1. Legal — and Banned
February 19, 2020. Courtroom of the 25th Criminal Collegiate Division, Seoul Central District Court. The presiding judge read the verdict. "Defendant Lee Jae-woong — not guilty. Defendant Park Jae-wook — not guilty." On one side of the public gallery, Tada employees wept. On the other side, taxi drivers shoved back their chairs and stood.
The court ruled that Tada's business model — operating an eleven-seat Carnival minivan with a driver included — fell within the scope permitted by Article 18 of the Enforcement Decree of the Passenger Transport Service Act. It was legal.
Sixteen days later, on the night of March 6, at 11:50 p.m., in the main chamber of the National Assembly, 185 lawmakers were seated. The amendment to the Passenger Transport Service Act — the bill the press was calling "the Tada prohibition law" — was put to a vote. In favor: 168. Opposed: 8. Abstentions: 6. A new provision was inserted: renting a vehicle with a driver requires a minimum rental period of six hours, or departure from an airport or seaport. Tada Basic's mode of operation was effectively banned outright.
The court had said it was legal. The National Assembly made a law that turned a legal service illegal. Between these two events: 16 days.
Tada, launched in October 2018, had within a year and a half accumulated 1.7 million subscribers and a fleet of 1,500 vehicles. Clean cars. Courteous drivers. No need to state your destination. Even deep in the small hours, when ordinary taxis were impossible to find, a car came with a single tap.
Consumers had welcomed it. The taxi industry was furious. Over the course of 2019, four taxi drivers set themselves on fire during anti-Tada demonstrations. One, a 70-year-old owner-operator known by the surname Ahn, doused himself in flames near Seoul City Hall Plaza. The resistance of entrenched interests was fierce, and politics took their side.
Park Jae-wook, CEO of VCNC, said at the time of his indictment: "When we started Tada, we designed and built the service strictly according to the letter of the law under a positive-regulation framework. In that process we met with officials from the Ministry of Land, Infrastructure and Transport, and we received legal review from a law firm." They had followed the law. Then the law changed.
Three years later, in June 2023, the Supreme Court issued a final verdict of not guilty for Lee Jae-woong and Park Jae-wook. Lee Jae-woong wrote on social media: "After four years of a long fight, we have received final confirmation that innovation is not guilty — but during that time, politicians who sided with the entrenched interests that feared innovation changed the law and crushed it."
In April 2020, CEO Park Jae-wook officially announced the suspension of Tada Basic. When a wave of late-night taxi shortages swept through Seoul in 2022, one outlet ran a headline: "If Tada had been here." But Tada was already gone. In the space left by the service, ride-hailing apps moved in, and late-night surcharges climbed higher.
In Chapter 6 we saw the Monetary Authority of Singapore (MAS) guarantee approval within 21 days through its fintech regulatory sandbox. At the same moment in Korea, a legal service was killed by legislative amendment. Tada's death is the most dramatic symbol of Korea's regulatory rigidity. But if it were only a symbol, there would be no reason to write this chapter. Tada is the starting point of a pattern.
2. Three and a Half Years in the Waiting Room
The AI Basic Act (인공지능 기본법) confirms that pattern.
In July 2021, an AI-related bill was first introduced in the Science, Technology, Information, Broadcasting and Communications Committee of the 21st National Assembly. It was a moment when Korea could have produced the first comprehensive AI framework law in the Asia-Pacific. Three and a half years later, on the night of December 26, 2024, the bill passed the main chamber by an overwhelming margin — 260 in favor out of 264 members present. One opposed, three abstaining. It was effectively unanimous. The outcome alone looks like a triumph of consensus. The process is a textbook case of delay.
What happened during those three and a half years? Dozens of AI-related bills were introduced in the 21st National Assembly, but none cleared the standing committee threshold. Under the National Assembly Act, unprocessed bills expire automatically when a parliamentary term ends — and when the general election of April 2024 arrived, every related bill was scrapped in its entirety.
When the 22nd National Assembly convened, lawmakers from both ruling and opposition parties rushed to re-introduce bills. On a single subject, more than 20 bills were registered under different names. A culture that evaluates legislative performance by the number of bills introduced blocked consolidation. Because each lawmaker wanted his or her own bill to serve as the base text, negotiations toward unification stalled. One subcommittee member said: "There were so many bills that just sorting out which one to use as the reference took several weeks."
During those three and a half years, the world did not stop. China promulgated and implemented Interim Measures for generative AI within four months — from the draft published in April 2023 to enforcement on August 15. The EU coordinated the competing interests of 27 member states and completed its AI Act in three years. Korea, a single country with a single parliament, needed three and a half years. A single nation, slower than a 27-country consensus body.
The substance of the compromises was equally problematic. As dozens of bills were merged, core provisions were diluted. The category of "prohibited AI practices" that the EU AI Act specifies does not exist in Korea's law. Strict limits on real-time biometric recognition or emotion-recognition systems are absent. The definition of high-impact AI was narrowed to ten domains, and the imposition of fines was deferred under a guidance period.
One legal scholar's assessment summarizes the three and a half years' outcome: "The result of 3.5 years of collision between the view that protection of safety and civil liberties is inadequate and the industry view that regulatory minimization is necessary is a law that fully satisfies neither side." When the law took effect on January 22, 2026, the government declared Korea "the first country in the world to fully implement AI regulation." That "full implementation" in substantive regulatory effect does not begin until 2027 or later. The title is fast; the substance is slow.
It is worth noting the date on which the bill passed: December 26. Exactly 23 days earlier, on the night of December 3, martial-law troops had entered the same chamber of the National Assembly. On December 14, the presidential impeachment motion had passed. The AI Basic Act was enacted in the middle of the impeachment crisis. On the same day, an amendment to the Income Tax Act postponing cryptocurrency taxation to 2027 was also processed. Legislation in chaos. Meanwhile, ChatGPT had launched, and hundreds of domestic AI services had entered the market with no regulatory framework in place. Even when the law is late, reality does not wait.
The AI Basic Act is not the only case. Telemedicine has been hitting a wall for 20 years. Since the legalization of physician-to-physician remote consultation in 2002, direct remote diagnosis between doctor and patient has not been fully permitted through a single administration — not under Roh Moo-hyun, Lee Myung-bak, Park Geun-hye, Moon Jae-in, or Yoon Suk-yeol. Five governments.
A 2006 pilot project was abandoned under opposition from the medical community. A 2010 amendment to the Medical Act submitted to the 18th National Assembly was never once placed on the agenda of the Health and Welfare Committee's bill-review subcommittee, and expired when the term ended. Re-introduced in the 19th National Assembly in 2014, it again expired automatically without reaching the standing committee. As of 2025, Korea is the only OECD country that categorically prohibits doctor-patient telemedicine in principle.
Over the three years of COVID-19, tens of millions of non-face-to-face consultations were conducted, and no large-scale safety incidents were reported. Yet the logic of "not yet validated" continues to block expansion. In a Korean Medical Association survey, 61.5 percent of respondents opposed to telemedicine said a "full-scale response including a general strike" would be necessary if it were permitted. Interest groups blocking the market through politics — the same pattern already seen in the Tada case.
The fate of DoctorNow (닥터나우), Korea's top telemedicine platform, proves the point. This company, which had accumulated three million subscribers under the COVID special exemption, was forced to lay off 40 of its 80 employees in June 2023 when the government converted the regime to a pilot program and narrowed the eligible scope.
A DoctorNow representative said: "If the innovative attempts of new-industry startups are uniformly restricted without adequate review and communication, the harm ultimately falls on the people who use healthcare." Half a year later the government again eased the regulation to permit first-time night and holiday consultations. A door opened, closed, reopened. Between those changes, the workforce had already been cut in half.
The same structure recurs in the financial regulatory sandbox. Since 2019, roughly 500 designations have been made, but the rate at which formal regulatory revision was completed after the special-exemption period expired stands at only 10 to 17 percent. The share of small and mid-sized fintech companies in the sandbox fell from 62 percent in 2019 to 5 percent in 2025. The sandbox is not permanent permission for innovation — it is a temporary exception, and even that has been distorted into a large-enterprise-centered mechanism.
The K-Chips Act (반도체 특별법) is another variation. In 2024, as Samsung Electronics and SK hynix engineers were pulling all-night shifts to meet NVIDIA's requirements in the HBM development race, a provision exempting semiconductor R&D personnel from the 52-hour weekly work limit had been blocked for more than a year in a ruling-opposition deadlock. In Chapter 11 we saw Korea's two wings — chips and batteries — as the two elements of indispensability. The National Assembly could not pass a single provision to give the engineers who build those wings more flexible working hours.
In December 2025, the Assembly passed the law with that provision deleted. The Seoul Economic Daily expressed it in a headline: "K-Chips Act Without the Filling." The chair of the Industry Committee, explaining the background of the agreement, said: "It is regrettable that the working-hour flexibility provisions including relaxation of the 52-hour limit were not reflected in this legislation. There was a sense of urgency that support for the semiconductor industry could no longer be deferred." The urgency was there. The core was missing.
The bill carried a supplementary opinion: that the Assembly "recognizes the need for working-hour exemptions for research and development personnel in the semiconductor industry, considering its characteristics, and will continue to discuss alternatives in the relevant standing committee." Not a solution — a deferral. During the same period, TSMC secured $52.7 billion in U.S. CHIPS Act subsidies and began operating its Arizona plant.
The three-way fragmentation of cybersecurity governance seen in Chapter 5 is a variation on the same structural problem. In a structure where the Korea Internet and Security Agency (KISA), the National Intelligence Service, and the Cyber Command each guard their own jurisdictions, an integrated response framework cannot precede legislative coordination — the same logic of dispersed interests at work in the AI Basic Act and the K-Chips Act now operates in the security domain.
The pattern is clear. Where law is needed, law is late. When law passes, the core has been removed. Sometimes even after the law passes, it goes unenforced.
Cryptocurrency taxation is the extreme case. The taxation bill passed by the National Assembly in 2020 has been deferred three times — from 2021 to 2023, from 2022 just before the presidential election to 2025, and in late 2024 amid the impeachment crisis to 2027. From passage to implementation: a minimum of seven years. The timing of each deferral aligns precisely with the electoral cycle.
More than ten million cryptocurrency investors — roughly 20 percent of the total population — form a political bloc that blocks regulatory enforcement. Experts project that "implementation even in 2027 is uncertain," citing inadequate taxation infrastructure, investor resistance, and the absence of a data-submission framework for exchanges. If any party judges that enforcing the regulation risks losing votes, it will not push ahead. This is not legislative delay. It is deliberate non-enforcement.
One structural finding about Korea's regulatory system emerges here. In domains directly tied to national security or industrial competitiveness, legislative speed is fast regardless of which party is in power — expanding the semiconductor tax credit to 25 percent under the K-Chips Act is the evidence. But in domains where interests are dispersed — AI regulation, platform labor, cryptocurrency taxation — the pace is extremely slow.
Korea's political system is far less efficient with regulation that has a "diffuse distribution of gains" structure than with regulation that has a "clear loser" and a "clear winner." The incentive structure of the bureaucratic system makes this worse. If a regulator relaxes a rule and innovation grows, the official in charge receives no credit for that outcome. But if an accident follows regulatory relaxation, the responsible official bears the blame. This asymmetry sustains the practice of "regulation as the default, permission as the exception" rather than "permission as the default, regulation as the exception."
3. A Shrinking Country
While institutions move slowly, the population is shrinking. In a country with slow institutions, the population contracts even faster — because a country that spends 20 years on telemedicine and three and a half years on an AI framework law operates by the same logic at the same speed when it comes to changing the structures of child-rearing, housing, and employment.
Korea's total fertility rate in 2023 was 0.72 — the lowest in the world. This is not simply a low figure. Korea is the only country among the 38 OECD members with a fertility rate below one. Against Japan's 1.2 and Italy's 1.2, Korea is at half that level. In the fourth quarter of 2023, the figure hit a historic low of 0.65.
This number began at 6.1 in 1960. In 64 years it has fallen 88 percent. A 2025 study published in the PMC medical journal stated explicitly that the "scale and pace" of this phenomenon are without precedent in human history.
There is an ironic history behind this. In the 1960s the Korean government worried that the population was too large. It promoted birth control under the slogan: "If you keep having children without thinking, you'll never escape poverty." That policy was a resounding success. So resounding that the reverse direction now threatens national survival.
From 2006 through 2024, the government poured approximately ₩378 trillion into low-birthrate countermeasures. That is roughly ₩57 billion spent every day for 18 years. The result is the world's all-time lowest fertility rate.
Why did it fail? Sixty-two percent of the budget was allocated to indirect programs — civilian-official personnel costs, tourism promotion, university development. The Ministry of Culture, Sports and Tourism's "family leisure program development" project was counted as low-birthrate spending. The Ministry of Education's PRIME project budget was also included. Spending directly connected to low birth rates was only 38 percent of the total. Family support spending stood at 1.56 percent of GDP, far below the OECD average of 2.29 percent, ranking 33rd among 38 OECD countries.
The problem was not only that the money was insufficient — the direction was wrong. The government diagnosed the cause of the fertility decline as a "cost problem" and distributed cash. The actual causes were multiple: Seoul apartment prices 25 to 30 times the average annual income, ₩23.4 trillion in annual private tutoring costs with 75.5 percent of students participating, the largest gender wage gap in the OECD, and an extreme imbalance in household labor — women performing 54 minutes of domestic work to men's 17. Money cannot change this structure.
One young person from the MZ generation said in an interview: "I thought that if my child doesn't make it into the upper tier of the labor market, my child could end up dying like that. I cannot leave this kind of world to my child." Choosing not to have children is not selfishness. It is a calculation.
The number of marriages fell more than 55 percent — from 434,900 in 1996 to 191,700 in 2022. Korea's out-of-wedlock birth rate stands at roughly 3 percent, in stark contrast to 40 percent in the United States and 55 percent in Sweden. In a structure where childbirth is dependent on marriage, fewer marriages mean fewer births. Research on baby-bonus programs found that more than 74 percent of the payouts went to couples who were already planning to have a child — the money rewarded existing decisions rather than changing them.
Population decline is not an abstract statistic. It is the reality of military units being disbanded.
Korea's military headcount fell from 560,000 in 2019 to 450,000 in 2025 — a drop of 110,000 in six years. Army enlisted personnel alone declined by more than 100,000 over the same period, from 300,000 to 200,000. Of the division-level and above units that numbered 59 in 2006, 17 have been dissolved.
In February 2024, the recruit training centers at the 1st Division in Paju, Gyeonggi Province; the 9th Division in Goyang; and the 25th Division in Yangju closed their doors. The final induction dates were December 26, 2023 at the 1st Division, December 18 at the 9th Division, and December 19 at the 25th Division. The 28th Division is scheduled for complete dissolution in December 2025. Units that had guarded the DMZ for decades since the Korean War are disappearing one by one before the adversary of low birth rates.
The Defense Ministry requires roughly 200,000 draftees per year to maintain force levels. The standard for active-duty classification was relaxed, raising the active-duty rate from 69.8 percent to 86.7 percent — but the troop decline was not arrested. Around 2039, the pool of eligible draftees shrinks to 180,000. That is the point at which maintaining an army of the current size becomes mathematically impossible — precisely when North Korea is strengthening its drone, AI, and nuclear capabilities.
In Chapter 10 we saw Japan's lost thirty years. Japan's demographic cliff began earlier than Korea's, but Korea is arriving at the same crisis far faster than Japan did. Japan's total fertility rate has never fallen below 1.0. Korea is already the country with the world's lowest fertility rate.
Villages are disappearing. Uiseong County in North Gyeongsang Province has an extinction-risk index of 0.093 — one of the fastest-vanishing localities in the country. Population aged 65 and above: 45 percent, the highest share nationwide. The 50,000-person threshold has been breached. The entire population of Sinpyeong-myeon is 814.
Schools began closing in 1992. Of the 33 elementary schools still remaining, six have fewer than ten enrolled students. Uiseong County ran a pilot program for extended evening obstetrics clinic hours in 2024, at a budget of ₩68 million — every Tuesday and Thursday evening from 6 p.m. to 8 p.m. That was the only solution the local government could offer.
In South Jeolla Province, 20 of the 22 cities and counties — 90.9 percent — are classified as extinction-risk zones. The number of cities and counties at high extinction risk surged from one ten years ago to 13 today. In the border regions from which military units have departed, the drawdown of soldiers' families has pushed Yanggu County and Injae County into new extinction-risk classifications. Population decline generates military contraction, and military contraction accelerates population decline further — a cascade.
Of 250 cities, counties, and districts nationwide, 77 — 30.8 percent — have no obstetric facility whatsoever. Sixty more have exactly one. More than half of all localities are in a delivery-care crisis. The number of delivery-capable medical facilities nationwide fell 40 percent over eleven years, from 706 in 2013 to 425 in 2024. The number of obstetrics clinics in the Seoul metropolitan area grew 11 percent while non-metropolitan areas collapsed. Pregnant women in Uiseong County must travel to Andong or Daegu for childbirth. Even being born has become difficult.
Foreign workers are filling the emptied spaces. As of 2024, E-9 visa foreign workers surpassed 300,000. Annual E-9 arrivals, at roughly 56,000 before COVID, tripled to 160,000 in 2023. Entire workforces in cabbage fields are Thai; entire crews repairing seaweed nets are Filipino. This has become ordinary. One farmer said: "Without foreign workers, we can't grow cabbage."
Korea's foreign resident ratio stands at roughly 5.3 percent (2024). In Chapter 3 we saw Lee Kuan Yew design Singapore's population at 30 percent foreign nationals. Korea is a country that depends on migrant labor while failing to build the immigration system to support it.
Programs to attract high-skilled talent — the K-STAR visa, the Top-Tier Visa — began in 2025. The Top-Tier Visa covers eight advanced industrial sectors including AI, robotics, quantum technology, and space aviation, requiring a master's or doctoral degree from a global top-100 university plus experience at a global company. Intake capacity in the first month: 20 people. The expansion target by 2030 is 350. Compared with Singapore's ONE Pass, the scale and speed belong to different worlds.
One reality must be faced directly here. In a country with a fertility rate of 0.72, the population cannot be fixed by law. ₩378 trillion has proved it. But the talent pipeline can be maintained by opening the national borders. The problem is that Korea refuses to call this "immigration policy." Public opinion on expanding immigration is 30 percent in favor, 55 percent opposed. The ethno-national narrative remains strong. And so Korea's immigration expansion — as Japan's did — will likely have to proceed without using the word "immigration."
Japan provides the precedent. Japan has never formally declared itself an immigration country. Yet after introducing the Specified Skilled Worker visa in 2019, foreign workers surged, surpassing two million for the first time in 2024. Japan constructed a de facto immigration system without touching the political detonator of "immigration." Germany was more direct. In 2023 it introduced the Chancenkarte — Opportunity Card — scoring education, experience, and language ability, and issuing job-seeker visas to those above a threshold. Canada's Express Entry system has been operating for more than 20 years, accepting hundreds of thousands of permanent residents each year.
What Korea needs is not piecemeal visa programs but a systematic talent-intake pathway. The first track is high-skilled: a fast-track visa for semiconductor, AI, and biotech engineers, at a scale of 3,000 to 5,000 per year. The current Top-Tier Visa target of 350 falls short of the foreign-hire demand of a single Samsung Electronics semiconductor division. The second is mid-skilled: a pathway that moves beyond the short-cycle rotation structure of the E-9 visa, allowing workers in manufacturing, caregiving, and services to earn permanent residency after three to five years of verified competency. The current structure, in which Thai workers in the cabbage fields remain "temporary" after a decade, is failing on both labor supply and social integration. The third is a regional-settlement track: a visa incentivizing international students and skilled workers to take up permanent residence in extinction-risk localities like Uiseong County. Annual international students at Korean universities already exceed 200,000 — but only 30 percent settle in Korea after graduation. The remaining 70 percent take their education and leave.
In Chapter 3 we saw Lee Kuan Yew design Singapore as a multi-ethnic society. Singapore was not always favorable toward immigration. After the ethnic riots of the 1960s, it laid the infrastructure of coexistence first — through HDB quota mixing, which mandated ethnic-ratio targets in public housing developments. Korea does not need to become Singapore. But as the foreign-resident ratio rises from 5.3 percent to 8 percent, to 10 percent, failing to design the institutional foundations of social integration in advance means repeating Europe's failures.
The population cannot be fixed by law. But to maintain an indispensable node above a shrinking population, the pipeline must function beyond the national border.
4. A Country That Resets Every Five Years
Where does the slowness of institutions come from? The structure reveals the answer.
Article 70 of the Korean Constitution limits the presidential term to five years, non-renewable. The provision was designed after democratization to prevent the recurrence of authoritarian long rule. But this structure has introduced five-year cycles of discontinuity into foreign policy and industrial strategy. The analysis of the U.S. Congressional Research Service (CRS) is blunt: "The constitutional constraint of a single five-year term creates distinct phases in the Korean presidential cycle, and from the third year onward, as attention shifts to selecting a successor, significant impairment of policy execution occurs. Even though they are limited to a single term, Korean leaders face the risk of lame-duck status from the moment they take office."
The single-term system is not the only cause. The standing-committee structure of the National Assembly deepens inter-ministerial silos by interlocking with departmental parochialism, and the civil-service rotation posting system prevents the accumulation of policy expertise. All of these factors erode policy continuity. But the five-year horizon is the most visible ceiling — it is the source of the inertia that defers long-term strategy to the next administration even when its necessity is recognized.
The discontinuity is clearest in foreign-policy direction. President Moon Jae-in chose "strategic ambiguity" — a posture of supporting the U.S.-centered order while preserving cooperation with China. Korea declared the THAAD Three Nos and held back from joining the Quad.
President Yoon Suk-yeol made a 180-degree turn to "values diplomacy." The Camp David Korea-U.S.-Japan summit declaration, formalization of participation in CHIP4, and the third-party settlement agreement on Japanese wartime forced labor were all concluded. But this turn came at the cost of deteriorating relations with China. With China still accounting for more than 20 percent of Korea's exports, the economic cost of choosing sides was a real pressure.
President Lee Jae-myung then reversed course again to "pragmatic diplomacy," declaring the restoration of the South Korea-China relationship to a "strategic cooperative partnership." Three administrations. Three foreign-policy lines. Eight years. Each administration begins by negating or dismantling the diplomatic legacy of its predecessor. This is not policy adjustment. It is the product of political differentiation.
Foreign Affairs probes the root of this phenomenon: "Power is highly concentrated in the presidency, and deepening polarization generates an ever-fiercer struggle over that power. Digital polarization is fragmenting the public sphere, making cross-partisan consensus virtually impossible."
The Carnegie Endowment for International Peace puts it precisely: "The most urgent task is to stop the ideological pendulum and create a national strategic framework that functions despite changes of government." Foreign partners find it difficult to regard Korea as a predictable actor. When the center of gravity between the U.S. alliance and China cooperation shifts every five years, no partner can have confidence in a long-term commitment. To stand in the between, one must be trusted by both sides simultaneously — but the current structure is one in which gaining one side necessarily means losing the other.
December 3, 2024 was the day this structural vulnerability was exposed in its most extreme form. President Yoon Suk-yeol declared emergency martial law. Within six hours, the National Assembly voted to lift it. The Hankook Ilbo expressed it thus: "Seventy years of carefully built diplomatic achievements collapsed in six hours." The planned visit of Japan's Prime Minister Ishiba Shigeru was suspended; the visit of Sweden's Prime Minister Ulf Kristersson was cancelled; U.S. and Japanese defense-ministerial level schedules fell apart.
In April 2025, the Constitutional Court upheld the impeachment by a unanimous 8-to-0 vote of its justices, stating in its written decision: "The abuse of emergency state authority caused negative effects on the country's international creditworthiness and produced diplomatic and economic disadvantages from the expansion of political uncertainty."
This impeachment crisis coincided precisely with the inauguration of the Donald Trump second administration. The principal party for defense cost-sharing negotiations was absent; the continuity of the Camp David commitment became uncertain; extended-deterrence consultations on the North Korean nuclear threat were disrupted. The Stimson Center's analysis: "The leadership vacuum occurred at the moment when China's military assertiveness in the Indo-Pacific and North Korea's missile threats made Korea's security cooperation more important than ever." The South China Morning Post, in the same period, diagnosed "pervasive foreign-policy uncertainty in Korea's power vacuum."
In Chapter 2 we saw Finland's NATO accession. Abandoning 75 years of military neutrality and joining NATO — a historic transformation — was approved by the Finnish Parliament 188 to 8. More than 95 percent in favor. A cross-partisan consensus. In Finland, the direction of foreign policy is not an element of political identity. Facing the question of national survival, factions that had historically fought one another spoke with a single voice.
Even earlier, in the Cold War period, Finland under President Kekkonen's "active neutrality" doctrine maintained trade with both NATO and the Warsaw Pact while preserving its independence. Hosting the 1975 Helsinki Final Act and contributing to East-West détente was the product of that consistency.
In Korea, North Korea policy, the scope of the U.S. alliance, and the weighting of China relations are themselves the political identity that divides progressive and conservative. To stand in the between, one must not waver. To not waver requires consensus that transcends the political cycle.
In Chapter 3 we saw Singapore's People's Action Party (PAP) guarantee strategic consistency through political continuity. Korea chose democracy, and that choice was right. But even within democracy, strategic consensus is possible. Finland proved it. Korea has not yet done so.
5. Two Clocks
This chapter has examined two crises. The crisis of slow institutions. The crisis of a shrinking population. These two are not separate problems. Each one makes the other worse.
When population falls, the labor force falls; when the labor force falls, growth slows; when growth slows, distributional conflict intensifies. As distributional conflict intensifies, political consensus becomes harder; when consensus becomes harder, institutional reform slows further. Slow institutions delay the AI transition; when the AI transition is delayed, the country cannot absorb the shock of population decline. A vicious cycle.
Applying the core formula, Korea stands at the formula's third stage. In Chapter 11 we saw technological innovation and concentration of capital — HBM, batteries, hundreds of trillions of won invested by the chaebols. Now the symptoms of the formula's third stage — social instability — are clear. Regulation confines innovation; the population shrinks; the military contracts; villages vanish; the national strategy resets every five years. The transition from concentration of capital to social instability is underway. The formula's fourth stage — institutional redesign — has not yet arrived.
Technological innovation → Concentration of capital → Social instability → Institutional redesign
The problem is time. The clock of population decline is already running. Statistics Korea's median projection puts Korea's population at 36 million to 40 million by 2050. Those aged 65 and above will comprise 40 percent of the total. The old-age dependency ratio will rise more than 2.6-fold, from 29.3 in 2025 to 77.3 in 2050. By 2050, 100 people of working age will support 97.5 dependents — in effect, one working person must sustain one non-working person.
Where caregiving labor has disappeared, AI robots have begun to fill the gap. The AI companion robot Hyodol, distributed to 1,230 elderly people living alone, demonstrated measurable reductions in depression and isolation. It showed socioeconomic value 3.7 times the budget invested. A society in which a robot stands in for a grandchild. The paradox of prosperity.
The clock of the AI transition has only just begun to accelerate. If AI raises productivity sufficiently, per-capita GDP can grow even as total GDP contracts. A smaller population may mean faster policy experimentation and more flexible social change. Israel, with a population of 9.5 million, has become a core hub of the global advanced-technology ecosystem — proof of that possibility. Korea already ranks first in the world in semiconductor design and first in 5G penetration. The material foundation for the AI transition is in place.
But for that possibility to become reality, institutions must keep pace. The AI Basic Act in three and a half years. Telemedicine in 20 years. The K-Chips Act passed without its core provision. A National Assembly that made a legal service illegal in 16 days. At this pace, the clock of the AI transition cannot outrun the clock of population decline.
In technology competition, speed is a matter of survival. The mechanism by which unprocessed bills expire automatically upon parliamentary-term expiration under the National Assembly Act means that semiconductor, AI, and energy legislation returns to square one on a four-year cycle. Technology industry investment cycles and legislative cycles are structurally out of alignment.
If Korea completes this transition quickly enough, population decline becomes a manageable challenge. If the population collapses sharply while the transition stagnates, that becomes a genuinely fatal crisis.
Two clocks are racing. The dual crisis of institutional speed and population speed. Which side wins this race will determine Korea's twenty-first century. If chips and batteries are Korea's shield — and Chapter 11 confirmed that they are — the arms holding that shield are thinning, and the speed at which the rules that make the shield larger can be changed is too slow.
In Book 4, Slow Justice, Fast Order, we saw that the speed of institutions divides the fate of nations. In Korea that proposition operates in double. When institutions are slow, the technological transition is late; when the technological transition is late, the impact of population decline cannot be arrested. Institutional speed is, for Korea, a question of survival.
The next chapter examines the shadow that falls over Korea's industrial power from within — the chaebol structure itself.