded by payroll contributions while employed, with eligibility and amount based on past work; after they run out, there is a separate relief system. Hong Kong has no such system. When someone loses their job, their Mandatory Provident Fund cannot be withdrawn early on grounds of unemployment, and the only safety net is CSSA.
Taiwan is a clear contrast. Taiwan has employment insurance: those who lose their jobs involuntarily can claim unemployment benefits of 60% of their average insured monthly salary over the previous six months, for up to six months, or nine months for those over 45 or with disabilities (Ministry of Labor). Unemployment is caught first by insurance, so social assistance only has to deal with genuine poverty: at the end of 2023, low-income and middle-to-low-income households together numbered about 564,000 people, about 2.4% of the population (Ministry of Health and Welfare). Hong Kong puts both into the same CSSA.
As a result, Hong Kong squeezes “temporarily unemployed” and “permanently unable to support oneself” into a single system. Two different uses collide inside one scheme, and what onlookers see is “welfare abuse”.
The time limit that Lorey says “is practised abroad” is modelled on the lifetime 60-month limit of the US TANF programme from the 1996 welfare reform. Caseloads did fall sharply after the reform, but not all of those who left had escaped poverty; some had simply been kept out. The share of poor families covered by TANF kept falling, and some research found deep poverty actually increased. What a time limit reduces is the number of cases on the government's books, not necessarily poverty itself.
7. Britain has already tried it
Most of Chiu Sin-hin's proposals have in fact been tried in Britain.
Third-party assessment: from 2008, Britain outsourced work capability assessments to Atos and later to private firms such as Maximus. The result was large numbers of wrongful refusals and high rates of decisions overturned on appeal; Atos left its contract early in 2014. One commenter put it in a sentence: third parties have KPIs too, just as private prisons in the US tend to keep incarceration high. Handing assessment to an agency without sympathy brings not neutrality but a different set of incentives. Outsource sympathy, and what comes back is KPIs.
Rent subsidies: Britain's Local Housing Allowance is already set at the 30th percentile of local rents, well below the average. The “government renting people three-bedroom houses in Cambridge” that he worries about is already capped under current rules.
As for his figure that “one in every 5.1 working-age people claims Universal Credit”, the number itself deserves a closer look. Universal Credit merged several older benefits, and a large share of claimants are in work — they have jobs, but earn so little that they still qualify.
What this number measures is not how many people don't work, but how many people work and still cannot make ends meet. That is exactly the problem neither video faces.
8. The same premise, moved to Paris
After this essay was written, large-scale protests broke out in France. They began with secondary school students: from mid-September 2026 they blockaded schools, protesting teacher shortages and run-down buildings, and university students joined on 1 October (Easy France Now). On 17 October, students joined crowds protesting fuel prices and a budget that froze benefits and civil service pay, and the protests spread nationwide (Bernama citing Franceinfo).
On 19 October, Chiu Sin-hin made a twenty-minute video about it. In the first three minutes he is clear: the students' demands are reasonable, and at first half of France supported them; teachers are short because leave is too easy to take, wedding and maternity leave are long, many after the pandemic took psychiatric leave on full pay, and the suburbs cannot find substitute teachers because talent flows to Paris and Lyon.
This is the same diagnosis as Fu Tai Estate. In Tuen Mun, the problem is welfare recipients getting money without working; in France, it is teachers getting paid while on leave. On pensions, he says the French “are used to a very elegant life”, “living off grandpa”. The subject changes from a Hong Kong child-abuse case to a French student protest, but the cause is the same: someone gets fed without working.
This diagnosis does not survive checking either. He says France's retirement age is rising from 62 to 64 and “will be 64 by next year”; in fact, in November 2025 the National Assembly voted to suspend the 2023 pension reform until after the 2027 presidential election, freezing the retirement age at 62 years and nine months (JURIST). His prescription is to use tax rebates to encourage private schools to take middle-class students, saying France has not used a public-private approach; but France already does exactly that: about 13% of primary pupils and about 21% of secondary pupils attend private schools under contract, the share of pupils from advantaged families in private contract lower-secondary schools is more than twenty percentage points higher than in public ones (French Ministry of Education, DEPP), and public funding for these schools is about €9 billion a year (IPP). The plan he proposes France has used for decades; its side effect is precisely the social segregation he says the left worries about.
Then the video takes a third step. After teacher shortages and deficits, it turns to the far left and far right, a survey of Muslims, and finally lands on the Western Jin official Jiang Tong's Treatise on Relocating the Rong: foreign peoples settled in the heartland will, if not dealt with, eventually rise up. The video's title also cites the treatise. To be fair, he himself favours a middle path — since they are here, help migrants integrate; but none of the causes he lists in the first three minutes — the leave system, suburbs that cannot keep staff, tight finances — has anything to do with immigration, yet the title leads readers to immigration.
This is exactly how the Fu Tai Estate videos worked. A ten-year-old girl became a welfare case after two minutes; a group of secondary students asking for teachers became, after ten minutes, a question of whether foreigners would rise up. Different entrances, same exit. One commenter saw through it: in the end, these all seem to be questions of public finance and distribution, and attention should not be only on left-right ideology, refugees or religion. Another asked more directly: if the problem is pensions, how would deporting migrants help pensions or the deficit?
How those migrants first came to France's suburbs is discussed in section 12.
A day later, on 20 October, Chiu made another video and stated his position more plainly. He said the right stands for “carrots and sticks”: carrots for the rich and middle class, so they work hard and pay tax; sticks for the poor, cutting welfare so they work hard, with welfare only for emergencies and not a long-term way of life.
This is a modern plain-language version of the 1834 principle, with one difference. Carrot and stick were originally a way of driving the same donkey: a carrot dangling in front, a stick behind, both applied to one donkey. Give the carrot to one group and the stick to another, and it is no longer an incentive but stratification.
In the same video he talked about school lunches. He argued that schools should use uniform ingredients and need not provide different menus for religion; if someone says the poor cannot afford to bring their own lunch, his answer is: that's your choice — you can keep your faith, or accept free food. Rich children need not choose; poor children must choose between their faith and a meal. This is the same thinking as the vouchers in the first episode: food is not just food but leverage. He also lumped food allergies together with religion, saying that accommodating everything costs too much. But peanut and seafood allergies are medical matters, which schools already handle individually with a doctor's certificate, and have nothing to do with religion.
France in fact already has a simpler approach. Since November 2019, every school canteen in France must offer at least one vegetarian meal a week (French National Assembly); as for pork-free alternatives, France's Council of State ruled in 2020 that whether to offer them is for local authorities to decide, but offering them does not breach secularism (AFP). One vegetarian meal already caters to those who don't eat pork, beef or meat at all, without tailoring menus religion by religion.
Factory canteens solved this long ago. I have seen many factories in mainland China and Taiwan, and their canteens all use the same system: employees queue with a card and tap it, take a bit more today, eat vegetarian tomorrow, add a bowl of noodles after, take whatever they like. The card works only in the canteen, so the money always becomes food; what to take is up to them. No one has to declare a religion, and no one has to choose between faith and a meal.
The only difference is how much goes in. With the same card, in some mainland factory canteens the food is barely edible; in some Taiwanese car-factory canteens, chicken chops and sausages are laid out in a row like a night market. Same system, different budget. Protection is the same: the same framework can provide bare subsistence, or let people eat with dignity — a point taken up again in section 14.
Ironically, the video opens by repeating Trump's claim that France's problem is “not an economic problem, it's an immigration problem”; yet later in the video, the causes Chiu himself lists are underfunded schools, no substitute teachers, and government spending at 57% of GDP. One commenter hit the nail on the head: the most frightening thing is when fiscal and management problems are solved by political means; no one treats the actual illness, and instead they dodge it to talk about ideology and religion. Problems of too few teachers and too little budget are recast as problems of immigrants and faith — just as the fiscal pressure on welfare was recast as the character of one girl's family.
9. What the comment sections are really angry about
Back to the two Fu Tai Estate videos. Read the most popular comments one by one, and the targets of anger turn out not to be only welfare families.
The elderly HK$2 transport fare concession cut to 20% of the fare; students' annual HK$2,500 subsidy cut; Legislative Council seats increased from 60 to 90; civil servants getting a pay rise. One parent said she was born and raised here and has never stopped working, yet may not earn HK$18,000 a month, and her children go to Disneyland only when relatives pay. Another said many working-class families live on one person's HK$18,000 a month, with HK$15,000 left after rent.
What these comments share is that their own share is being cut. Under fiscal austerity, welfare families are the only people in the whole chain who are both visible and unable to fight back, so the anger flows there. The anger is real, but it compares the two lowest groups with each other, not with where the money is actually going.
Some commenters saw further. One pointed out that rich neighbourhoods need a great deal of low-paid labour that cannot be outsourced — waiters, security guards, cleaners, carers — and that public housing nearby amounts to a stable, close-by labour pool that holds down businesses' labour costs. Scholars of Hong Kong housing policy have long pointed out that during the years of economic take-off, public housing served as a “social wage”: the government bore housing costs, so employers could pay lower wages for a long time. Seen this way, it is not only public-housing tenants who are subsidised, but also those who employ them.
The working class itself understands this function of public housing well. The June episode classed “deliberately keeping income low to apply for public housing” as a strategy of active lying flat, calling it “get public housing, get the world”. A commenter did the sums for the host: a 200 sq ft flat rents for HK$9,000 to HK$12,000 a month; a full-time job paying HK$25,000 leaves half after rent, and little after utilities and food. The host replied personally to agree. The so-called lying-flat strategy, worked through to the end, is just rent arithmetic.
Another viewer regretted saving for a down payment to buy a flat, working for the bank while also paying tax, rates and government rent. What he did not say is that Hong Kong's public finances depend heavily on land sales and land-related revenue, so much of the spending on welfare, public housing and healthcare is ultimately paid, through land prices, by those who buy property.
He and welfare families are not opposing sides, but two positions in the same land-based fiscal system: one pays the entry fee with thirty years of mortgage payments, the other gets a little back from the system.
All of the above concerns a minority: those considered “able to work but not working”. The next question is what the same principle becomes when automation makes this “minority” ever larger, and where the extra share from technological progress has actually gone.
10. A hundred workers, swapped for ten engineers
Even looking only at those “able to work but not working”, the premise is shaky.
Automation is often described as “a hundred workers reduced to ten”, as if ninety lose their jobs and ten stay. In reality, all hundred production jobs often disappear, replaced by ten jobs for automation, control and maintenance engineers, filled by different people.
What disappears is not just a number of jobs but a whole career path — from operator to senior technician, foreman, supervisor. Ten years on a production line will not turn anyone into a control systems engineer.
The productivity gains are real: higher output, lower unit costs, steadier quality. But the gains flow to the company, shareholders, newly hired engineers and, when prices fall, consumers. The workers replaced own none of the relevant assets and have no mechanism for sharing profits. Technology solved efficiency but did not answer distribution.
It is not only machines that take jobs away. People used to say Americans were lazy, so production went overseas. But no matter how hard an American worker works, he cannot compete with a place where wages are a fraction, even a few per cent, of his. Factories move not because someone refuses to work, but because the same goods are cheaper to make elsewhere. Hard work can narrow the efficiency gap but not the wage gap. By that logic, American workers were lazy for forty years and Chinese workers hard-working for twenty — until Vietnamese workers were even more hard-working.
“Willingness to work” thus becomes a convenient explanation: it recasts the results of a cost structure as a group's character flaw. “Lying flat” in the welfare debate and “lazy Americans” back then are the same kind of claim — a structural problem rewritten as personal character, which shifts responsibility from those who make the decisions to those who bear the results.
A structural contradiction follows: the production system needs a certain group's labour less and less, yet the income and protection systems still require them to prove through work that they deserve to live. Say “those on welfare must not live better than the working poor” at this point, and the question becomes: if the working poor are themselves sliding downward, this floor will sink with them all the way down.
Work has two meanings: producing goods and services, and serving as modern society's entry point for distributing income, status and eligibility for benefits. Automation has weakened the first, while the second remains untouched.
A firefighter with no fires to fight is not without value; a person who puts in eight hours a day does not necessarily create more value than someone without a job who cares for their family. Labour can create value, but labour itself is not value.
Push this reasoning to its end and you get the script of Lorey's second episode.
11. When “no work” becomes most people's future
The second episode cites former Google X chief business officer Mo Gawdat's interview on The Diary of a CEO, projecting that within ten to twelve years AI will replace jobs all the way up the ladder, from entry level to middle class to CEO. Once the labour side collapses, the consumer side disappears with it: however much and however cheaply companies produce, no one can afford to buy. To prevent unrest, governments can only cooperate with tech giants, tax them, and introduce UBI.
Part of this projection holds: productivity and purchasing power can come apart. The hundred workers in the previous section are a miniature of that split — output rises while income flows elsewhere. The only difference is scale. With a hundred workers, onlookers can still say that group failed to adapt; when the same thing climbs the ladder, there are no onlookers left.
But “AI will replace all jobs” itself deserves caution. It need not actually replace everyone to have an effect; the claim “AI will replace you” is already enough to justify layoffs, wage cuts and anxiety-driven courses. In this chain, AI is not the cause but the licence. The video takes the claim as a premise and builds fear on top of it.
Then comes what section 1 mentioned. The video warns that UBI will eventually be paid not in cash but in survival points for food, healthcare and electricity; combined with something like a social credit score, wrong opinions cost points, and falling below a threshold cuts off food. This is the same design as the first episode's welfare plan: rations in kind instead of cash, eligibility decided by the issuer, revocable at any time. Applied to the minority who are “able-bodied but choose to lie flat”, it is called fairness; applied to “us, after AI has put most people out of work”, it is called shackles.
Whether a system of protection turns into control does not depend on whether it is called CSSA or UBI, but on a few things: whether the issuer can revoke eligibility at will, whether recipients can use what they receive freely, and whether there is independent appeal and accountability. By these standards, the welfare plan proposed in the first episode falls squarely on the side of shackles. The future the second episode fears, the first episode had already written the specification for, complete with the list of exemptions.
Such a specification is acceptable on one condition only: that the speaker assumes he will always be on the giving side, or at least the taxpaying side. When automation turns “no work” from a minority's situation into the majority's future, the same principle changes from a tool for punishing others into a threat hanging over one's own head.
The June episode had in fact already given a different answer. There, the unemployed were pushed out by structure: gains harvested by the top, tech giants avoiding tax, and the state should rein in capital as Germany does. Four months later, facing a family on welfare in Tuen Mun, the same host switched back to the principle of 1834. He can see the distribution problem; it is just that when the unemployed don't look like him, the explanation switches from structure to character.
In the June episode's comments, one remark got nearly a hundred likes: when one person doesn't work, it is a personal problem; when a large group doesn't work, it is a social problem. The logic of the first episode is precisely to break a large group back down into individuals.
The shackles need not wait ten years either. Controlling a person does not require cutting off their food; make receiving assistance shameful, and they will put up with worse jobs and keep quiet. That is exactly what the first episode does. In the second episode's comments, one viewer put it more bluntly: if you don't use AI, you'll be eliminated and lose your job immediately. What chains people today is not points but work itself — the tighter survival is tied to work, the less need there is to forge extra chains.
12. Imported labour and welfare are two ends of one lever
Two demands from the employers' camp often appear together: tighten welfare, import labour. They look unrelated but aim at the same goal — weakening workers' ability to walk away.
The most-liked comment on the June episode was a single line: Hong Kong's unemployment problem is clearly more about imported labour than AI. Nearly three hundred people liked it. Compared with distant AI, imported workers are a visible competitor. But this anger is just as easily misdirected — it is never imported workers themselves who decide to import labour.
A welfare floor gives local workers an exit option: they can turn down bad jobs, so employers must raise wages to hire. Importing labour bypasses that option: if locals won't do it, find people with no choice. Without foreign migrant workers, wages in some of Taiwan's industries ought to rise; they don't, because labour can be imported. As of February 2026, Taiwan had more than 860,000 industrial and care migrant workers, with figures published monthly by nationality by the Ministry of Labor.
Taiwan has a deeper layer still. Live-in domestic caregivers are not covered by the Labor Standards Act and have long been paid below the minimum wage. They not only hold down wages but directly replace long-term care services the state ought to provide.
Hong Kong's foreign domestic helpers are the same: they keep dual-income families running, and spare the government from providing childcare and elderly care at scale. What the welfare state does not do is outsourced to a group with no membership and no benefits. Many of the angry dual-income parents in the comments are able to cope only because of this arrangement.
Here one must be careful with talk of “imported poverty”, or it easily slides into the logic of “new immigrants cheat welfare” — exactly the downward comparison this essay criticises.
The French suburbs mentioned in section 8 are the clearest example: after the war, factories were short of workers, and France actively recruited labour from North and West Africa; then came deindustrialisation, the factories left, the people stayed, and poverty in the banlieues followed. What was imported was not poverty but labour; poverty was manufactured after the work disappeared. Capital enjoyed cheap labour and, when it left, left the cost to the welfare system. This is the same story as the hundred workers, with a border added.
Hong Kong's post-1997 One-Way Permit migrants are mostly family reunions, the result of working-class Hong Kong men marrying across the border in the 1980s and 1990s. They are not outside poor, but part of Hong Kong's own class structure, separated by a border. One-Way Permits are approved by the mainland, so Hong Kong does not fully control its own membership — one reason the related anger is especially strong. How the population's points of entry work, and why they cannot be measured, is the subject of Part 4 of this series.
13. Global in principle, national in practice
In the end, the problem is unequal distribution. But at what level should things be made equal?
The pressure of globalisation is real, but on whom it finally falls and how hard depends on how each country distributes (for the global figures, see Part 2, “Global Top, Local Bottom”). Hong Kong's manufacturing workers are a case in point: many of those hundred workers' jobs were not taken by machines but moved to Dongguan and Vietnam.
In principle, equality should be global. But the political philosopher Thomas Nagel argues that distributive justice requires a sovereign able to tax, legislate and enforce; there is no such body globally, so between countries only a duty of humanitarian aid remains. In practice, only states can distribute.
Accept the state as the only realistic unit, and the next question is: has the state used all its capacity? Distribution at the national level needs at least three conditions — control over the tax base, control over membership, and decision-makers accountable to those being distributed to.
Capital can move, but taxes stay within borders, so countries compete to cut taxes on capital. What states can reliably collect is only what cannot leave: labour, consumption and land, so the money for redistribution at home is increasingly paid by workers themselves. The anger that “taxpayers support welfare” is therefore not baseless — it is just that the party that should share the burden has long since left the tax base.
The second episode's whole script rests on this step: the government taxes tech giants and pays out UBI. But if states are cutting even ordinary capital taxes against each other, taxing the companies best at moving capital and profits is exactly the hardest step in the script. The video imagines the giants growing powerful and only then bargaining with governments; in reality the bargaining began long ago: companies choose where to report profits, governments choose whom to make up the shortfall from, and the answer is usually today's workers.
Hong Kong meets none of the three conditions fully, but not for lack of capacity. Low tax is an identity Hong Kong chose; relying on land prices for revenue shifts the cost of redistribution onto property buyers; and the government still holds fiscal reserves in the hundreds of billions. Income per head is among the highest in the world, yet the Gini coefficient is high among developed economies — the problem is not that the pie is too small, but how it is cut.
One remark by Chiu Sin-hin exposes this neatly. He said that if welfare households could move to the Greater Bay Area, HK$18,000 could become HK$9,000, saving taxpayers half, and only “one country, two systems” prevents it. The essence of the plan is to use the cross-border gap in living costs to export welfare households beyond the bounds of membership. When the topic is saving money, global inequality becomes a tool to exploit; when the topic is distribution, we are told it can only be calculated within the country.
14. Survival is the floor, not the destination
The discussion so far has argued against “those who don't work should be worse off”. But one question remains unanswered: beyond basic support, if a poor person is to live comfortably and with dignity, where does the money come from?
This was always the point of technological progress. As productivity rises, the same output needs less labour, and the surplus should let people live better, not just survive. The productivity line keeps climbing; wages just haven't risen to match. The problem is that the gains are not effectively distributed.
US data show it most clearly. The Economic Policy Institute's long-run tracking shows that from the end of the Second World War to the 1970s, productivity and typical workers' pay rose roughly together; after 1979 the two lines split: from 1979 to 2019, net productivity grew about 60% while a typical worker's pay grew only about 16% (EPI). EPI argues the split comes mainly from policy choices that weakened workers' bargaining power — deregulation, de-unionisation, erosion of the minimum wage, tax cuts for high earners — rather than from technology itself (EPI).
The same holds globally. A 2024 International Labour Organization report found that labour's share of global income has fallen 1.6 percentage points since 2004; had it stayed at the 2004 level, workers worldwide would have earned about $2.4 trillion more in 2024 alone. The report lists technological change such as automation among the main causes, and warns that generative AI may push the share down further (VOA citing the ILO).
Hong Kong's version is more direct. The Society for Community Organization notes that since the handover, GDP has grown substantially in real terms while the real median monthly household income has fallen rather than risen (SoCO). The government's Hong Kong Poverty Situation Report 2020 showed 1.653 million people in poverty before policy intervention, a poverty rate of 23.6%; and when measuring the effect of poverty alleviation, public housing alone, a non-cash benefit, “lifted” 265,000 people out of poverty (Hong Kong Catholic Commission for Labour Affairs). This is how the social wage from section 9 appears in official statistics: the shortfall in wages is made up by public housing, then counted as the government's anti-poverty achievement.
Look somewhere else. Imagine a young man from Hong Kong; call him Kelvin. In 2019 he went to Queensland on a working holiday visa to pick strawberries, and to renew for a second year he had to complete 88 days of specified work in a regional area. The farmer, Gary, treated him well and bought him beers after work.
In 2022, Gary bought a harvesting machine. That year his harvest hit a record and his labour costs fell sharply. Kelvin's visa expired, and he went back to Hong Kong.
Gary's good year will appear in Australia's agricultural productivity statistics; Kelvin's departure will not appear in any country's unemployment statistics. In the June episode's comments, someone said that after coming back from Australia, they spent two years as a NEET.
This is not an isolated story. Over the three pandemic years, labour on Australian horticultural farms fell by about a fifth, about 29,300 people, mostly overseas working holidaymakers; output actually rose about 3% over the same period as farms replaced people with machinery (Good Fruit & Vegetables citing ABARES). By 2022–23, about 40% of horticultural farms were using harvesters and other machinery to cut labour (ABARES).
The land belongs to the farmer. Automation raises output and cuts costs, and the gains stay with those who own the land and machines; those replaced are mainly casual hands and backpackers on working holidays. They own no land, have no membership, and leave when their visas expire. The same automation, in a factory, hurts a hundred workers; on a family farm, it benefits the farmer's family. The difference is not the technology but who owns the asset being automated.
This also answers “where does the money come from”. A decent life cannot depend only on relief after the fact; people need a share where the gains are produced. There are roughly four routes: wages rising with productivity; public services holding down the cost of housing, healthcare, education and transport; letting residents own part of the productive assets and share the returns; and taxing land and other rents, rather than taxing only labour that cannot leave.
There was also an older route, which Hong Kong dismantled itself. In the past, a poor person who could afford only a bedspace and could not read could still push a wooden cart as a hawker, earn enough for two meals, and raise a child. In 1974 Hong Kong had more than 50,000 hawker licences; from the 1970s the government almost stopped issuing new ones, and by the end of 2016 fewer than 6,000 remained (LegCo Research Office, 2017). The Sheung Wan night market, Dai Dat Dei, disappeared in 1992 when the land was reclaimed (Our China Story). With the space for the poor to make their own living taken away, only two routes remain: a job or welfare; and then society points at people on welfare and says they won't work.
Asset dividends are no cure-all either. Since 1982, Alaska's Permanent Fund has paid residents a share of oil revenue every year, one of the few long-running examples; but in 2025 each person received only $1,000, the lowest ever in real terms, because of falling oil revenue and budget cuts (Alaska Public Media). Even when assets belong to the public, how much is shared out is still a political decision — the criteria in section 11 apply here too.
So the question is not only “should those who don't work be worse off”, but “why don't those who work get a share either”. The parent in the comments earning under HK$18,000 a month is far closer to the welfare family than to the productivity line.
Final chapter: society needs people to work; that doesn't mean people must work to deserve to live
How to design universal protection is an extremely complex question. This essay has no answer and will not pretend to. The problem with these videos is precisely that they jump to solutions too fast: the first episode and Chiu designed a system around an unverified case; the second episode designed fear around a ten-year prophecy; the June episode substituted a bestseller for an answer; and Chiu's video on France carried a fiscal problem all the way to immigration.
The ways out offered by the three Lorey Flash Talk episodes point the same way. The June episode's answer was ultimate creativity: the host cited a bestselling book of more than two hundred pages containing only the word “meow”, to show that the market still has spending power if you are creative enough. The exit from structural unemployment, it turns out, is a book full of “meow”. The first episode's answer was to tighten collective protection; the second's was for every family to hold twelve to twenty-four months of cash and, while the financial system still works, convert it early into “resources that will have value in future”. One tells people to rely on creativity, one says collective protection will be abused, one says it will become control; all conclude: don't rely on the collective, rely only on yourself. Nor did the three episodes ignore design. The first episode's design is clear: protection must carry punishment, so that people are forced to work. The second fears protection becoming control, but does not recognise that this is exactly the first episode's design. What none of the three asks is another question: can protection do without punishment, and society still have people willing to work?
But one idea, at least, can be dismantled: that more work, more pay equals worth, and that people who don't work should live worse.
Society needs people to keep it running; that is a fact. Healthcare, care, maintenance and infrastructure all need hands. But this is a collective need, not a condition for an individual's survival. Tie the two together, and employers benefit: if not working means misery, a person cannot refuse any job, and importing labour further weakens what bargaining power they have left.
If society really needs people to do hard, low-paid work, the reasonable thing is to raise wages to hire them, not to make those who don't work so miserable that they have no choice.
There are always people who say that nothing in this world is deserved. The line is rarely used upward: no one tells landlords their rent is not deserved, or shareholders their dividends are not deserved. It is only said downward, to workers asking for a raise and families applying for welfare. “More work, more pay” on one hand and “nothing is deserved” on the other are simply two lines used at different times to hold down the same people. “Nothing is deserved” was never a neutral observation, but a principle allowed to run only downward; accept it, and any exploitation can be called natural.
A reasonable safety net asks very little: anyone who falls, under any circumstances, should be fed and kept warm. It need not make life comfortable, but it should not be designed to make life unbearable. The attitude behind “no enjoyment of life whatsoever” is: you're poor, so you shouldn't have it easy. The trouble is that the height of the safety net decides more than the bottom floor. If the bottom has somewhere to retreat to, people need not fight to the death over a bad job; once the bottom fights for its life, the pressure travels upward floor by floor: when the working poor fight for jobs, clerks' pay is squeezed; when clerks fight for positions, white-collar hours get longer. Many people think that because they earn above the median, the safety net has nothing to do with them; but as the floor keeps sinking, everyone standing on every floor above it sinks too.
Fraud should be investigated, approval errors should be held to account, and children should be protected. These can and should be handled separately. But none of them needs to rest on “the poor must suffer more”, and people without work should not be humiliated for it.
In the second episode's comments, one viewer wrote just three words: alternative welfare, CSSA by another name. He was half right. It is not another name; it is the same thing. Whether it is acceptable depends only on whether we believe we will never be on the receiving end. On the day Mr Chan clicked like, he believed exactly that.
Work exists to keep society running; people should not have to work to earn the right to survive. And a society whose productivity keeps rising should let people do more than survive.
“Snatching Rice from the Beggar's Bowl” series: Prologue | Part 1 “Suffer If You Don't Work?” | Part 2 “Global Top, Local Bottom” | Part 3 “The Fraud You Can See, the Numbers Nobody Measures” | Part 4 “Those Who Leave and Those Who Arrive” | Final part “After the Counting”
The next essay asks: if the problem is distribution, where do you yourself stand in it?
References
The videos discussed (in Cantonese, unlinked): Lorey Flash Talk, Sun Channel, episodes of 18 June 2026, October 2026 and 16 October 2026; Chiu's Reading Life, episodes of 8, 19 and 20 October 2026. All other sources are linked inline throughout the essay, including the LegCo Research Office (2016, 2017, 2018), EPI, the ILO via VOA, ABARES, Alaska Public Media, JURIST, the French Ministry of Education (DEPP), IPP, the French National Assembly, AFP, Taiwan's Ministry of Labor and Ministry of Health and Welfare, and HK01's reporting on the Fu Tai Estate case.