2026-07-26 · 9 min read · Raw Drafts / Learning / NCKU
I Tried to Read an Entire University
I knew NCKU as buildings, paths, studios, and cafeterias. Mapping the books behind every department revealed more than a hundred incompatible ways of deciding what reality is.
I asked ChatGPT to list the books students from every department at National Cheng Kung University are supposed to read. This was obviously an unreasonable request. NCKU has more than a hundred academic units, depending on whether graduate institutes, medical specialties, interdisciplinary programs, research centers, and the newer semiconductor academy are counted separately. There is no shelf somewhere in the administration building containing the official intellectual contents of the university. Professors choose different books. Courses change. Some classes use journal articles, lecture slides, standards, case studies, or a pile of PDFs whose filenames make sense only to the person who uploaded them.
But I wanted the list anyway.
I spent several years at NCKU while knowing almost nothing about what most of the university was doing. I knew the Planning and Design building, the studios, some professors, the convenience stores, the cafeterias, the places where I could charge my laptop, and the route I took across campus. Nearby, other students were apparently learning orbital mechanics, parasitology, semiconductor packaging, constitutional law, tropical plant science, psychiatry, naval architecture, archaeology, gerontology, and dozens of other things I could barely name.
We were members of the same university, but not really the same intellectual world.
A university is easier to see as buildings
The physical university makes the separation look normal. Chemistry happens in one building, architecture in another, medicine across the road, engineering in a cluster of taller blocks. Departments have their own entrances, noticeboards, laboratories, vocabulary, internal politics, and ways of dressing. Even the elevators can feel disciplinary. An architecture building has models leaning against walls. A medical building has people moving with a slightly more urgent purpose. An engineering corridor contains equipment whose function is obvious to everyone except the visitor.
The administrative map says these are parts of one institution. The books suggest something stranger. Each department is teaching a different answer to the question of what deserves attention.
A mechanical engineer learns to see loads, tolerances, stress, heat, friction, failure modes, and safety factors. A historian learns that evidence was produced by somebody, preserved by somebody else, and selected by the historian asking the question. A doctor learns to translate a patient's account into anatomy, mechanisms, probabilities, tests, and treatment, while trying not to erase the person in the process. An accountant learns to convert messy economic activity into standardized statements, and then learns where judgment is hiding inside the standardization. A designer learns that what appears to be human error may be the result of a bad mapping, weak feedback, or an interface that expected the wrong behaviour.
These are not only bodies of knowledge. They are perception-training machines.
After enough time in a field, certain things begin to glow. The urban planner notices the width of the pavement, the distance between crossings, the mixture of uses, and the blank wall that makes a street feel longer than it is. The materials scientist looks at a cracked object and sees processing history, microstructure, defects, and stress. The political scientist watches a public decision and asks which coalition has to remain satisfied for the decision-maker to survive. The interaction designer notices the button that is visually available but behaviourally impossible.
The degree changes what becomes visible.
The university contains incompatible realities
NCKU presents itself as one university. The reading atlas contains at least 134 different operating systems.
For physics, reality can be expressed through fields, particles, energy, symmetry, and conservation laws. For literature, reality is also produced through language, narrative, translation, memory, and power. For medicine, the body is a living system that must be interpreted under uncertainty. For management, organizations become processes, incentives, constraints, information, and value. For architecture, space is not emptiness between walls but a sequence of thresholds, proportions, movement, light, climate, and social expectation.
The departments sometimes study the same object without agreeing on what the object is. A city can be a transportation network, a property market, a legal territory, an ecological metabolism, a political coalition, a collection of buildings, a field of memories, or an interface that makes some actions easy and others nearly impossible. A body can be anatomy, chemistry, movement, behaviour, identity, labour, risk, or lived experience. A semiconductor can be matter, a device, a manufacturing process, a supply chain, a national-security problem, and a business model.
The object does not change when you cross the campus. The questions change, and then the object appears to change with them.
This is probably obvious to anyone who has seriously crossed disciplines, but the university structure usually hides it. A student experiences a narrow corridor through the institution. They learn their own prerequisites, methods, canonical arguments, and professional anxieties. The rest of the campus remains physically nearby and intellectually distant. An electrical engineering student and a Taiwanese literature student may eat at the same table while being trained to recognize almost non-overlapping forms of evidence.
The atlas was a way of reading the university sideways instead of completing one curriculum vertically.
The books are not neutral
The list also became less authoritative the closer I looked at it. There is no definitive book for a department. A field contains arguments about what is foundational, who should be canonical, what is outdated, and which methods have been allowed to define competence. Many supposedly universal textbooks were written in the United States or the United Kingdom and then imported into a Taiwanese institution whose local history and professional conditions are not identical. A title can be excellent and still represent a particular geography of knowledge.
A reading list is a quiet exercise of power. It decides which concepts arrive first and therefore feel fundamental. It decides which thinkers become names every student recognizes and which appear later as alternatives, critiques, regional cases, or optional material. It decides whether a field introduces itself through theory, practice, history, calculation, or professional procedure.
Even the status labels in the atlas needed to remain awkward. Some books could be traced to publicly indexed NCKU syllabi, but that only proves that one instructor used one edition in one course at one point in time. Other books are field standards, selected because they are widely used or because they make the discipline legible to an outsider. Literature and philosophy also required primary works, where treating the book as a textbook would misunderstand why it matters.
The atlas is therefore not a recovered official curriculum. It is a reconstruction, and the seams should remain visible.
AI makes intellectual trespassing cheap
Without AI, this would have been the sort of project I imagined and then abandoned after opening twenty browser tabs. The university is too large, the course information is inconsistent, and every unfamiliar discipline creates another branch of research. AI made it possible to build a first map quickly, check the structure, find obvious gaps, and compress each book into a few ideas that explain why it might belong on the shelf.
This does not mean I have read 212 books. It means the cost of approaching 212 books has changed.
That distinction matters. A summary cannot reproduce the experience of reading a difficult text, the accumulation of examples, the arguments with earlier authors, or the point at which a concept finally rearranges something in your head. But the summary can tell me where a door might be. It can show that a book exists, why a department values it, and which question I might carry into the first chapter.
AI makes horizontal movement across knowledge much cheaper than it used to be. It does not turn the outsider into a surgeon, structural engineer, historian, or biochemist. It makes it easier to identify which discipline might already have language for a problem I have encountered without knowing its name.
This feels useful because most of the problems I now work on refuse to remain inside one department. A public Wi-Fi map is software, urban infrastructure, data governance, accessibility, maintenance, and institutional memory. Green Filter began as interaction design but kept expanding into finance, behavioural science, sustainability, data systems, and agentic commerce. A medical interface cannot be understood through interface design alone because the surrounding workflow, regulation, incentives, evidence, and risk are part of the product.
Traditional education separated these areas so they could be studied deeply. Actual work keeps collapsing them back together.
Building the education I did not get
There is also a more personal and slightly embarrassing reason for making the atlas. I am often trying to assemble the education I feel I should already have.
I studied audiovisual media and later interaction design. Both gave me ways of seeing that I still use, but my curiosity keeps escaping their borders. I want to understand physics without becoming a physicist, biology without pretending to be a biologist, finance without reducing everything to money, and engineering without treating equations as decorative symbols. I want enough structure to know when I am asking the wrong field to solve a problem.
Another degree might provide depth, deadlines, teachers, and a cohort, but it would still choose one corridor. The atlas suggests a different kind of post-university curriculum: identify the foundational shelf of each discipline, understand the questions that organize it, and follow the connections that repeatedly appear between fields.
This is not the same as mastering the university. It may be a way of becoming less trapped by the department in which I happened to begin.
The first version contains 134 academic units and 212 books. That is already too many, and also clearly not enough. The medical college alone can keep subdividing until every organ, technique, and patient population has its own literature. Engineering books branch into standards, handbooks, and specialties. The humanities resist the idea that two books could represent a department at all. Every shelf could be challenged, expanded, localized, or replaced.
That incompleteness may be the point. A university is not a finished body of knowledge. It is an argument about how knowledge should be divided, preserved, taught, and made useful. The atlas is another argument layered on top of it.
I used to think I knew NCKU because I knew how to move through it. I knew which gate to enter, where to park, which building was too cold, which staircase led to the studio, and how long it took to walk across campus in the Tainan heat. Now I am less certain. Knowing a university spatially is not the same as knowing the different realities being constructed inside it.
I may need to return through the books.
Open the NCKU Reading Atlas and enter through any department.
AI authorship note
This essay grew from Kris Haamer asking for the foundational books associated with every academic unit at National Cheng Kung University, then asking what Raw Drafts could do with the result. AI helped reconstruct the university's academic structure, organize the reading atlas, summarize the books, design the interactive experience, and draft and implement this article. Kris supplied the personal history, the central framing of departments as different ways of seeing, and the editorial direction.