A few friends I know around the age of twenty sleep on buses. They feel that sleeping in a bed is a waste of time. They spend almost all their remaining hours absorbed in writing code or making things. Asking these people what college means to them is a bit like telling someone already swimming to sign up for a pool first.
Lately, I have been spending time with some talented graduates of high schools specializing in software development. Some are Fellows at Hashed Vibe Labs. Watching how fast they grow makes me seriously question whether college is still the best default. This does not mean the value of college has disappeared. It means the functions it long held a monopoly on are being unbundled, one by one. Information, relationships, selection. These three used to come in a single package. That is no longer the case.
The first is information. College was once a water filter for knowledge. There was a clear information gap around what to learn and in what order. But now the river itself has become clear. MIT OpenCourseWare, GitHub, YouTube lectures, open-source documentation, global communities. An abundance of information does not guarantee good learning. Still, the days when college alone controlled the entrance to learning are over.
The second is relationships. The school name on a business card used to do the handshaking for you. Now other kinds of handshakes are emerging. Commit logs shake hands, open-source contributions serve as introductions, and hackathon projects speak before business cards do. A school name is still a strong signal, but it is not the only one. It is no longer unusual to see code posted on GitHub by a developer from a specialized high school start a conversation with a senior developer in Silicon Valley. Entrepreneur Jungseok Roh calls these young developers “sages” as a mark of respect, for much the same reason. These are moments when ability, already alive and at work, speaks before an old institution gives its approval.
The third is selection. This was the core function that made college what it was. Gather people who have passed similar exams in one place for four years, and a network forms, shaping their paths for the next ten or twenty years. But this selection mechanism now has competitors too. GitHub stars, hackathon results, open-source contributions, community reputation, actual user numbers and revenue. These sometimes offer a higher-resolution signal than a 4.5 GPA. People who have actually built, shipped, and fixed things are starting to come into sharper focus than people who are good at taking tests.
Michael Spence’s signaling theory is worth revisiting here. The idea is that education does not directly raise productivity; rather, people who are already highly productive use degrees to prove it. If this model is right, a degree is not evidence of learning but a signal of the ability to learn. What happens to the price of a degree when faster, cheaper, harder-to-fake signals emerge? If a degree was gold under the gold standard, a GitHub profile is closer to a new signal that is faster and more transparent. The moment a better signal appears, the premium on the old one begins to erode.
Peter Thiel was raising the same question when he called college a bubble. Can a state in which price has outstripped value last forever? IBM, Apple, and Google did not drop degree requirements for some roles because they rejected the value of education. They did so because they could now measure more directly what they had been measuring through degrees. Total student loan debt in the United States has exceeded a trillion dollars, and in Korea, too, college is becoming an increasingly expensive option when tuition and opportunity costs are considered together. As costs rise, people ask more insistently. What exactly is this price buying?
The more fundamental problem with college lies in how it structures time. It binds talented people, at the age when they can grow fastest, to curricula built around decades-old paradigms and the rhythm of midterms and finals. Isn’t the time almost every college student spends truly absorbed in studying concentrated around exams? It is less a system that teaches people to love knowledge than one that gets them through assessments. The romance and friendships of youth, and time for unhurried exploration, are not without value. But it seems increasingly clear that for some people hungry to grow, college is not the best option.
Another variable has entered the picture. Vibe coding. A senior developer’s experience and feel for architecture still matter. But what makes them matter is changing. In the past, a senior developer’s value was closer to “How much have you implemented yourself?” Now it comes down to “Can you judge what to build and what not to build?” By contrast, the strengths of juniors without legacy baggage show up elsewhere. Stamina, focus, an ability to adapt without preconceptions, and less loyalty to established ways of doing things. The strength of that friend who sleeps only on buses is not that they know more. It is that they are not doing anything else.
An empty hard drive writes faster than a fragmented one. The strength of being twenty may lie less in genius than in simplicity. Someone free of complexity can push certain kinds of work forward much faster than someone full of it. The slowness of forty is often not a lack of ability, but the weight of what has already been written. Paul Graham said the advantage of youth is having nothing to lose. I would add one more thing. Having no legacy to maintain yet.
Perhaps that explains a strangely funny scene. Those friends from software-focused high schools envy elementary school children younger than themselves. They feel they have already spent some time learning the rules of an earlier era, while those children will enter fully into the age of AI with a cleaner slate. The sight of twenty-year-olds envying ten-year-olds is both funny and apt. It means that generational divides are beginning to depend not on age, but on how long someone lived before a particular tool appeared.
Of course, this does not apply to every young person. Only in a very small number of cases does a junior’s focus translate directly into results. Does college create that one percent, or does it gather early the people who would have been in the one percent wherever they went? If it is the latter, what college selects is not talent but probability. It is also true that jobs are hard to find, internships are scarce, and many juniors are struggling. At the same time, a small few are encountering far greater opportunities than the generation before them. Competition has grown fiercer, and the field has grown wider. There are more doors. There is just no map yet showing which door leads where.
That is why college sometimes feels like a swimming certificate these days. Proof that you can enter the water. But what we need now is not a certificate. It is someone already cutting through the water. If the ocean changes twice in the four years it takes to earn that certificate, we need to reconsider the relationship between proof of qualification and actual speed. Some knowledge comes from failure before it comes from a classroom, some networks form faster in open-source communities than in alumni associations, and some careers begin well before graduation.
I often think about how to support friends with this kind of talent more systematically. I do not yet have a clear answer, but one thing is certain. Those who sleep only on buses are already on the highway. The moment we park that bus at a stop called a diploma, it could take four years to merge back into traffic. The highway will not wait. If the system does not become more flexible for them, they will forge a new world outside it.
What do these people need most right now? A good curriculum, strong mentorship, a peer group, or their first paying customer? Perhaps the best school of the future is not a campus, but an ecosystem that lets talent already in motion keep moving.
