A Stone From Another Mountain — Why Korean Society Cannot Avoid Reform

Tasan-jiseok (他山之石).

It means that even a stone from another mountain can be used to polish my jade.

I love this saying. Because when looking at the world, it makes me think not about what I like or whose side to take first, but about what words I can extract from the reality in front of me.

In “even a sheet of paper is lighter lifted together,” you can see collaboration. In “too many boatmen steer the boat up the mountain,” you can see responsibility and decision-making structure. The two sound contradictory, but real organizations need both at once. The OECD's 2026 report on Korea likewise recommends moving to a structure where a job's difficulty, responsibility, required skills, and performance are reflected more directly in pay and careers.

In “tap even a stone bridge before crossing,” you can see verification. In “pull the bull's horn while it's hot,” you can see speed and decisiveness. Both variables coexist in the startup market too: total Korean venture investment in 2025 was 13.6244 trillion won, up 14.0% from the prior year, yet the share going to early-stage companies within three years of founding fell to 16.6%, a trend the Ministry of SMEs and Startups itself described as “a preference for investing in proven growth companies.”

In “dig one well if you dig at all,” you can see focus and expertise. In “don't put all your eggs in one basket,” you can see diversification and risk management. In 2025 venture investment as well, 2.2666 trillion won went to early-stage companies while 7.4156 trillion won went to late-stage companies past seven years — late-stage share rising to 54.4% of the total.

What matters, in the end, is not memorizing one sentence as the correct answer. It is seeing which variable is actually at work in front of me right now.

I consider this a positive attitude. Not believing reality is unconditionally good, but finding what to replicate in good cases and what structures never to repeat in bad ones.

And the most important object of observation is, in the end, oneself.

If I keep observing what I learn quickly, in which environments I produce results, and how I tend to fail, experience becomes data rather than mere time. The OECD's 2026 report also finds that people who learn repeatedly on the job or participate in adult education show higher competency, while Korea's adult education/training participation and learning-at-work indicators rank low within the OECD.

Yet in Korea, even as the need for lifelong learning grows, analysis shows the labor market's dual structure and seniority-based wages are weakening the incentive to learn. The OECD explicitly points out that Korea's regular/non-regular duality and seniority wages depress the incentive to invest in lifelong learning.

From here, I look at Korean society.

This is not a metaphor.

In 2024, 14,872 people in Korea died by suicide. That was 894 more than the prior year, a 6.4% increase, and the suicide rate of 29.1 per 100,000 was the highest since 2011. Averaged daily, it was 40.6 people.

More important are the leading causes of death by age. In 2024, suicide was the number-one cause of death for people in their teens, twenties, thirties, and forties alike. It was the first time since the statistics began that suicide overtook cancer as the top cause of death among people in their forties.

Suicide accounted for about 48% of teenage deaths. About 54% in the twenties, about 44% in the thirties, about 26% in the forties. In other words, more than half of all deaths in the twenties were suicides.

Even in OECD comparisons that remove differences in national age structure, Korea's age-standardized suicide rate in 2024 was 26.2 versus an OECD average of 10.8 — roughly 2.4 times the OECD average.

To conclude from this that “everyone dies by suicide because of Korean society” would be false. The national statistics office likewise explains that suicide involves multiple interacting factors — mental health, economic problems, physical illness — so cause-of-death statistics alone cannot establish individual causal relationships.

But a different question is possible.

If suicide is simultaneously the leading cause of death from the teens through the forties in one country, can all of it be viewed as a purely individual problem?

At minimum, this is social data beyond the stories of a few individuals. The fact that in 2024 alone, 14,872 distinct personal histories gathered under the same statistical category does not change.

On April 22, 2021, 23-year-old Lee Seon-ho, working at Pyeongtaek-Dangjin Port, was crushed to death by a collapsing container wall while removing cargo-securing timber. The investigation found the work had been improvised without prior planning; Lee was deployed without even basic safety equipment, and the container's own safety mechanisms were not functioning properly.

On October 15, 2022, at the SPL Pyeongtaek plant of the SPC group, a 23-year-old worker died caught in a sauce mixer. She was a young employee about two and a half years into the job, and structural safety questions were raised about the machine in use at the time.

On June 24, 2024, a fire at the Aricell plant in Hwaseong killed 23 people and injured 8. Of the 23 dead, 18 were foreign nationals; the Aricell CEO was later sentenced in the first trial to 15 years in prison for violating the Serious Accidents Punishment Act, among other charges.

Nor is this a story assembled from a few famous accidents.

In 2025, the Ministry of Employment and Labor's investigated workplace-accident deaths numbered 605. 286 died in construction, 158 in manufacturing, 161 in other industries.

By workplace size, 351 of the 605 died at workplaces with fewer than 50 employees. Of those, 174 deaths occurred at workplaces with fewer than 5 — up 22 from the prior year, a 14.5% increase.

The link between company size and on-site risk is not mere conjecture. In the cumulative 2025 industrial-accident statistics the ministry released in 2026, roughly 58% of all accident deaths occurred in the under-50 segment.

Nor were large companies free of accidents. The ministry disclosed that POSCO E&C had 1 serious accident in 2023, 3 in 2024, and 5 in 2025 since the Serious Accidents Punishment Act took effect, and that inspections of 55 sites and headquarters confirmed 403 violations of the Occupational Safety and Health Act.

The words I see here are authority and risk.

The person with the authority to decide work plans and safety measures at Pyeongtaek Port and the 23-year-old standing beneath the container wall were not the same person.

At SPL, the person who decides production-equipment structure and safety investment and the 23-year-old working at the sauce mixer were not the same person either.

At Aricell too, the executives who decide the production and safety-management system and the workers who lost their lives in the fire did not stand in the same position. The court, in the first trial, recognized management's safety-and-health responsibility and handed down what was then the heaviest sentence since the Act took effect.

That is: the fact that the position that bears production's risks with its body and the position that decides production's conditions can be separated is confirmed by actual events.

Korean youth really do search a long time for good jobs.

As of May 2025, the population aged 15–29 was 7.974 million, with an employment rate of 46.2%, down 0.7 percentage points from a year earlier. Labor-force participation also fell 0.8 points to 49.5%.

The average time from graduating (or leaving) one's final school to the first paid job was 11.3 months. For youth with high-school education or less it averaged 1 year 4.5 months; for university graduates and above, 8.8 months.

Among economically inactive youth, 14.5% were preparing for employment exams. Fields of preparation: private companies 36.0%, general civil service 18.2%, technical certifications and the like 17.8%.

The OECD does not view this as simply young individuals' choices. The 2026 Economic Survey explains that because of a dual labor market with large gaps in wages, job security, and social protection, young people compete excessively for elite universities and good jobs.

The same report notes that non-regular workers are concentrated in SMEs while regular workers are concentrated in large firms, and that non-regular workers, moving between low-quality jobs, struggle to accumulate firm-specific skills and to climb into more productive jobs.

The actual wage gap is large too.

Per the labor ministry's 2026 release, non-regular workers' total hourly wages in June 2025 stood at 65.2% of regular workers' — the gap widening 1.3 percentage points from the same month a year earlier.

A 2025 survey put the ratio at 66.4%. The size of the gap shifts with measurement method and whether part-time work is included, but the existence of substantial wage differences by employment type is confirmed in official statistics.

So from a young person's standpoint, the advice “just take any job first and build a career” is not always rational.

The OECD directly notes the phenomenon of young Korean workers delaying labor-market entry to prepare credentials and exams rather than accepting low-quality jobs.

This bears directly on how we view “resting” youth.

We cannot assume every resting young person has finished a rational economic calculation. But in a society with large gaps in pay, stability, and social protection between good jobs and bad ones, waiting for a better entry point is behavior that is economically explicable.

Another feature of the Korean labor market is seniority.

The OECD repeatedly points to the large influence of tenure in Korean firms' wages and career progression. The 2026 report again recommended moving from seniority wages to job-centered structures reflecting difficulty, responsibility, required skills, and performance.

The problem is not being old, nor having worked long.

Experience is a real asset. Someone with twenty years on the same site may judge far better than a novice.

But the problem begins when tenure itself is treated as identical to current ability.

The OECD analyzes that in Korea's seniority-wage structure, pay growth can outpace productivity growth as careers progress, leading firms to induce honorary retirement of regular workers around age 50, after which workers move into lower-paid second careers.

That is, the seniority system is not unconditionally good even for incumbent older workers.

On one side, insiders' wages rise with tenure; on the other, the paradox arises of firms unable to sustain that wage structure inducing early exits and sending people back into lower-reward jobs.

The OECD indeed holds that Korea's dual labor market and seniority wages reduce lifelong-learning incentives. For those already inside, keeping the seat can pay more than continuously building new abilities; for outsiders, opportunities for long-term firm-specific training are weak in the first place.

The core I see here is the separability of ability and position.

Winning a position through ability at the start, and then keeping the same authority regardless of current ability, are different problems.

Korea's education fever is hard to explain as mere parental greed.

The OECD's 2026 report directly connects the large-firm/SME and regular/non-regular gaps to a structure that makes competition for elite universities and good jobs excessive.

University rank connects to actual labor-market outcomes.

In research the OECD cites, graduates of top-tier Korean universities earn 24.6% more than graduates of lower-tier ones at labor-market entry, and the gap widens to **50.5%** by ages 40–44.

Then students and parents pouring time and money into entrance competition cannot be explained by cultural obsession alone.

If the name of the university one entered connects to labor-market rewards decades later, participating in the competition is a considerably rational choice for the individual.

But a different problem arises for society as a whole.

The OECD points to the possibility that test-centered education and credential preparation for good jobs delay youths' labor-market entry and institutionally exclude people with narrow, high-level abilities needed in actual jobs.

The same issue has been observed in civil-service recruitment, where career-based and open recruitment tracks beyond exams have expanded; the OECD notes these alternative routes already account for a substantial share of hiring.

Here a collision arises with the “ability to observe oneself” I mentioned at the start.

In a society where securing the socially recognized admission ticket matters more economically than discovering what one does well and deepening it, individual learning, too, ends up shaped to the test.

In Korea, the same housing price means entirely different things to someone who already owns a home and someone entering the market for the first time.

In 2023, 43.6% of all households did not own a home. But among young households headed by someone under 39, the non-ownership rate rose from about 65.9% in 2015 to **73.2%** in 2023. Over the same period, the rate among those aged 40–59 and 60+ declined gently or changed little.

This does not mean the entire young generation is poor and the entire older generation is rich.

Indeed, the relative poverty rate among those 66 and older was a very high **39.8%** in 2023; elderly poverty is a grave Korean problem in its own right.

But it is simultaneously true that on the specific variable of housing entry, young households are at a disadvantage.

In 2024, 53.4% of Seoul households were renters. As of 2020, the monthly-rent share among households headed by someone 29 or younger was **55.9%** — the highest of any age group.

The shift from jeonse to monthly rent is confirmed in real numbers too.

Of roughly 2.16 million lease contracts nationwide granted fixed-date status in January–September 2025, about 1.36 million — **62.8%** — were monthly-rent contracts. The national statistics office analyzed this monthly-rent shift as a structural change constraining young people's housing stability and asset formation.

Housing quality differs as well.

In 2024, 3.8% of all households fell below the minimum housing standard — but 8.2% of youth households, and 9.5% of youth households in the capital region.

The share living in “non-housing dwellings” — gosiwon rooms, lodging-facility rooms, corners of workplaces — was 2.2% of all households but 5.3% of youth households, 5.7% in the capital region.

Floor area differs too.

In 2024, average dwelling area was 81.3㎡ for owner households, 63.5㎡ for jeonse households, 39.7㎡ for monthly-rent-with-deposit households, and 25.2㎡ for monthly-rent-without-deposit households.

The price-to-income ratio (PIR) stood at a national median of 6.3 in 2024, generally 8–10 in the capital region, and around 15–17 in Seoul.

That is, the person trying to newly acquire a median Seoul home on labor income alone and the person who already owns that same home are looking at the same real-estate market with opposite economic effects.

When prices rise, for existing owners it is an increase in appraised assets.

For the homeless young person who must buy, the same rise is an increase in required initial capital and debt burden. The under-39 non-ownership rate reaching 73.2% is one result displaying this difference.

Korean households' average assets stood at 566.78 million won as of end-March 2025, average debt at 95.34 million, average net assets at 471.44 million. But a single average tells you nothing about how assets are distributed.

In the same survey, the 2024 Gini coefficient of equivalized disposable income rose to 0.325 from the prior year; the quintile share ratio was 5.78, and the relative poverty rate 15.3% — all higher than the year before.

Nor can the elderly be bundled into one wealthy bloc.

Households headed by someone 65+ had average net assets of 465.94 million won in 2024, yet the relative poverty rate at 66+ was 39.8%. High average assets and high income poverty can exist at once.

So the vested interest I speak of is not age.

Making the older generation the enemy actually distorts the explanation of reality.

The real question is not “who is old,” but which assets, positions, and institutions, once secured, keep generating rewards independent of current ability, responsibility, and value creation.

I saw this structure firsthand doing IT outsourcing work.

In projects I experienced, the actual developer understood the client's systems and code most deeply, yet price, schedule, contract terms, and key decisions were set several levels above.

One experience cannot generalize all of Korean industry.

But research exists showing that subcontracting dependence and wage gaps between large firms and SMEs are a long-standing structural problem. KDI research has analyzed how many manufacturing subcontractors are highly dependent on their buyers, and how higher subcontracting dependence correlates with wider large-firm/SME wage gaps.

The reason the Fair Trade Commission still maintains subcontract payments, protection of technical data and trade secrets, and technology misappropriation as distinct core policy areas is precisely that transacting parties' bargaining power may not be equal. The FTC's 2025 work plan likewise included subcontract technology theft and SME technical-data protection among key tasks.

The OECD 2026 report, in discussing Korean labor-market reform, also separately addresses raising social dialogue and predictability in indirect-employment structures spanning prime contractors and subcontractors.

What matters here is not “big companies bad, small companies good.”

Large firms provide real value: capital, brand, large-scale sales networks, quality control, legal accountability, project coordination.

But problems arise when the position that creates the actual value and the position that holds bargaining power drift too far apart.

When I once described it as “prime 100, first-tier 70, second-tier 50, actual developer 30,” that was a simplified illustration of the structure I felt on the ground — not official statistics for all Korean contracts.

But the question that metaphor points to can be checked against statistics.

Why does compensation for the same hour of labor differ so greatly by firm size and employment type?

Why were non-regular hourly wages at 65.2% of regular wages in 2025?

Why does the OECD treat Korea's large/small-firm and regular/non-regular gaps as problems important enough to link to society-wide education competition and delayed labor-market entry?

Skip these questions and say only “work harder,” and you fail to see the structure.

AI is not the same as transferring predefined tasks onto computers.

You must keep judging which tasks can be automated, which must retain human judgment, how much accuracy to demand, whether data may leave the premises, and what inference costs are.

In an AI job-task analysis cited by the Korea Labor Institute, roughly **9.8%** of Korean employment was estimated to be in areas highly likely to be automated by AI, and **15.9%** in areas where AI is highly likely to augment human work.

These numbers do not mean “10% of people will soon be fired.”

The Korea Labor Institute likewise explains that at the current stage AI more often replaces or complements parts of tasks rather than whole occupations, and that the technology's impact varies by occupation and by how firms adopt it.

Then the core is transition.

When AI can draft documents, you must choose: protect the hours spent drafting, or eliminate them and move people up to verification, decision-making, client consultation.

When AI can write code faster, you must choose: keep existing development hours as they are, or produce more products in less time.

If AI automates repetitive accounting, you must decide: protect the repetition itself as the profession's identity, or move accountants toward judgment, interpretation, audit, consulting.

This problem is already beginning in the labor-market numbers.

The finding that jobs likely to be augmented by AI outnumber those likely to be automated suggests that at this stage the gap between people who use AI and people who don't may matter more than “human versus AI.” This is an interpretation built on the research.

It is therefore dangerous to confine the AI transition within an organization's existing reporting hierarchy.

If the person who knows the field data best is excluded from decisions, the old subcontracting structure — tech budgets above, tech information below — can simply reproduce itself. The OECD, too, treats Korea's indirect-employment and subcontracting structures and labor-market duality as distinct reform tasks.

Korea's venture-investment market itself is not shrinking.

Venture investment in 2025 was 13.6244 trillion won, up 14.0% from 11.9457 trillion in 2024, with 8,542 deals recorded.

But look at where the money goes and a different story emerges.

The share invested in early-stage companies within three years of founding fell: 19.2% in 2021, 26.9% in 2022, 24.6% in 2023, 18.6% in 2024, and **16.6%** in 2025.

Conversely, the share going to late-stage companies past seven years rose from 44.4% in 2021 to **54.4%** in 2025.

The Ministry of SMEs and Startups itself interpreted the change as “a preference for investing in proven growth companies,” and announced it would double the fund-of-funds' early-startup allocation in 2026 to boost early investment.

So there is no need to exaggerate that “Korean investors hate new technology.”

Government statistics alone confirm the reality: total market investment is growing while the share allocated to the risky early stage is falling.

In AI as well, the unicorns newly confirmed in 2025 included AI-semiconductor firms like Rebellions and FuriosaAI. Counter-examples exist of large capital flowing into companies building genuinely deep technology.

So my point is not “funded AI companies are fake.”

The more precise question: let us not treat funding and valuation as the same concept as technical depth.

A company using APIs can build excellent products.

Conversely, a company with its own model is not a good company if it fails to solve customer problems.

What must be examined, in the end: cost, data, control over the technology, customer value, repurchase, margin, switching costs.

Korean film has not vanished either.

In 2024, Korean films drew 71.47 million admissions, up 17.6%, with Korean-film revenue up 15.5% to 691.0 billion won. “Exhuma” and “The Roundup: Punishment” each passed ten million admissions.

But total theater admissions were 123.13 million, down 1.6% from the prior year, and 44.3% below the 2017–2019 average.

Theater revenue was 1.1945 trillion won, down 5.3% year over year and 34.7% below the pre-pandemic 2017–2019 average.

Box-office success was heavily concentrated in a few titles.

2024's top film “Exhuma” drew 11.91 million, “The Roundup: Punishment” 11.50 million, “Inside Out 2” 8.80 million, “I, the Executioner” 7.53 million. The upper ranks included multiple sequels and already-known IP.

To conclude from this alone that “Korean film has no creativity” is overgeneralization.

The same year produced cases like “Exhuma,” a major hit built on a novel genre combination.

But the Korean Film Council itself points out that investment contraction and declining production counts are actually happening.

KOFIC analyzed that post-pandemic investment contraction reduced production counts, and that box-office polarization and skewed screen allocation left too few films to offset big-budget failures.

Re-releases in 2024 numbered 228 — the most since tracking began in 2013 — with re-release admissions reaching 2.5 million.

So the industry's reality is more complex than “the talent is gone.”

The market as a whole is smaller than before the pandemic, big budgets are harder to stake, and investors want surer outcomes — these conditions exist in real numbers.

In that environment, the cost of giving newcomers a chance to fail becomes more expensive.

This is not moral condemnation but economic inference: a shrinking market likely lowers risk tolerance.

The Korean game industry is not a small industry.

Per the Korea Creative Content Agency's 2024 Game Industry White Paper, domestic game-industry revenue in 2023 was about 23 trillion won, up 3.4% from the prior year.

Of that, mobile-game revenue was 13.6118 trillion won — 59.3% of the total. PC games were 5.8888 trillion won, 25.6%.

That is, more than half of domestic game revenue comes from a single platform type: mobile.

From this number alone you cannot say “every company uses the same business model.”

But when it is already clear where the biggest market lies, the economic incentive to allocate budgets around mobile is naturally explained.

In 2025 venture investment, the game sector drew 271.0 billion won, up 69.4% from 160.0 billion the prior year. So capital for new games has not disappeared either.

Therefore, in explaining the game industry too, one must look at structure rather than the single sentence “it's ruined.”

A giant 23-trillion-won market, 59.3% platform concentration in mobile, and rising new investment all exist at once.

And within that structure, how much failure budget and decision authority to give new directors and developers is, in the end, the company's choice.

In 2023, domestic webtoon-industry revenue passed 2 trillion won for the first time. KOCCA's 2024 webtoon-industry survey tallied 2023 industry revenue at about 2.189 trillion won.

In 2024, webtoon revenue was tallied again at 2.2856 trillion won, growing 4.4% year over year.

Yet the median annual gross income of webtoon artists who serialized throughout 2023 was 38 million won. Industry revenue of 2 trillion won and one artist's income are different indicators.

Work intensity has numbers too.

In the 2023 survey, webtoon artists created an average of 5.9 days a week, with 26.9% responding they worked all seven. Average working time was around 10.1 hours a day.

Contract structures are not singular either.

In 2023, per-work serialization contracts accounted for 54.8% and employment contracts 25.9%, with creators linked to platforms, CPs, and studios in multiple forms.

Therefore the equation “if the industry grows, creators automatically grow stronger” does not hold.

Industry revenue passed 2 trillion won, while the median income of year-round serializing artists is 38 million won and they create an average 5.9 days a week — these numbers exist simultaneously.

This does not mean the single conclusion that platforms exploit artists.

Platforms provide real functions — distribution, payments, recommendation, overseas expansion, marketing — and artists' reasons for preferring platform contracts include more readers and revenue opportunities.

What is clear, however, is that the industry's growth rate and individual creators' bargaining power, income, and rest must be measured separately.

In 2024, domestic popular-culture-arts industry revenue was surveyed at 15.3845 trillion won — 34.5% larger than 2022's 11.4362 trillion.

The claim that the industry is shrinking does not match the numbers.

Affiliated popular-culture artists numbered 12,092, up 6.5% from 2022.

But surveyed entertainers' average monthly personal income was 3.151 million won, of which income related to artistic activity averaged 1.507 million won. Production staff averaged 3.275 million won a month.

Trainees numbered 963, down 207 from 2022, a 17.7% decrease. Of them, 613 — 63.7% — were aspiring singers.

At the same time, the standard exclusive-contract adoption rate rose to 95.3% and the written-contract rate for production staff to 97.9%. Positive data also exists showing contracts' formal transparency improving over the past.

That is, K-pop's reality is also wrong if explained in only one direction.

The industry is growing greatly and the contract system is improving, while individuals' artistic-activity income moves on an entirely different scale from total industry revenue.

So “it succeeded globally, therefore its insides are healthy” is wrong — and declaring “the whole industry is an exploitation structure” also runs ahead of the data.

What must be watched: through which contracts, to whom, and how much of revenue growth's share moves.

In the Korean advertising market, online is not a peripheral channel.

Per KOBACO's 2025 broadcast-communications ad-spend survey, the online ad market in 2024 was 10.1011 trillion won. For 2025 it was estimated to grow about 6.1% to 10.7204 trillion.

That is, the center of ad spending has already moved online.

As measuring impressions, clicks, conversions, and purchases became easy in online advertising, the growing importance of metrics like CTR, CPA, and ROAS is a technically natural result.

To take this as “we mustn't look at numbers” would be wrong.

If anything, numbers let us measure effects the old ad industry could hardly know, far more precisely.

But measurable short-term performance and long-term brand memory are not the same metric.

So what the ad industry needs is not an either/or.

Measure performance more precisely — while refusing to treat hard-to-measure brand equity as nonexistent. That is the core.

In industrial accidents, the question was whether the person bearing the risk and the person deciding the conditions are the same. The number behind that question: of 605 accident deaths in 2025, 351 occurred at workplaces under 50 employees.

In the labor market, the question was how connected current ability is to current reward. The background: non-regular hourly wages at 65.2% of regular in 2025, and the OECD defining the gaps in pay, stability, and social protection as a “dual labor market.”

In housing, the question was whether those who already hold assets and those now entering compete on equal terms. The background numbers: a 73.2% non-ownership rate among under-39 household heads, and Seoul PIR around 15–17.

In startups, the question was whether new entrants and proven players have equal access to capital. The background: early-stage investment share at 16.6% in 2025, late-stage at 54.4%.

In webtoons, the question was whether industry revenue growth and creators' personal finances move in the same direction. Industry revenue passed 2 trillion won, but year-round artists' median income was 38 million won and average creation days were 5.9 a week.

In film, the question was who has the capacity to bear risk. The background: KOFIC's analysis of contracted investment and production counts with total admissions 44.3% below the pre-pandemic average.

In AI, the question becomes who holds the authority to use the technology and who tries to guard existing roles. Research already shows 9.8% of Korean employment has high automation potential and 15.9% high augmentation potential.

Entirely different industries — and the same words keep surfacing.

Authority. Responsibility. Risk. Reward. Entry. Position. Assets. Learning. Transition.

Define vested interests as “people with money” and you cannot explain reality.

The statistics show 65+ household heads with average net assets of 465.94 million won alongside a 39.8% relative poverty rate at 66+. You cannot bundle a group into one economic class merely because they are old.

Working at a large company does not automatically mean holding unjust power.

Those who create high productivity and bear weighty responsibility at large firms have every reason to receive correspondingly high rewards.

The same goes for licensed professionals.

That doctors, lawyers, and accountants who pass difficult study and exams and take responsibility for complex judgments receive high compensation is not economically strange in itself.

The vested interest I mean is narrower.

It is the state in which a position, credential, asset, or internal organizational status secured in the past keeps producing rewards without current ability and responsibility being continually verified.

Under this definition, the focus of the problem moves from generation to structure.

A young person who secures a monopoly position can become a vested interest, and an older person who keeps competing, learning, and creating new value can be irrelevant to the criticism.

This question matters especially in the AI era.

In large IT projects of the past, collecting requirements, handing them to a developer, and inspecting deliverables could work to some degree.

In generative-AI projects, model performance and prices change fast, and what is possible can change weekly with the state of the data.

Then the moments multiply in which the person on the ground holds more information.

But if organizational authority is fixed around rank and tenure, information and authority end up in different places.

The reason the OECD recommends shifting Korea's pay and career systems from seniority to job, responsibility, skill, and performance is precisely to connect the value of the work currently being done more closely to its reward.

Nor is this reform only for the young.

The phenomenon of honorary retirement concentrating on regular workers around 50 because of seniority wages, followed by moves into lower-paid re-employment, is also a Korean problem the OECD has flagged.

That is, redesigning roles and rewards to current ability is not a reform that drives out the old for the sake of the young.

It is closer to a reform that lets people work long regardless of age, keep learning, and be paid to match the job.

Look again at the industrial-accident cases and the answer is not simple.

Lee Seon-ho, a 23-year-old day laborer, bore the site's risks but held no position deciding work plans or the safety system.

The 23-year-old in the SPL accident likewise held no position designing the production equipment or deciding the firm's safety-investment budget.

In the Aricell fire too, management stood trial under the Serious Accidents Punishment Act over ultimate responsibility for the safety-and-health system, and a heavy sentence came down at first instance.

This does not mean capital holders and managers bear no risk at all.

Entrepreneurs, too, can lose invested capital and bear criminal, civil, and administrative liability.

Precisely for that reason, the more accurate question is whether the risk of bearing economic loss and the risk of one's body dying or being maimed can be treated identically.

What 2025's 605 accident deaths show is that in the labor market, certain risks concentrate on the front line in forms money cannot restore.

Then positions with great risk must carry corresponding safety investment, compensation, and participation in decision-making.

This is a matter closer to the basics of contract than to ideology.

Labor-market insiders may already hold regular status, tenure, corporate welfare, social insurance, and in-organization networks.

Outsiders pay first — in education, exams, job-search time — to win that seat.

The statistics that youths' first job takes an average 11.3 months in 2025 and that 14.5% of inactive youth are preparing for employment exams are precisely part of this entry cost.

Housing is the same.

For homeowners, price increases are asset growth; for the 73.2% of under-39 households without homes, the same increases are increases in the price they must eventually pay.

University is the same.

Someone who already holds a top-tier diploma can use it on a lifetime résumé, while the next generation must compete anew for that admission ticket.

The research finding that top-tier graduates' wage premium grows from 24.6% at entry to 50.5% by the early forties shows the ticket's economic value.

In this situation, telling the young only “don't fear competition” is half an explanation.

Yes, they must compete.

But if incumbents' competitive winnings are protected permanently while only new entrants must keep competing afresh, that is not circulating competition.

The Korea Employment Information Service published a separate 2025 study on “resting” youth.

That the government surveyed, as an independent research topic, these youths' previous jobs, finances, anxieties, employment perceptions, and policy needs itself means the phenomenon is not viewed as mere lack of individual will.

The national statistics office's youth survey likewise shows first employment taking an average 11.3 months after graduation, with many inactive youth preparing for company, civil-service, and certification exams.

The OECD explains still more directly the phenomenon of young Korean workers delaying entry and upgrading credentials rather than accepting low-quality jobs.

So within “resting” there can be health problems, exam preparation, burnout, and failure to find a suitable job.

Bundle the whole group under one character and you can explain nothing for policy.

This part must be kept distinct.

Suicide has multiple causes — mental health, relationships, illness, economic problems. The statistics office likewise explains that cause-of-death data alone cannot determine specific individual causes.

Industrial accidents arise from different variables — safety management, firm size, industry, equipment, training, work planning. The 2025 statistics splitting accident deaths into construction 286, manufacturing 158, other 161 show this.

Youth inactivity mixes different situations — exam preparation, childcare, illness, rest.

Low birth rates likewise cannot be explained by housing costs alone; labor markets, gender roles, education costs, and values all operate. The OECD, too, analyzes Korea's low fertility as a compound problem including labor-market duality and high housing costs.

Even so, there is a reason to put these data in one essay.

The causes differ, but within the same society we can see together who bears the risk, who pays the entry cost, and who already holds secured stability.

Refusing to bundle causation carelessly and comparing structurally simultaneous phenomena are different acts.

Lay out the numbers this far and a pattern appears.

Youth non-ownership under 39 keeps rising.

Between good jobs and the rest, gaps in pay, stability, and social protection are large.

Non-regular hourly wages stand at 65.2% of regular.

Seniority wages make tenure's influence larger than current job value.

The venture-investment share allocated to early-stage firms is 16.6%; late-stage, 54.4%.

Youth spend an average 11.3 months reaching their first job.

The single largest cause of death for Koreans from their teens through their forties is suicide.

On industrial sites, 605 people died in accidents in 2025 alone.

This does not mean all these numbers share one cause.

But they interlock substantially with the description of a society where conditions differ greatly between those already inside and those now trying to enter. The OECD likewise analyzes labor-market duality, elite-university competition, migration to the capital region, housing costs, and differing mobility gains by parental income and assets as one structural problem-set.

The OECD's 2026 report even cites research that moving to the capital region is no longer a path to prosperity for everyone as it once was, with the economic gains of migration now varying greatly with parental income and assets.

That is, even two young people of equal ability can face different conditions for developing that ability, depending on parental assets, housing base, educational opportunity, and the quality of the first job.

This is why, when I speak of an ideal society, I keep saying: a society that keeps developing its people's abilities.

When starting position plays too large a role relative to ability itself, people spend their energy securing admission tickets instead of building ability.

The OECD linking Korean youths' excessive elite-university competition and credential preparation to labor-market duality is the same context.

Here the reform conversation begins.

Say technology eliminates some task.

Simply discarding the person is not good reform.

Unemployment benefits, retraining, vocational training, mobility support, and social insurance must carry that person to the next job.

But protecting people does not require preserving the existing task itself forever.

Take AI's impact: current research sees high automation potential in 9.8% of Korean jobs and high augmentation potential in 15.9%. When task structures may genuinely change, preserving existing job descriptions as-is collides with reality in the long run.

The same goes for regular-worker protection.

This is not the crude claim that firing regular workers should become easy overnight; the OECD's recommendation is to strengthen social insurance and mobility opportunities for non-regular workers while narrowing the excessive protection gap between the two.

The same goes for seniority.

Not that older workers' wages should simply be cut — but that responsibility, difficulty, required skill, and performance should connect more directly to reward.

The same goes for housing.

Not that homeowners should be punished — but that if policy's only goal is permanently lifting the asset value of homes already owned, the entry cost for the generation without homes rises with it. The under-39 non-ownership rate of 73.2% and Seoul PIR of 15–17 show that reality.

That is: people can be protected. But not every existing position can be protected permanently.

This is the most uncomfortable part.

Moving to job-based pay may reduce the expected future wages of some who fared relatively well under seniority.

Narrowing the regular/non-regular protection gap may make incumbent regular workers feel their relative job-security premium shrinking. The OECD explains that this is precisely why seniority wages enjoy strong support among large-firm regular workers and union insiders.

If housing supply expands greatly and price-rise expectations fall, existing owners' expected asset returns may drop.

When new technology arrives, firms and occupations that earned from the old tasks may lose part of their revenue.

When new companies enter, incumbents' market share shrinks too.

Allocating more policy capital to early-stage startups moves some capital that could have gone to proven late-stage firms toward riskier places.

That is, real reform always has distributional effects.

A reform in which no one loses, everyone's existing share is preserved intact, and new people receive even more opportunity is, in reality, often mathematically impossible.

Someone actually gets hurt.

So complementary measures are needed to reduce that harm.

But halting all change merely because harm exists means the existing structure almost never changes.

Korea's labor-market duality is not a newly discovered problem.

Yet the OECD in its 2026 report again presents narrowing the regular/non-regular gap, reforming seniority wages, and expanding social insurance as core tasks.

Housing is not a new problem either.

Yet the under-39 non-ownership rate rose from 65.9% in 2015 to 73.2% in 2023.

The monthly-rent shift has long been underway too.

62.8% of fixed-date lease contracts in January–September 2025 were monthly rent, and the statistics office assessed it as a structural change constraining young people's housing stability and asset formation.

Industrial accidents are an old problem.

Yet in 2025, 605 people still died in accidents, and deaths at workplaces under 5 employees actually rose 14.5% from the prior year.

Suicide is an old problem.

Yet in 2024 the count and rate climbed to their highest since 2011, and for the first time suicide became the leading cause of death all the way through the forties.

So too much time has already passed to say “wait a little longer and it will naturally improve.”

Nor need reform be seen as a matter of social justice alone.

The OECD's 2026 Korea report estimates that pursuing product- and labor-market structural reform could lift GDP 1.1% by 2032 and 8.8% by 2060 versus baseline.

The same analysis presented easing labor-market duality, reforming seniority wages, expanding female and elderly labor supply, and product-market deregulation as elements that together raise growth potential.

That is, maintaining the existing structure is not always the economically conservative, safe choice either.

The cost of maintaining inefficient resource allocation accumulates invisibly every year.

Youth spending 11.3 months to first employment, repeating exam preparation for good jobs, failing to accumulate skills in low-quality jobs — all of these are costs in productivity terms.

Workers unable to move to high-productivity regions because of Seoul's housing costs, or bearing excessive housing costs, are likewise a labor-market efficiency problem. The OECD, too, notes that high Seoul housing costs burden labor mobility and welfare.

Reform, then, is not charity bestowed on someone — it is also the process of moving society's resources back toward where present value is higher.

I do not want a society where abilities differ but everyone receives the same reward.

The opposite.

I want a society where the person who is genuinely better receives more authority and more reward.

The OECD's job-based pay recommendation — reward by responsibility, difficulty, skill, and performance — points the same direction.

But that reward must stay connected to present value.

Having done well in the past is no reason to hold the most decision power forever.

Passing an exam in the past is one event that proved ability — not evidence that one can make the best judgments today after decades without learning new knowledge and skills.

That a company was innovative in the past is no reason to permanently preserve its current market position through regulation.

That someone bought a house in the past is not a social contract obliging us to preserve prices the next generation's labor income cannot reach.

That is: a society is needed that rewards achievement without turning the reward into permanent status.

This question is easiest to see inside organizations.

Say a team has a manager with 20 years' experience and an engineer with 5.

Managing people and negotiating with clients — the 20-year veteran may do better.

Designing new AI models and data pipelines — the 5-year engineer may do better.

A healthy organization does not decide which of the two is the higher human being.

It gives decision rights, problem by problem, to whoever knows best.

In a seniority structure, every important decision risks following a single rank.

One reason the OECD recommends tying pay and careers more to a job's responsibility, skill, and performance is precisely to align rank more closely with actual work value.

I believe this applies to society as a whole.

When a new person does better, they should rise.

When the incumbent still does better, they simply stay.

The core is not age.

It is whether the structure allows continuous proving.

If startups truly want innovation, they must also accept that early companies may fail.

The government itself judged the 16.6% early-stage share of 2025 as “preference for proven growth companies” and rolled out early-investment expansion — out of the concern that capital may not flow sufficiently to new companies.

Film is the same.

When the audience market shrinks far below pre-pandemic levels and investment and production counts contract, the capacity to absorb newcomers' failures shrinks. KOFIC likewise flagged investment contraction and fewer productions as key 2024 problems.

In webtoons too, industry revenue growth does not mean artists can work infinitely more.

An average 5.9 days a week, 10.1 hours a day — that workload is data showing creative sustainability must be watched as closely as industry performance.

Permitting failure does not mean paying failed people forever.

It means the channel must not disappear: fail small, bring back the data, try again.

The logic of the stone from another mountain lives here too.

If failure is the end, the economic reason to learn from failure shrinks.

When there is a chance to try again after failure, failure becomes data.

In 2024, suicide was the leading cause of death from the teens through the forties.

About 48% of teenage deaths, about 54% of twenties deaths, about 44% of thirties deaths were suicides.

These numbers must not be explained by labor-market duality alone.

Nor by housing prices alone.

Nor should industrial accidents and suicide be called the same cause.

But looking at these numbers and speaking only of individual willpower is also an explanation far too small for the data.

It is entirely reasonable to view together the conditions of one society: youth non-ownership rising, the gap between good and bad jobs large, labor-market entry delayed, monthly-rent conversion advancing, and hundreds dying in workplace accidents every year.

I do not bind all these phenomena into one simple formula.

I only pose the question: how convincingly is this society showing its young that if they develop their abilities, the future will be better than now?

Frame it as generational war and only generational war remains.

The 39.8% elderly poverty rate alone shows that bundling the whole older generation into one vested-interest bloc does not fit reality.

The criterion of reform must be structure, not age.

A 70-year-old who keeps developing current ability should have a path to work.

The OECD likewise recommends changing the Korean structure in which seniority wages trigger exits around 50, so that people can work longer according to job value rather than age.

Conversely, high positions cannot be handed out merely for being young.

Without proven ability, one must learn and prove.

That is: what I want is not generational replacement itself.

It is rebuilding the connections among ability, responsibility, risk, and reward.

When industry automates, people's livelihoods must be protected.

But that does not mean people must keep performing repetitive work that can be automated.

When companies struggle, workers need transition support.

But a company that has lost its competitiveness cannot be protected in the market forever.

If some lose out from seniority-wage reform, transition periods can be provided.

But there is no reason to hand a structure with weak links between job and performance down to the next generation.

Households strained by housing-price adjustment need policies managing acute debt risk.

But permanently raising every homeowner's asset price cannot be the state's everlasting duty.

This distinction matters.

Protecting people and preserving existing structures are different things.

The professions in the AI era are the best example.

That AI has the potential to automate parts of repetitive document and analysis work is already confirmed in labor-market research.

That does not mean lawyers, accountants, doctors, and developers disappear.

If anything, the value of judgment humans must answer for — explanation, verification, client relationships — may grow.

The choice professionals should make then is not to block the technology but to climb on top of it.

An accountant need not make out-typing AI at spreadsheet entry a competitive edge.

A developer has no reason to make typing boilerplate faster than AI a mark of expertise.

Expertise should be not a monopoly on repetitive work but the ability to move to higher-order judgment even as new tools appear.

Having been a big company in the past confers no right to be a big company in the future.

Startups likewise.

Having once raised large funding is no guarantee of the next round or of market share.

That 2025 venture investment recovered to 13.6244 trillion won and new unicorns appeared in AI semiconductors, data, and cloud is a live example that power within industries keeps moving too.

A good market is not one that deliberately topples incumbents.

It is one where a new company that does better can rise.

And if the incumbent still does better, it simply keeps winning.

What exam you once passed.

What company you once worked for.

When you bought your house.

What project you once made succeed.

All of that is important data.

But it must not become the identity card that permanently determines society's authority and rewards.

What do you know now?

What can you build now?

What responsibility do you bear now?

What risk do you carry now?

What value are you providing to whom, now?

I believe these questions matter more.

The OECD's recommendation to move Korea's seniority-centered pay and career systems toward job value, responsibility, skill, and performance is, in the end, a structural reform that values present work over the past.

Korea has already accumulated too much data.

In 2024 there were 14,872 suicides, and suicide led causes of death from the teens through the forties.

In 2025 there were 605 industrial accident deaths; 351 died working at sites with under 50 people.

The under-39 non-ownership rate was 73.2% in 2023, and 62.8% of lease contracts in January–September 2025 were monthly rent.

Youth spend an average 11.3 months reaching their first job.

Non-regular hourly wages stand at 65.2% of regular workers'.

The OECD diagnoses that Korea's dual labor market — wide gaps in pay, security, and social protection — drives youths' excessive competition for good jobs.

The OECD points out that seniority wages weaken lifelong learning and even trigger early retirement.

The venture share reaching early startups fell to 16.6%, and the government assessed a visible preference for proven growth companies.

Total Korean film admissions remain 44.3% below the pre-pandemic average, with investment contraction and fewer productions.

The webtoon industry passed 2 trillion won, but year-round artists' median income is 38 million won and they create an average 5.9 days a week.

The popular-culture industry passed 15 trillion won, but surveyed entertainers' monthly artistic income averaged 1.507 million won.

The online ad market passed 10 trillion won, shifting to platforms.

AI already holds the potential to automate part of Korean jobs and augment a larger part.

Before this data, “keep the current structure as it is” is also a choice.

But that choice has a cost.

And the cost is already showing up in the numbers.

So my conclusion arrives here.

Reform must be accepted.

Even if I lose by it, it must be accepted.

My rank's authority may shrink.

AI may do the work I did, faster.

Part of my professional work may be automated.

A new company more efficient than mine may appear.

The housing returns I expected may fall.

The future wages I expected under seniority may change.

The market share I hold may shrink because of new competitors.

That is reform.

A reform in which no one loses, no one lays anything down, every existing position is preserved intact, and new people receive even more opportunity — in reality, that is nearly impossible.

So protect the person, not the position.

People must be protected from industrial injury and unemployment risk. That 605 died on industrial sites even in 2025 shows amply why safety nets and safety regulation are needed.

But inefficient tasks themselves need no eternal protection.

Expertise must be respected.

But repetitive work that new technology does better need not be preserved forever under the name of expertise.

Experience must be respected.

But a structure where authority and reward climb automatically on tenure alone, regardless of present judgment, need not be maintained forever. The OECD repeatedly recommends reforming exactly that structure.

Regular workers' lives must be protected.

But preserving intact a dual structure in which non-regular workers receive 65.2% of regular wages with weak ladders up cannot be the definition of protection.

Homeowners' lives must be protected too.

But with under-39 non-ownership at 73.2% and Seoul PIR at 15–17, not every policy can aim solely at raising existing asset values.

Incumbent firms too deserve fair competition.

But if a new company built it cheaper, better, faster, consumers must also be free to move there.

Creators must be protected.

But freeze the whole industry into making only unfailable works, and the space for new creators and new IP shrinks. Post-pandemic contraction in Korean film and webtoon artists' workloads show why failure and creative sustainability must be designed together.

In the end, I do not think the opposite of vested interest is dispossession.

I think it is circulation.

Authority goes to whoever does it best now.

Reward goes to whoever bears responsibility now.

To those who carry the risk go matching reward and protection.

A new person who does better can come in.

An incumbent who keeps doing well simply stays.

When new technology appears, you learn it.

When old work disappears, you learn new work.

When you fail, you bring the data and go again.

And when I am no longer the one who does it best, I hand it over.

Tasan-jiseok.

Back to where we began.

From “even a sheet of paper is lighter lifted together,” learn collaboration; from “too many boatmen steer the boat up the mountain,” learn responsibility.

From “tap even a stone bridge before crossing,” learn verification; from “pull the bull's horn while it's hot,” learn speed.

From “dig one well if you dig at all,” learn expertise; from “don't put all your eggs in one basket,” learn risk management.

From what is good, take.

What is wrong, fix.

What doesn't work, discard.

When something better appears, switch.

And even if that change disadvantages me, if it is a change that lets more people in this society exercise their abilities, I must be able to accept it.

Korea is already sending too many signals.

Suicide.

Industrial deaths.

Youths' delayed entry into the labor market.

The gap between regular and non-regular work.

Seniority.

Youth without homes.

The shift to monthly rent.

Early companies' access to capital.

The platformization of industries.

The task reshuffle AI brings.

See all of this and still refuse to touch a single existing interest, and in the end nothing changes.

So the final conclusion is very simple.

To some degree, we must shut up and accept reform.

Not demanding reform only of others,

but including the possibility that what I hold may be shaken too.