Kyle Chan

Learning how a country learns

Kyle Chan is an American sociologist who studies how countries acquire the ability to make complicated things. He is best known for explaining Chinese industrial policy, but that description misses the question that gives his work its continuity: how do organizations turn resources, knowledge, and ambition into something that actually functions? His objects have changed from railways to electric vehicles, semiconductors, and artificial intelligence. The recurring problem is the distance between an intention and a working system. He is now a fellow at the Brookings Institution’s John L. Thornton China Center and the author of High Capacity, a newsletter accompanied by a podcast.¹

His career has itself crossed several systems. Economics supplied one set of tools; sociology made room for questions those tools did not settle. Travel in China and India turned development from an abstraction into a daily experience of unequal possibilities. Doctoral fieldwork required access to officials, construction projects, and documents that did not advertise themselves as an accessible research collection. A university supported the slow accumulation of this knowledge. Later, newsletters, interviews, and policy institutions made it usable at the much faster speed of public argument.

The resulting figure is neither adequately described by his institutional appointments nor reconstructed by assembling his opinions about China. What matters is how he learned to change the scale of observation: from an individual official’s fear of making a decision to the organization that generates that fear; from a firm to its suppliers; from a successful product to the industrial environment that makes a succession of products possible. The strongest interpretation of his vocation is that he tries to make collective competence visible. The difficulty, increasingly central to his public work, is that competence in building something does not settle whether it should be built, who benefits, or what it costs.

A development question before a China specialization

Chan grew up in Los Angeles in a family of Hong Kong immigrants. His parents chose Mandarin rather than Cantonese for his language education because they thought it would be more useful. In his telling, the choice eventually proved consequential for research. It is a small example of a much larger feature of this formation: a family connection to China did not automatically provide a finished professional identity. Language was an acquired resource, directed toward a future whose precise form was not yet known.²

His interest sharpened during college visits to China. He remembered the speed at which highways and buildings appeared, but also polluted air and exhausted migrant workers. After college, he volunteered for several months at an NGO in Delhi before entering management consulting. India’s ambitions and difficulties made the comparison unavoidable. Two enormous, rapidly changing countries were confronting development with very different organizational arrangements. The difference seemed too consequential to explain merely by labeling one authoritarian and the other democratic.³

This matters because the original problem was not American strategic competition. It was why people born in different places have such radically unequal prospects. In a later conversation with the technology commentator TP Huang, Chan connected his interest to inexpensive travel through different countries and the contrast between strenuous work and limited opportunity. Infrastructure was an especially tangible part of that inequality: reaching work, moving goods, or obtaining reliable power could be an everyday struggle. He described China and India as ways into the larger development question, rather than development as background for a pre-existing fascination with China.⁴

The distinction changes how his later writing should be read. A researcher concerned principally with identifying an adversary’s strengths might see a railway as evidence of national power. A development researcher must also ask how the railway changes the range of possible lives. Chan’s eventual move into Washington has not erased the second concern, although the public settings in which he now speaks often foreground the first.

He studied economics at the University of Chicago and political sociology at the London School of Economics. A contemporaneous Peking University seminar biography confirms those disciplinary identifications, which are more revealing than a list of university names alone.⁵ Economics and sociology were not simply successive credentials. They offered different ways to decide what counted as an explanation.

In his account to the CSIS AI Policy Podcast, economics provided valuable statistical tools, while sociology allowed him to combine quantitative material with corporate records, government documents, and interviews. It also legitimized questions about how the parts of a bureaucracy fit together.⁶ The move was an expansion of method, not a claim that quantitative evidence was dispensable. It was possible to retain the demand for rigor while becoming dissatisfied with an analysis that left the organization itself largely unopened.

Consulting belongs in this history too. An older author biography at the German publication The European describes management-consulting work in San Francisco and London through 2012 and writing about economics and American politics.⁷ The available record does not establish a complete client history or justify turning consulting into the secret cause of everything that followed. It does, however, prevent a misleadingly seamless story of undergraduate study followed immediately by an academic career. He had worked in another profession before deciding to return to sustained research.

Princeton and the social conditions of inquiry

The acknowledgments to Chan’s 2023 dissertation, Fast Track, reveal his intellectual apprenticeship. Reading Atul Kohli at LSE helped persuade him to pursue sociology. His adviser, Yu Xie, assisted with Chinese fieldwork and scholarly introductions. Miguel Centeno discussed Weber and helped develop manuscripts. Patricia Fernández-Kelly became a supporter through teaching together.⁸

Kohli’s role connected the initial development question to an established comparative tradition. Xie’s role shows something equally important about research training: an adviser need not reproduce himself in a student. A scholar can provide protection, criticism, access, and confidence without sharing every method or object. Centeno and Fernández-Kelly, meanwhile, place Chan’s formation inside conversations about states, cities, inequality, and the practical responsibilities of teaching. The eventual technology analyst emerged from a much broader sociology.

Institutional support was concrete. Princeton’s Center on Contemporary China records a 2018 research award for his high-speed-rail project.⁹ His later postdoctoral appointment connected the Sociology Department with both the China center and the M. S. Chadha Center for Global India.¹⁰ Those affiliations supplied a place for a comparison that would have been awkward inside an exclusively China-focused account of the world.

The dissertation’s personal acknowledgments add another dimension. Chan recalls his father’s weekly library trips and his mother’s example of compassion. His wife, Thalia Gigerenzer, shared the decision to undertake Princeton doctorates and fieldwork abroad. He credits Raine Daston’s comments and Gerd Gigerenzer’s advocacy. Scholarship appears here as a relational undertaking.¹¹

These details should not be converted into a mythology of predestination. A childhood full of books does not explain a theory of bureaucracy, and an intellectually rich circle does not do the fieldwork on the researcher’s behalf. The point is more exact: a long project requires conditions under which an uncertain question can remain open. The time to read, the confidence to keep investigating, the introduction that produces an interview, and the criticism that makes a manuscript publishable are all part of the production of knowledge.

By January 2019, Chan was presenting the China–India comparison at Peking University’s Center for Social Research, under the Chinese name 陈凯欣. The seminar already distinguished concentrated organizational responsibility from overlapping authority.¹² A Princeton presentation in March 2023 returned to the same puzzle. The continuity is striking beside the pace of his later public commentary: the central railway argument developed over years rather than appearing fully formed in a timely essay.¹³

Getting inside the railway

Railways offered an unusually demanding test. A train journey looks like a single service, but producing it requires land, finance, civil engineering, equipment, safety systems, operating institutions, and coordination across jurisdictions. A failure at one connection can frustrate competent work elsewhere. The research problem was therefore not simply why China spent more money. It was how decisions and responsibilities were arranged so that a project could move through this sequence.

Chan’s dissertation places his main fieldwork in 2016–2019, based chiefly in Delhi and Beijing. It records difficult access in China, introductions through railway academics, and assistance from Delhi’s Institute of Rail Transport. Wu Jianhong, Cheng Zhongxin, and Pramod Uniyal helped make inquiry possible. He also bought and digitized more than a dozen Chinese railway project books.¹⁴

This is a useful corrective to the impression that an American university affiliation opens every door. Access was worked for, relationship by relationship.

A December 2022 appearance on The Belt and Road Podcast made the research process itself part of the discussion: riding trains through both countries, following their bureaucracies, and deciphering the unglamorous names of Chinese companies. His recommendations were the documentaries The Chinese Mayor and Powerless, both concerned with the lived politics of infrastructure and administration.¹⁵ Even the cultural recommendations sit close to the same interest: what public power looks like when it encounters a place and the people trying to live there.

The published study of Chinese state-owned enterprises reveals the documentary labor behind the visible travel. It draws on interviews with industry participants and thousands of pages of project records, contracts, bond prospectuses, and corporate material. Chan separated parent corporations from operational subsidiaries and followed the allocation of managers, equipment, and technical knowledge. The object was the working organization, not the name at the top of an annual report.¹⁶

This way of working has an important advantage. Documents produced for financing, procurement, or internal administration often expose relationships that disappear in a public speech. An institutional slogan announces a purpose. A contract allocates responsibility. A prospectus must explain an entity to creditors. A personnel record shows where expertise moves. Their different purposes make comparison between them valuable.

Yet documents and access also set boundaries. Interviews are encounters with people who agreed to speak; administrative records make some forms of action more visible than others. The resulting account cannot be treated as an unobstructed view from nowhere. Its strength lies in asking whether several kinds of evidence converge, and in specifying what the comparison can actually establish.

That qualification is especially important for China and India. A comparison of their railway bureaucracies is not a controlled experiment in regime type. It cannot make differences of historical inheritance, fiscal resources, geography, politics, and social conflict disappear. The organizational argument is valuable because it adds mechanisms to that larger picture. It would be weakened, rather than strengthened, by being made to explain the entire picture alone.

The discoveries beneath the megaproject

A sociologist beyond industrial policy

Chan’s interests during graduate school were wider than railways. His March 2020 talk on “Hyper-Quantification in China” considered social credit, customer programs, employee evaluation, education, and matchmaking. It placed contemporary Chinese digital life in a sociological conversation reaching back to Simmel and Weber. The question was how numerical classification spreads across institutions and everyday experience, with China potentially at the forefront of a process often imagined as originating in the West.²²

The connection to the railway work is more than a shared country. Both ask how apparently neutral systems of organization reshape what people can do. A measure is not merely a description of an activity once it becomes the basis for evaluation, reward, or access. A bureaucratic chart is not merely a picture once it governs whose approval is necessary. Chan’s breadth is most coherent at this level of inquiry, rather than as an ever-growing list of fashionable technologies.

Teaching supplied another setting for this interest. Princeton lists him as the instructor for Introduction to Sociology in fall 2024. The course description emphasized the social structures behind apparently individual choices and introduced varied conceptual and methodological tools.²³ Earlier, the spring 2021 Social Networks course listed Chan alongside Matthew Salganik for office hours; assignments included replicating studies and writing newspaper-style editorials.²⁴

These records establish teaching responsibilities, not student evaluations or a complete account of his pedagogy. They nevertheless matter to the career. A public explainer’s ability to move between a general proposition and a concrete example can be developed in places where the audience has not already accepted the vocabulary. The classroom is one such place. His later newsletter did not require him to begin learning that task from zero.

High Capacity and a different scale of explanation

By March 2024, High Capacity was carrying essays on industrial development. “How China Uses Foreign Firms to Turbocharge Its Industry,” published that month, treated foreign investment as a route to learning across a domestic production system. The key move was to follow what a multinational’s presence did for suppliers, skills, and manufacturing capability, rather than count the factory as a self-contained addition to output.²⁵

The railway research had opened this question. A country can buy a technologically advanced object without becoming able to reproduce, improve, maintain, and expand the system that makes it possible. Learning is therefore an organizational process as well as a transaction. Chan’s public writing progressively made that distinction the hinge between older infrastructure work and newer industrial analysis.

His May 2024 Apple essay is particularly useful for seeing the method in action. He distinguished the nationality of a supplier from the location of its production, and the number of suppliers from their contribution to product value. He cross-referenced Apple’s supplier information with reporting and company-level evidence about manufacturing facilities. The apparent question, whether Apple was leaving China, became several different questions about ownership, location, components, and upgrading.²⁶

That decomposition is the actual service provided by the essay. It gives the reader a way to ask a better question before reaching a geopolitical conclusion. “Made in China” can refer to assembly, a domestic company, a foreign-owned plant, or a much larger accumulation of engineering know-how. Each implies something different about what might move and what would be difficult to reproduce elsewhere.

The January 2025 essay “China’s Overlapping Tech-Industrial Ecosystems” enlarged the frame again. Chan traced shared suppliers, demand, technologies, and scale economies across batteries, electronics, vehicles, sensors, and robotics. A battery producer could learn from and sell into more than one sector; an industrial capability could strengthen several downstream activities. The accompanying sketch made those connections visible. Chan credited a jigsaw-puzzle analogy from Dylan Patel on ChinaTalk, locating even the explanatory metaphor in a conversation rather than presenting it as wholly self-generated.²⁷

The essay’s significance for his career is that it gave a portable visual form to years of organization-centered thinking. A reader did not need to know the details of railway administration to see why capabilities accumulated in one place could matter elsewhere. The diagram could travel between a newsletter, a presentation, an investor’s analysis, and a policy report more readily than a dissertation chapter.

That portability also changes the risk. A diagram makes connections legible by leaving things out. It cannot, on its own, distinguish a strong spillover from a merely plausible one, a productive redundancy from expensive duplication, or a mutually beneficial relationship from dependence. The best use of Chan’s ecosystem perspective is as a guide to investigation. The weakest use would turn the existence of connected boxes into proof that the system inevitably wins.

An April 2024 essay on the relationship between carbon-intensive heavy industry and clean technology added an uncomfortable complication. Steel and aluminum production could supply the materials that made renewable-energy equipment and electric vehicles possible while remaining major sources of emissions.²⁸ This was a characteristic refusal to let an attractive category, “clean tech,” erase the industrial processes behind it. It also showed why the larger argument cannot be reduced to admiration for spectacular finished products.

Making research available to other people

High Capacity is best understood as a working publication rather than a secondary outlet for already completed academic findings. Its essays test categories, connect reporting from different sectors, draw charts, and give readers routes into further material. The publication’s own description reports more than 15,000 subscribers in over 140 countries, including officials, researchers, journalists, policy analysts, and investors. These are self-reported audience figures, not an audited statement of revenue or influence.²⁹

That audience mixture helps explain the writing’s level of detail. An investor, a scholar, and a congressional staffer may all need to understand the same industrial relationship, but they will use it for different purposes. Chan’s task is to preserve enough mechanism that the explanation remains meaningful after it travels. A vocabulary built entirely for insiders would shrink the audience; a sequence of dramatic conclusions would give the audience little to work with.

A June 2025 collaboration with Sinification shows another form of mediation. Thomas des Garets Geddes invited Chan to introduce a condensed translation of an interview with Peking University scholar Lu Feng; Ailsa Brown was also credited on the piece. Chan called the position “Chinese industrial maximalism.” The division of labor is important: he supplied the framing, while the translated interview carried Lu’s argument. Publishing and interpreting an argument does not establish that Chan adopts every claim it contains, still less that Lu is his personal mentor.³⁰

The episode places him in a contemporary English-language China-reading world where newsletters function as small editorial institutions. Translation, selection, framing, cross-posting, and specialist conversation become ways to enlarge what an audience can encounter. This is not the same as extracting a quotation from China to illustrate an American argument. At its best, it lets a Chinese intellectual’s account retain enough structure to be examined on its own terms.

Chan’s ChinaDocs database makes the infrastructural side of this role especially clear. The public collection identifies him as its manager and organizes official Chinese documents by title, date, issuing entity, Chinese name, and links or files in Chinese and English. Entries distinguish official translations from unofficial ones. It also points readers toward other established document and translation collections.³¹

A database is a different kind of public contribution from an opinion column. It reduces the cost of checking a claim and allows other people to formulate questions the compiler did not anticipate. The value persists even when a reader disagrees with Chan’s interpretation. It is therefore one of the strongest examples of his practice extending beyond commentary.

In March 2026, he used successive Five-Year Plans to track the recurrence and changing language of technological priorities. “China’s Technology Long Game” combined historical reading with keyword charts, and linked the underlying document collection. Its account emphasized persistence and adaptation rather than an immaculate master plan. The essay also acknowledged a correction from Jonathan Sine concerning institutional chronology.³²

This is a revealing combination of scholarly and online habits: inspect a documentary series, make the comparison visible, publish while the issue is live, and correct the record in place. It differs from the closed form of a completed journal article. Its quality depends on retaining the discipline of the first activity while accepting the speed and revisability of the others.

From a research audience to a policy audience

The year 2025 brought several kinds of institutional recognition together. Chan testified before the U.S.-China Economic and Security Review Commission on February 6. The session placed him on a panel with Liza Tobin and Barry Naughton, discussing the effectiveness of Beijing’s approach to innovation.³³ He was also selected for the Penn Project on the Future of U.S.-China Relations, whose third-cohort Washington workshop included his work on lessons from Chinese clean-technology policy.³⁴

These were not equivalent relationships. A hearing makes a scholar a witness and places his claims under questions from public officials. A fellowship creates a setting for developing policy work with a cohort. Neither makes him the author of government policy. Precision about those roles is particularly important when public biographies compress them into the general language of influence.

His June 2025 RAND publication, Full Stack, was explicitly collaborative. Chan wrote with Gregory Smith, Jimmy Goodrich, Gerard DiPippo, and Konstantin F. Pilz. It mapped Chinese AI policy across hardware, computing infrastructure, research, models, and applications, while identifying bottlenecks and possible waste. RAND disclosed that the work was independently initiated within its China Research Center and supported by a gift from Michael Tang, other supporters, and operating income; it stated that donors did not control findings. That is evidence about the project’s support, not Chan’s personal compensation.³⁵

The collaborative form matters. A single-author voice can connect sectors, but a report on rapidly changing technology also requires specialized knowledge and review. Full Stack should not be folded into a story in which Chan personally possesses every kind of technical expertise its subject demands. Its author list is part of how the work becomes possible.

Princeton’s 2025 China-center annual report records a book contract with Princeton University Press, alongside his public writing and policy work.³⁶ His own current website still describes a book in progress about Chinese tech-industrial policy.³⁷ The distinction between contract and publication must remain intact. The larger synthesis is an ongoing project, not a completed book whose argument or reception can already be assessed.

Chan joined Brookings in October 2025. In his introductory interview with China-center director Ryan Hass, he described a move from studying railways toward patterns of learning across technological sectors.³⁸ The appointment did not create the underlying expertise. It provided a new institutional location for making it available to policymakers and the press.

One can therefore see a sequence of conversions without assuming that any step was inevitable. Fieldwork became academic arguments. Those arguments supported broader industrial interpretation. Public explanation helped make the expertise visible beyond a discipline. Hearings, collaborative reports, fellowships, and finally a think-tank appointment put that expertise into settings organized around current decisions. Each conversion increased reach while changing the demands placed on the research.

Conversation as a way of extending expertise

The High Capacity podcast began in January 2026 with Georg Stieler, a consultant with extensive experience in Chinese robotics and industrial automation. Chan asked about changing clients, production requirements, the shift from conventional cars to EVs, and the distinction between a robot’s visible form and the work needed to deploy it. The interview was built around experience in factories and supply relationships, not simply predictions about humanoids.³⁹

This is a productive answer to the problem created by his increasingly wide scope. He does not need to pretend that comparative sociology makes him the leading specialist in every component. He needs enough understanding to identify which specialist can answer which question, and to connect the answer to problems his wider audience can recognize.

A conversation with Paul Triolo about semiconductors followed relationships between toolmakers, chip manufacturers, designers, and AI customers. Chan’s questions repeatedly drew out feedback between adjacent parts of the production system. The podcast also directed listeners toward Triolo’s own writing.⁴⁰ Hosting becomes a form of research in public: the interviewer has a hypothesis about what connects the pieces but invites someone with different experience to fill in, qualify, or redirect it.

The guest list also prevents the show from being understood as a series of interchangeable technology enthusiasts. Abigail Coplin brought a sociologist’s account of biotechnology and the knowledge that firms and scientists supply to policymakers. Ruby Wang brought medical and healthcare-strategy experience. Grace Shao discussed the differing strategies of Chinese AI companies. Each conversation entered a world whose practical knowledge could not be generated from national policy documents alone.^{41, 42, 43}

Chan’s exchange with Kevin Xu about open source explored a further organizational problem: attracting contributors can require a company to relinquish some control. An ecosystem has to be cultivated, and outside participation cannot simply be commanded into existence.⁴⁴ This is close to the old railway question in a new setting. An organization may have resources and a strategic objective while lacking the relationships that make its preferred arrangement work.

His appearances on other people’s programs are equally revealing. In a ChinaTalk discussion of Patrick McGee’s Apple in China, Chan served as a cohost, connecting the company history to the movement of industrial development across East Asia. He asked whether China’s manufacturing strength would leave room for India to follow the same path.⁴⁵ The development question remained active beneath the better-known American–Chinese rivalry.

Kaiser Kuo’s February 2026 Sinica introduction described Chan as an unusually frequent recommendation by other guests. The episode pointed listeners toward Katrina Northrop and Grace Shao; Chan’s reading recommendations included WIRED’s Made in China newsletter and The Wire China.⁴⁶ That record establishes a circulation of attention and practical intellectual affinity. It does not justify describing every recommended writer as a close collaborator.

The economics and limits of a public research practice

There is no adequate public basis for estimating Chan’s income, the profitability of High Capacity, a book advance, or the division of his time between institutional work and independent publication. Subscriber counts cannot supply those answers. Nor does a prominent byline establish whether an article was paid, how much preparation it required, or what other work subsidized it.

What can be reconstructed is the institutional sequence that supports the practice: graduate research assistance, a postdoctoral position, teaching, an adjunct research affiliation, collaborative policy publication, a think-tank appointment, and an independently identifiable public outlet. The newsletter and podcast are valuable partly because they let him maintain continuity across those positions. The evidence supports a mixed institutional and public-facing practice; it does not support a financial portrait of a self-sustaining media business.

The distinction is not pedantic. Public explanation often looks cheap because the reader encounters a short essay. Behind it may stand years of training, difficult field access, unpaid reading, subscriptions, relationships with specialists, and an employer that makes some of the time possible. Conversely, public visibility can produce opportunities that support future research. Both directions are plausible in Chan’s career, but their financial magnitude is undisclosed.

A more demonstrable kind of return is the reuse of his work. The U.S.-China commission’s 2025 report cited his testimony and academic writing, and reproduced his overlapping-industries diagram with attribution. The commission expressly cautioned that the diagram illustrated a concept rather than precise degrees of overlap between industries or companies.⁴⁷ This is concrete evidence that his explanatory work entered an official analytical document. It is stronger than saying that he “shaped the debate,” and narrower than claiming that he caused a policy decision.

His practice also depends on an ecosystem of other people’s labor. Financial and technical journalists identify companies and events; translators make difficult texts accessible; industry specialists explain operational constraints; academics supply historical and comparative arguments. Chan’s skill lies partly in relating this material without making every contribution disappear into a single expert persona. Citation, guest credit, and links are therefore not ornaments around the work. They preserve the structure on which its authority depends.

What changes when explanation becomes advice

In his April 2026 congressional opening statement, Chan distinguished three meanings of winning in AI: leading models, a strong supporting technological system, and gains for ordinary Americans. He asked policymakers to consider jobs, energy costs, and the distribution of benefits alongside technical leadership.⁴⁸ His written testimony was delivered on April 16; the newsletter version followed on April 22.⁴⁹

The sequence brings the career’s original concern back into view. A technology can be impressive without improving the lives of the people expected to support it. That is the same gap between aggregate achievement and lived conditions that made development compelling to him in the first place. The public language has changed from infrastructure provision to technological competition, but the question of what capacity is for has not vanished.

His January 2026 discussion of a “great tech reversal” similarly complicated a one-directional account of learning. The argument was that technological knowledge now also travels from China toward Western companies in selected fields.⁵⁰ As an interpretive move, it follows from the earlier attention to suppliers and cumulative capabilities. As advice, it raises a more contested problem: how should another country learn from a competitor while managing dependency and security risks?

Chan’s writing is most useful when it keeps such tensions open. In April 2024, he argued that successful Chinese EV manufacturers would not necessarily eliminate other automakers. Shared suppliers, technology partnerships, consumer attachments, and policy responses could alter the outcome. He explicitly acknowledged the weakness of treating Japan’s earlier auto expansion as a straightforward precedent, since the geopolitical relationship was different.⁵¹ The essay is an argument about possible mechanisms and contingent responses, not an already verified prediction.

His essay on overcapacity likewise set consumer and climate benefits alongside jobs, distributive conflict, and other countries’ industrial ambitions. It recognized the costs and risks borne within China and argued that Chinese firms and policymakers needed to take foreign concerns seriously.⁵² Yet its sharp presentation of competing worldviews also shows the compression inherent in newsletter argument. Real policy disputes combine macroeconomic balances, security, industry structure, employment, and environmental goals. No single contrast can resolve all of them.

By August and September 2026, Chan was writing more directly about Chinese perceptions of AI risk and the prospects for cooperation. His climate comparison argued that safety concerns would become effective in China when understood as domestic priorities rather than merely foreign demands. A subsequent newsletter directed readers to his New York Times essay and New Yorker interview on the subject.^{53, 54}

This development should not be read as an abandonment of capacity-building for an unrelated ethical subject. It asks how institutions recognize a problem, what incentives make it actionable, and whether international cooperation can become compatible with domestic objectives. Those are organizational and political questions of the kind he has long pursued.

There remains a danger in making industrial capability the privileged window onto a society. Workers can appear chiefly as skills; universities as sources of talent; cities as productive clusters; politics as a mechanism for coordinating investment. The human meaning of those institutions exceeds their contribution to output. Chan’s own earlier work on bureaucratic emotion and his later insistence on broadly shared benefits provide resources for resisting that narrowing. They are most valuable when applied to his favored examples as rigorously as to their competitors.

An unfinished vocation

Chan’s career is still young enough that its current visibility can easily distort the proportions of its formation. The well-circulated explanation of a new Chinese technology rests on a much slower apprenticeship in asking how organizations work. India, graduate teaching, the sociology of emotion, and the difficult acquisition of railway evidence belong near the center of the account, not in a preliminary paragraph before the “real” career begins.

The work’s coherence lies in a recurring refusal to treat the completed object as self-explanatory. A fast train does not explain the authority needed to build a railway. A national champion does not explain the subsidiaries that do the work. An advanced AI model does not explain the power supply, chips, researchers, and users on which it depends. A widely read expert does not explain the institutions and relationships that made his knowledge possible.

That last parallel is an interpretation of the career, but a useful one. Chan’s own practice becomes clearer when subjected to the same scrutiny he brings to industry. Its capacity is distributed across teachers, interlocutors, documents, coauthors, editors, institutions, and audiences. His contribution is to bring certain relationships into focus and give them an intelligible form.

The unresolved test is whether that form remains open to correction as its audience and consequences expand. A strong explanation should help readers inspect a mechanism, not merely exchange one confident national story for another. At his best, Chan equips them to ask what is being built, who can actually make it happen, how learning travels, and who will gain from the result. Those questions give the public technology analyst continuity with the young researcher who wanted to understand why effort and opportunity were so unevenly distributed across the world.

Notes