Alan Chan

Tainan: investigating mathematics with other people

Chan’s account of an early science-fair project begins with a teacher, a visiting professor, and two other students. A geometry puzzle involving rotations and buried treasure, demonstrated with GeoGebra, became a collaborative investigation. The group varied the configuration, looked for invariants, worked toward proofs, and repeatedly revised the written presentation. Teachers supplied time outside ordinary classes and space in which to work. His recollection includes playing games as well as preparing for judges. The project won first prize in mathematics at Taiwan’s national science fair in 2012.³

This was a form of independence produced inside an educational institution. A student could pursue a question beyond the normal syllabus because adults made room for it and peers sustained the work. The work also had several different standards of success: a construction that behaved interestingly on screen, an explanation that survived mathematical scrutiny, and a report that other people could understand. Later accounts of self-education sometimes foreground the solitary reader. This earlier scene shows a more distributed arrangement of permission, equipment, criticism, and companionship.

His public project record traces the rotation research into three dimensions and lists a third-place award at the Taiwan International Science Fair in 2014. It also records later calculus materials and lectures on relativity. These were attempts to make mathematical and physical understanding communicable, alongside the effort to acquire it. They belong to his formation as a maker of explanations.⁴

Physics competitions and university mathematics before university

Physics offered another route out of routine schoolwork. In a 2015 essay, Chan describes struggling with Olympiad questions, studying independently, narrowly missing an earlier selection threshold, and subsequently reaching the national selection camp as a reserve. The camp altered his estimate of his own ability because he met peers whose capacities surprised him. Competition also mattered instrumentally: admission opportunities could free time that would otherwise go into conventional entrance-exam preparation.⁵

The distinction between a national selection result and an international medal matters. Chan’s later biography reports a top-ten national selection ranking. The contemporary account supports a reserve position, not a claim that he won an international Physics Olympiad medal. What the episode establishes more securely is his early use of an institution’s selective routes to obtain greater control over what he studied.

At Tainan First Senior High School, the science-class arrangement also enabled him to take courses at National Cheng Kung University. A January 2016 reflection describes mechanics, electromagnetism, and advanced calculus. The calculus course was especially consequential: proof and the construction of mathematical structures became attractive in their own right, rather than merely serving the solution of physics problems. He recalls the movement toward topology and more abstract spaces as a change in what counted as an interesting question.⁶

His dissatisfaction with later undergraduate requirements therefore did not begin as an aversion to demanding formal study. He had encountered teaching that expanded his ambitions. One source of conflict was the mismatch between that experience and a degree’s standardized sequence. The existing system had given him both unusually rich opportunities and reasons to object when its next stage offered less flexibility.

Football, writing, and the work of belonging

School football occupies a different register in Chan’s writing. In his account of a class tournament final, he remembers the responsibility of being captain and the missed penalty that accompanied a runner-up finish. The disappointment concerned a shared undertaking: the game mattered because teammates had trained and expected things of one another.⁷

A later football reflection is more specific about the labor behind a team. Organizing training and resolving disagreements had become burdensome. Chan describes changing how he played and taking inspiration from a teammate, including a change of shirt number. Read beside his academic writing, these pieces reveal different kinds of competence. Understanding a proof does not settle how to carry responsibility for others, enjoy playing with them, or respond when collective effort does not produce the hoped-for result.⁸

Writing supplied a recurring place to revisit such experiences. In August 2021, he looked back on monthly blogging since June 2013 and reported that the site had passed a million views. The same essay recalls joining a self-learning group in 2016 through a friend. Members gave presentations and pursued projects across technology, human-computer interaction, and business; its membership and work evolved over time. This was an important setting for learning to make an interest legible to peers, and later for finding collaborators.⁹

The blog should consequently be read as a continuing practice, not just a publicity channel created for Heptabase. It is also an interested source. Chan chooses which episodes to tell and often reconstructs earlier decisions using distinctions developed later. Contemporary student essays, retrospective accounts, and current self-descriptions can illuminate one another without being collapsed into a single stable account of his personality.

National Taiwan University, 2016–2017

Chan entered National Taiwan University in 2016 and pursued physics and mathematics. His first-year reflection describes discovering programming through Udacity’s introductory computer-science course. Building a working program offered a different relation to knowledge from reconstructing physical laws: one could assemble relatively simple operations into something that had not existed before. He also describes repeatedly revising an elaborate course plan. At this stage, the effort to direct his education still took place partly through the university’s timetable.¹⁰

The scale of his planning is worth noticing without making it a diagnosis. He was trying to reconcile several subjects with limited time, prerequisites, and credit rules. Learning programming enlarged the set of things he wanted to make, while his mathematical interests exceeded what he regarded as the instrumental mathematics of the physics curriculum. A degree that looked coherent from a department’s perspective could look inefficient from the perspective of his particular combination of prior knowledge and new interests.

In December 2017 he publicly explained his leave from NTU. He objected to required courses, restrictions on course substitution, and the expenditure of time on material he believed he already knew. The criticism was directed at the organization of learning and credential requirements rather than a claim that all university teaching was worthless.¹¹

His longer argument used a deliberately provocative contrast between becoming oneself and becoming a standardized component in a system. It proposed work on actual projects as an alternative means of intellectual development. That polemical form helped the essay travel, but it also made his individual case easy to mistake for a universal prescription. The relevant biographical fact is narrower: a highly prepared student with previous access to advanced teaching decided that the current bargain was poor for him.¹²

A subsequent question-and-answer essay complicates the image of a student interested only in technical subjects. He mentions auditing literary studies, investigating hip-hop, talking about philosophy and cooking with roommates, and approaching people outside his own field. His mother taught Chinese in a university general-education setting, and he explicitly acknowledged the care such teaching could involve. His objection was to compulsory definitions of breadth, not to the humanities or to every general-education course.¹³

Programming projects and an education assembled around them

Chan’s early software work was varied in scale and purpose. Beact, developed with Vibert Thio and Joey Huang in 2017–2018, was an audiovisual instrument in the browser. It joined musical controls, visual response, and the technical machinery required to save and sequence what a user made. NTUAA.io, a multiplayer project for the NTU Tainan student community’s 2018 event, brought game logic, animation, and a networked audience into the same experiment. These were social and expressive projects, not early versions of a notes application.¹⁴

Working on such artifacts changed the practical meaning of self-education. A missing skill could become visible as an obstacle in a project: an interaction that did not work, state that could not be preserved, or a server that had to coordinate multiple players. The object being made supplied a reason to learn and a partial test of whether the learning had worked. It also exposed the learner to limitations that a successful exercise might hide, including usability and the difference between a demonstration and a maintainable tool.

In an October 2018 essay, Chan described understanding as something that had to become concrete enough to be grasped and used. He moved between examples from computing and other subjects, looking for structures that could transfer across contexts. The interest was already broader than efficient recall. He wanted ways of noticing that differently named things could share an underlying organization.¹⁵

The High Quality Resources Collection, another project from this period, organized useful learning material through contributions. Chan selected submissions, and people whose resources were accepted received access to the compilation. He declined requests to sell it because payment might undermine participation. The arrangement made curation and contribution part of the same system: access was connected to helping improve what others could learn from.¹⁶

Mission, military service, and the decision to try Minerva

The freedom to choose a project did not immediately answer what to devote a life to. In a June 2019 essay, Chan recalls a period of uncertainty about purpose and investigations into religion, history, psychology, neuroscience, and philosophy. He distinguishes a profession from the skills it uses and from the larger purpose it might serve. He also describes keeping a bullet journal after military service and returning to it periodically to review what he was doing.¹⁷

These reflections are neither a settled philosophical system nor evidence that he had discovered the eventual company in advance. They show a young adult trying to establish a scale at which choices could be compared. A job title seemed too narrow; an abstract ambition could become too large to guide daily action. The journal and the projects were ways of moving between those scales.

Minerva appealed as a different educational arrangement rather than simply another prestigious credential. His 2019 account of the decision describes conversations that preceded applying and an interest in an institution built around a more deliberate model of learning. The willingness to enroll again is important: leaving NTU did not end his interest in organized education or in the people who gather around it.¹⁸

Chan formally ended his NTU enrollment after the period of leave and began at Minerva in San Francisco in 2019. His November 2020 account of leaving Minerva is notably appreciative. He valued its intellectual habits and the people he met. The conflict was again time, but with a different texture: coursework and work-study competed with startup experiments and relationships. He estimated roughly thirty-five weekly hours for the formal obligations and described the absurdity of trying to attend to class and a startup meeting simultaneously.¹⁹

The two departures therefore should not be narrated as repetitions of the same rejection. The first produced a public argument against curricular constraints. The second involved giving up something he liked because its obligations no longer fitted the work and relationships he wanted to prioritize. Neither decision is adequately explained by the label “dropout.”

San Francisco friendships and the costs of moving

A 2026 retrospective supplies material details absent from the familiar story of academic departure. After leaving NTU in 2017–2018 and losing student accommodation, Chan spent several months in a small software-consulting firm’s office. In San Francisco, he and a close friend exercised and cooked together, including preparing lunch boxes for other students. These recollections make visible housing, food, companionship, and the help of other people, rather than presenting freedom as something obtained by will alone.²⁰

Leaving San Francisco during the pandemic also separated him from friends. He describes maintaining regular video calls with three of them over years and later reunions in Taiwan and New York. The relationship endured through a recurring commitment, not continuous residence in a single entrepreneurial center. This evidence matters to his later life in Britain: geographical mobility expanded possibilities but created maintenance work of its own.²¹

It would be misleading to turn the international trajectory into a frictionless story of a globally mobile founder. The relevant units are often ordinary arrangements: somewhere to sleep, a person with whom to cook, a scheduled conversation across time zones. His writing records these as parts of the life, not simply inputs to the business.

Reading computing history and finding a project, 2020–2021

By 2020, Chan was publishing a more explicit account of the technological work he wanted to undertake. “The Context” set it within an exceptionally large frame, connecting human understanding and the evolution of knowledge to a speculative long-term view of intelligence and the universe. This is evidence of his ambition and vocabulary, not a scientific finding or a practical company plan.²²

“A New City” approached the problem through the organization of environments. Its urban analogy made the relations between activities and places important, rather than treating a digital tool as a bag of isolated features. “A Forgotten History,” published in 2021, situated his interests in the history of interactive computing and the ambition to augment human intellectual activity. Together these essays show him trying to move from a personal educational problem to the design of environments in which other people might think.^{23, 24}

Douglas Engelbart, Alan Kay, Ted Nelson, and Seymour Papert supplied a vocabulary different from the conventional pitch for a productivity application. Their presence does not establish a direct line of inheritance or equivalence between their research programs and Heptabase. It does explain why an application built around cards and whiteboards could be described by its founder as an early part of something larger. The product would have to be small enough to make and specific enough to use, while his reasons for wanting it remained expansive.

The November 2021 Heptabase vision essay organized knowledge work around activities including exploration, collection, thinking, creation, and sharing. Its value as a source is partly the gap between these ambitions and what was then buildable. A founder’s vision can recruit collaborators and help decide what to exclude, but it can also remain underspecified until actual users reveal the tasks they will repeatedly perform.²⁵

Founding Heptabase and learning the chief executive’s work

Heptabase emerged through failed prototypes and changing collaborators. Chan eventually worked with Wu on design and users while Lin assumed principal development responsibilities.²⁶

By early 2022, the division of labor was explicit in an interview: Chan handled product direction, fundraising, and customer support; Wu worked on detailed design and simulations of use; Lin led development and technical management. Analytics were not reserved to one founder. This was a transition in Chan’s vocation from personally learning and building many things to sustaining an organization whose parts depended on different people’s judgment.²⁷

Heptabase entered Y Combinator’s Winter 2022 batch. Chan’s financing account reports $1.68 million raised across several kinds of investors by March 2022. The amount is best read as a historical disclosure, not a current balance-sheet figure. Venture capital supplied an option to expand; the company also pursued paid use early rather than treating monetization as a distant stage.²⁸

In a December 2022 retrospective on his departures from university, Chan divided the previous years into exploration and execution and described reaching a point at which the company could support itself. That is his own retrospective periodization. It does not mean the preceding years lacked productive work, or that profitability resolved his earlier questions about what kind of life he wanted.²⁹

His responsibilities now included domains far from the technical material he had first studied. Learning had to serve actual decisions about product priorities, cash, contracts, and customer expectations. The knowledge-work application was simultaneously a product he sold and an environment in which he tried to acquire the knowledge needed to keep the company operating.

Reading, whiteboards, and the public record of work

Chan’s account of learning with Heptabase describes a large personal collection of cards and whiteboards, but its more informative details concern their use. He brought user interviews into a system where recurring problems could be compared; read about retention and product analytics; and studied legal, financial, and computing-history material as needs arose. Different inquiries produced different structures rather than a single universal taxonomy.³⁰

His 2023 account of acquiring knowledge separates collection from understanding and connection. Highlighting a passage is an intermediate act. Restating an idea and deciding how it relates to another idea require additional work. The whiteboard provides a place to externalize that work, but does not perform it merely by existing. This is a description of a method, not controlled evidence that every learner benefits equally from it.³¹

The public library on his website broadens the picture. Scientific and philosophical interests coexist with fiction, cinema, manga, and sport. A list of things someone has consumed cannot establish influence by itself, and it would be careless to assign a product feature to a favorite book without a more specific source. It does, however, prevent the reduction of his cultural life to the handful of computing pioneers most useful to a company narrative.³²

Writing is another part of the working apparatus. Explaining a decision publicly requires selecting examples, articulating differences, and exposing claims to readers who may disagree. The record also preserves revisions. Readers can compare an early expectation with later accounts of what actually happened, rather than receiving only the successful founder’s final retrospective.

London, 2024, and a larger team

In March 2024 Chan moved to London with some colleagues after obtaining a UK Global Talent visa. His account describes preparing evidence and references while continuing to work, and argues that a remote company can relocate some people without concentrating everyone in one city. The move belongs to both a personal biography and a company’s administrative history; it does not mean that Heptabase ceased to have a Taiwanese team.³³

His subsequent comparison of Taipei and London is unusually concrete. It weighs food, rent, transport, parks, sport, museums, shopping, banking, and healthcare. He likes aspects of London’s urban and cultural environment while missing Taipei’s convenience and food. These are personal judgments, not a general ranking of two societies. They show that a relocation decision cannot be reconstructed adequately from tax rates, investor access, or a startup directory’s headquarters field alone.³⁴

Managing a larger organization raised different questions. In a May 2025 essay, Chan said the team had remained at six people until June 2024 and described later engineering recruitment. He stressed communication, capacity to learn, and experience in small companies, while acknowledging that a substantial portion of the previous year’s feature work had not been the best use of effort. These are an employer’s criteria and self-assessment, not independent evidence of employee experience.³⁵

The admission is useful because it breaks a simple equivalence between more resources and better outcomes. An additional engineer can increase what gets built without settling whether it should be built. For Chan, the practical problem had shifted toward the allocation and coordination of other people’s work, including the willingness to recognize his own prioritization errors.

AI and a change in learning practice, 2025–2026

In September 2025 Chan published a detailed account of studying with AI, using Christopher Bishop’s Pattern Recognition and Machine Learning as the example. His emphasis was on making difficult primary material approachable through a sequence of explanations, questions, and selected context. He distinguished spending a fixed amount of time on more superficial material from using assistance to attempt a harder text.³⁶

The method joined source selection, PDF parsing, conversations with a model, and the spatial arrangement of learning material. The demonstration was also a demonstration of the product he sold. That dual role deserves to remain visible: his enthusiasm may be sincere and his examples useful, while neither establishes general learning gains or removes the need to check an AI explanation against the source.

By 2026 the company’s emphasis on AI had generated disagreements among users about cost, interface changes, and development priorities. In a community letter, Chan acknowledged these concerns and explained the turn toward AI-centered learning. He also announced that he and Wu would handle in-app support daily. The letter documents commitments and a public response to criticism; it is not independent proof that the commitments were subsequently fulfilled.³⁷

This is a different kind of encounter with education from the university essays. He was now responsible for a commercial environment used by people whose needs and preferences did not necessarily match his own evolving method. The user who wanted a stable visual thinking tool and the founder excited by a new AI-mediated practice could have legitimate but competing expectations of the same subscription.

Work, friendship, and what he no longer wants to optimize

Chan’s January 2026 essay “錘子” questions the extension of business and engineering habits into every area of life. He explains his reluctance to return to San Francisco in terms of an environment where exchange, efficiency, and optimization could crowd out other ways of relating. He also describes wanting to be someone who can put work aside for people he cares about.³⁸

This is not evidence that he abandoned technology, business, or disciplined work. He continued leading Heptabase. The change concerns the jurisdiction of those methods: success at organizing a company does not confer the right method for friendship, beauty, or every decision about how to spend time. His public “Mindset” page similarly resists defining a person solely through a profession and gives relationships their own demands of attention and commitment.³⁹

A short April 2026 travel essay records walks and views in Edinburgh, encounters with the city’s historical associations, and a train journey to Manchester. Its presence in the archive matters partly because it does not need to be converted into a product insight. It also has a narrower geographical meaning than a headquarters announcement: a train destination does not establish a residence or a corporate relocation.⁴⁰

In October 2026 Chan wrote about his father’s death and their difficult history around his educational choices. He recalled a later conversation in which his father asked whether he was happy in his work, and described the reconciliation he had imagined but could no longer arrange. The account is his perspective on a family relationship, not a basis for assigning motives to the other person. It leaves grief and unfinished conversation in the record rather than resolving them through the success of the company.⁴¹

As of November 2026, the public record therefore contains both a rapidly changing software business and a founder questioning which parts of his life should be organized by its logic. The future of Heptabase remains open. So does the relation between the work he has chosen and the other commitments he has learned to protect.