This is my second post responding to Paul Graham’s essay on university entrepreneurship centers. In the first I focused on what he gets right, while defending what I see as the best of what we do at the Lassonde Entrepreneur Institute. Then I read the responses of people I respect — particularly Bill Aulet, managing director of the Martin Trust Center for MIT Entrepreneurship and author of “Disciplined Entrepreneurship” — and realized I conceded too much.
Graham argues that the hard part of startups is product: knowing what to build and being able to build it. I agreed and said we are downstream of the departments that teach it. That is true of building the thing. It is not true of building the company. Aulet, writing from a longer vantage, put it plainly: “The hardest part of startups is this carbon-based life form called humans.” He argues product is being commoditized, that product-market fit is becoming table stakes, and that what separates outcomes is team, go-to-market strategy, and execution — none of which is taught in the departments Graham names.
Graham folded those two halves into one. I accepted the fold. The half I gave away is the half this essay is about.
One constraint, or two
Graham names a single binding constraint on a prospective founder: time to build. He is right that unstructured time is a precondition and not a luxury. But he named one constraint and then treated it as the only one. There are others, and they don’t respond to inspiration.
The students I work with — including the most capable and driven of them — struggle to make the finances work. How to build, and also eat, and sleep, and pay tuition.
And one constraint shows up on neither a calendar nor a personal financial statement. Belief that you are the kind of person who could build something that matters.
Belief is a prior, not a switch
Graham hints at the mechanism, and he is close. Seeing a founder up close converts founding from a thing other people do into a thing a person like you does. He even gets the dosage right — the mid-twenties founder three years in does more work than the billionaire.
The limit is who it works on best.
I was talking with my high school son about a program we were about to run for students his age. His advice: make sure you have college kids doing most of the talking. He was making a serious point. Proximity is not only physical. It is also position in a life. The nearer a founder stands on both axes, the more of the work the exposure does, because the student has less distance to close on their own. Proximity without identification is merely entertainment.
Graham points to the gap between how often Harvard students apply to Y Combinator and how often students elsewhere do. But an application is what belief looks like after it has already formed. His instrument cannot see anyone on the far side of that threshold, which means it cannot tell him what moves them. That is not a flaw in the observation, but it is a flaw in generalizing it.
Belief is a prior condition. It updates on evidence about yourself, not on evidence about somebody else.
What works, and what gets built instead
What works best is small, participatory, cohort-based, consequence-bearing. Peers learning in front of each other, with something real on the line.
What often gets built instead is large, passive, and photographable. Speaker series, panels, firesides. Evidence about somebody else, delivered to a student in a chair, nodding politely.
The pull is for institutions to default to the legible version. That is the accreditation problem I raised last time, showing up again closer to the ground.
Here is the design error underneath it. An accelerator selects on belief and then supplies resources. The students we are talking about need the inverse. Most intro programming is an accelerator with the intensity turned down a smidget — same machine, weaker current. What these students need is a different machine, and a low-intensity one on purpose, because the cost of stepping onto the first rung has to be something a working student can afford.
How belief moves
So how does belief move? Not by argument, and not by exposure on its own. It moves when a student does one small, real thing with a consequence attached — something they did not believe they were capable of until it was done. The scale matters less than people assume. What matters is that it was real, that it could have gone badly, and that it was theirs.
Who is in the room matters just as much. Peers taking the same risk on the same day, so the act reads as ordinary rather than exceptional. And community masters bearing witness — someone further along, whose judgment the student respects, watching it happen and treating it as unremarkable. A student can talk themselves out of their own success in about a day. It is harder to do that when somebody who knows the work saw it and wasn’t surprised.
That is formation, and it happens well upstream of where Graham is standing. By the time a student is filling out a Y Combinator application, this work is years finished.
Which is also why we err toward volume. David Agogo, a colleague of mine at the Eccles School, puts it this way: “The path to entrepreneurial success is notoriously circuitous.” Not every student arrives on campus already sparked, and the ones who don’t need an environment stable enough to stretch them until they find out. So we build a lot of doors — roughly seven thousand individuals experience a Lassonde program in a year. A door is not a lecture. It is the smallest version of the thing described above: participatory, finite, with a real outcome at the end of it, and cheap enough in hours that a student working nights can walk through one. That figure is not a boast. Plenty of those encounters do nothing at all. We run them because we cannot tell in advance which ones won’t.
Survival
Graham’s population had two conditions met, not one. They believed they could create, and they knew they could eat while they tried.
A student carrying 20 credit hours and 20 hours of paid work has no hours to hand a project. Reading period — the open weeks Gates and Zuckerberg had at Harvard — has no equivalent on most campuses. The nearest substitute is independent study outside the core curriculum, and a student has to know it exists, find a sponsor, and pay for the credit hour.
The two constraints are not independent. A student watching from the outside learns that founding is for people with a cushion, and they are not wrong. Survival shapes belief, belief spends hours, and the hours are the thing they don’t have.
Unstructured time is a purchased good
So my job is to buy them space, and the cheapest purchase available is the one universities resist hardest. We do this through programs like the Master of Business Creation, Builder’s Track in the entrepreneurship major and minor, and the Sandbox accelerator. Credit for work that reaches a learning outcome faster and deeper than any of us can reach it by talking. Without it, building is a third job stacked on top of the coursework and the shift. With it, the building is the coursework. Nothing else about the student’s week has to change — the hours simply stop competing. Which is the whole of what anyone means by buying a student time.
The second purchase is scholarship. Tied to building and creating, and to leading the programs that make building possible, so the money converts tuition pressure into hours rather than into a line on a résumé.
The third is the building, and here I will say the uncomfortable part before somebody says it for me. Our slogan at Lassonde Studios is “Live. Create. Launch.” The first word buys the room for the other two — room to survive, and room to believe. But Lassonde Studios is paid for by the students who live in it. The people who benefit are the people who pay, which is why the building exists at all and why it does not depend on anyone’s continued generosity to keep operating. It also runs directly against the argument I have been making all essay. The students least able to afford a bed are often the students for whom proximity would do the most work. The same financial logic that built the building narrows who gets to live inside it. Scholarship relieves that pressure, and scholarship is not free either. Somebody funds it, every year, or the pressure comes back.
Which is why the ground floor is open to all, without a university key card. Most of the people in it do not live upstairs — commuters, students working thirty hours a week, alumni, people with no affiliation at all. A student who could not afford a bed still gets the room, the tools, and the person at the next bench. That does not solve the problem I just described. It keeps the problem from being the whole story.
Graham’s model has two states. A student is building, or they are not. That works when both questions have already been answered — when belief is settled and survival is handled, the only variable left is whether the calendar is clear. Our students arrive at every point along both axes. So the design has to be a ladder instead of a switch. Entry points that cost an hour and entry points that cost a summer, arranged so a student with one hour has somewhere to spend it, and can see the next rung from where they are standing. Some of them never climb past the second one. That is fine. The ladder is not a funnel, and I am not measuring it like one.
The asymmetry
Inspire a student to build when they can’t afford the hours, and you have handed them a frustration. Buy hours for a student who doesn’t believe building is for them, and they will spend those hours elsewhere.
Both, or neither.
Graham says his approach costs nothing extra. It costs nothing extra where somebody already paid. Belief and space cost someone. Find the universities willing to be that someone. I’m glad I did.
