Colleges have an education problem

Reflections on a broken system

When a measure becomes a target, it ceases to be a good measure — Goodhart's Law

---

My cofounder mentioned something interesting over lunch — that he wasn't sure if he would want to send his two kids to college. "I just hope they don't, like, have to go through all that. It's such bullshit, and the value just isn't there." He's right. I've been thinking the same. We're not the only ones.

I've mentioned a few times on this newsletter that I didn't love my time at uni. And since I've been thinking a lot more about credentials and epistemology, I want to lay out in full why I think our universities, and indeed our entire higher education system, are struggling. The problem is not woke, political bias, or funding; nor is it that too few people can attend, or that the price is too high (though the latter is symptomatic).

The problem is that we have jammed our society's education and research institutions together with its credentialing systems. And the former is crumbling under the weight of the latter.

A tale of two universities

When freshmen come through orientation during their first days on campus, they are bombarded with propaganda about the school and themselves. "You have been chosen among thousands and thousands of applicants", says the dean at the mandated one hour speech. "You have come to this place of learning to be the future leaders of the world. Here you will learn among the best and brightest" — that phrase appears a lot by the way, 'best and brightest', like we're all light bulbs at a Home Depot — "to take on the hardest challenges, whether it's curing cancer, writing the next great American novel, or running for political office. Through our classes, you will find mentors in your professors and push yourself to grapple with complicated theorems or philosophical questions" etc etc.

This is, in a nutshell, the first university, and also the public facing one. In this model of the world, the university is a place of learning. You go to college to learn things, things that will prepare you for future hardship and challenges. The point is to be in an environment where you can try a lot, and fail a lot, all in service of a higher education. And as a result, when you graduate you will come out more equipped to handle some of society's problems.

Every year, some poor kids buy the propaganda. They take the hardest classes and really push themselves, and even though they work like dogs they get Cs and Ds. They have some wicked hard classes on their resume, and there is a sense in which they have learned quite a bit, and a deeper sense in which they are now more prepared to actually make a positive impact in the world than they were previously. But they have a 2.5 GPA. And everyone else graduating from the school? Those kids who took the easy A class, they all have a 3.5 or 4.0 GPA. So no one hires the kids who worked really hard.

What happened?

Those hard workers all ran head first into a second, completely different university, one that hides behind the "higher education" facade. This second university isn't about education at all. In fact, if you're earnestly there for education, you're going to get screwed. No, this university is all about credentialing — that is, proving to other people that you are intelligent and capable so that you can land a sweet high paying job. In university-two, the only goal is to come out of the school with as high a GPA and as many accolades as possible, learning be damned. Oh, sure, you may learn a bit as you go. But it's entirely secondary, and often directly contradictory to the main goal. You try as hard as you can to take easy classes, you make up excuses to get extra time on exams and assignments, you try to become friends with smarter peers or TAs. Going out of your way to take harder classes? You're a chump.

If you play the game right, you are rewarded — massively. Get a good GPA from a good school and you'll have an easier time walking into any consulting or finance or tech gig and you'll be on your way to making millions. Sure, you'll have to pass an interview screen there too, but that's easy enough. The hard part is getting your resume through the reject pile. The right university name with the right GPA to get through the filters, though, and you're most of the way through to the good life. So you got to play the game. And the rest of those kids who are try-hards are just helping the rest of us with the curve.

---

Even though I'm being harsh here, I don't want the take away to be that we shouldn't have either higher education or credentialing institutions. Both of these are critical for a society to function. We need a way for people to dive deep into various subjects and really learn their stuff; and we also need a way to identify who can perform specific jobs well. The problem is that through the entirely organic machinations of the free market, we have stuffed both of these into a single place. These are, in a meaningful sense, contradictory goals; any single institution trying to meet both masters will inevitably fail.

Right now, universities are failing by increasingly abandoning their educational goals in favor of focusing on credentialing. As a result, our students pay thousands of dollars for McKinsey and Goldman to outsource their hiring pipelines. And the university can sit in the middle and rent seek, because their degree has essentially become the only way to even get in front of these businesses to begin with.

Note that everyone here is more or less following their incentives. Assume that Harvard starts as a place of higher education — the people who go there are there to legitimately learn and are willing to sacrifice years of their life to do it (and very few people actually go there as a result). McKinsey wants to hire the best people but put in the least work to find them, so it goes to Harvard and recruits from there. The average person who isn't interested in education but IS interested in money realizes that going to Harvard is a great way to get to McKinsey, so more people start applying to Harvard. The university realizes this and jacks up its tuition price. It may loosen its standards to capture more cash, or it may try to be more selective to retain its prestige, but it won't matter either way; it is easier for unqualified people to get into the university than it is for the university to filter people out.

We're now in a death spiral until, inevitably, the college degree loses its value as a quality signal. This process is already occurring. Approximately 35% of US Citizens 25+ have a Bachelor's degree! Once that rot hits even the best institutions, McKinsey will move on to other pastures. McKinsey will be fine, of course — they and everyone else more or less train new-hires in their first year on the job. But our educational institutions won't be. They will have been hollowed out and left for scraps. Goodhart's law in action.

Instruments of a Recruiting Pipeline

It's an open secret that university is not really about learning. Students know it, and they talk about it all the time. Some of the faculty know it; the more subversive ones will purposely give everyone As or otherwise screw with the grading scale as an act of protest. Parents know it, that's why so many of them are still willing to put their kids into debt or spend hundreds of thousands of dollars for a degree.

And yet, it's really hard to call this problem out for what it is. Why is that?

In my opinion, the most frustrating thing about all this is that the university administration — the deans, and presidents, and definitely the endowment people — will do whatever they can to uphold the charade of higher education. They can't stop talking about education. It comes up in public speeches and printed materials and alumni marketing and donation emails… And if you try to call them out on it, if you argue that the schools are becoming more and more like pay-for-play financial institutions than organizations with higher principles, they will double down on the whole education thing.

But if you know where to look, the mask slips.

The most obvious tell is curved grading. If the goal of a test is to quantify how much a student has learned for their own edification, why on earth would you stack rank the class? That's not going to help anyone on their educational journey. It certainly would not encourage people to push themselves to take hard classes. In fact, the only real benefit of curved grading is that you get a method to identify which students do better compared to their peers. That is simply useless data when it comes to learning. But it is extremely valuable if you care about credentialing.

Another example is the concept of 'academic dishonesty', colloquially known as cheating. Students who are caught cheating are punished severely, often with more draconian punishments than those who are physically violent. If the goal of the university was education, the folks who are cheating would really just be preventing themselves from learning; the only harm is to the individual engaging in 'dishonesty'. But if you care a lot about credentialing, then rampant cheating devalues the degree. The harm is spread out across everyone graduating from the institution who may now face more scrutiny from recruiting and hiring pipelines. And at the same time, if the primary goal of university was getting a credential regardless of the actual learning that is happening, there is an extremely high incentive to cheat.

Did you know that somewhere between 60-70% of students admit to some kind of academic dishonesty?

School administrators like to frame this as a moral issue, which is either incredibly naive or purposely cynical. One or two people cheating on a test is a personal failing. 3/5ths of your student population cheating is a structural incentive. Those numbers should not be surprising to anyone who is paying attention. Of course the students will cheat! They are paying hundreds of thousands of dollars to get a degree with the sole purpose of leveraging that degree for a high status high paying job! Schools can be as draconian as they want, they will never fix this problem. Rather, what will happen is a few kids will get their lives ruined because they got caught doing what everyone else is also already doing.

A good friend of mine was almost one of those kids. He got caught up in an academic dishonesty case while we were in college. If I remember the details right, he was working with a few other folks on some comp sci project that was meant to be individual,¹ and the teacher caught on. Sent the whole group through the Dean's academic dishonesty process, where they were each individually interrogated by an admin — not a faculty member, just another faceless bureaucrat. Apparently, he had asked the admin exactly this question: "why does cheating even matter if the only people being harmed are the people cheating?" And the admin said the quiet part out loud. "Academic dishonesty is existential because it will devalue the degree. People will not want to hire students from our university if the degree is not valuable." Mask-off moment indeed. That story always sat with me because it was such a bitter lesson.

While we're on with the anecdotes, here's another. Every semester, a single required CS class would send approximately 200 kids through the academic dishonesty process. That's not a typo; it was about half the class every semester. The professor of this class took cheating really seriously, and made it a personal mission to root out as much of it as possible. And so, every single semester he would send out the same set of emails, about how he was so disappointed that so many people cheated, offering amnesty for folks who turned themselves in. One semester, in his traditional tsk-tsk email, he included a story of a student who was struggling in the class and had ended the year with a D. The professor proudly exalted this student who, despite struggling, refused to engage in academic dishonesty. "What a paragon of honesty!" said the professor. " Truly an inspiration!" All I could think was that kid still ended the class with a D. The extra honesty really hurt his job search. Chump.

GPAs, grading curves, academic dishonesty tribunals, job fairs, on campus interviews. The modern university is a four year talent and recruitment pipeline for major corporations. People wonder why the price tag of the degree didn't change at all during COVID, even though the actual learning opportunities were obviously significantly less. But they are basically answering their own question: the price tag didn't change as a function of learning because the price tag was never about learning to begin with.

People like to make fun of English majors, but honestly I think they are the real victims of this whole farce. The average English literature student is rather obviously not in it for trying to game the system. There's no system to game! Having an English degree doesn't obviously help you get a job. And as a result, the English department is the only one that is actually treating the educational system the way it was meant to be. No one cheats through a class on Ulysses, and the teachers mostly give out As anyway. Everyone is there to learn. Maybe not surprisingly, I adored my English literature classes. I thought they were some of the most useful parts of my 'higher education', because it was only there that I actually learned anything interesting or new. The problem is, those English majors are all paying the sticker price that the CS and econ grads are paying. Those latter two groups can probably trade their degree up for a job at a cushy tech job or a management consulting firm. But the English majors just end up with debt and an appreciation for Chaucer.

Silicon Valley Isn't It

Look, I know my audience. I know a lot of us are startup founders and libertarians and tech people. A certain type of person is probably banging the table right now. "Yes!" they might say. "Yes! This guy gets it! This guyyyy! He gets it! College is a waste of time! People shouldn't spend their most creative years locked up in a cage! They should be building things! They should be starting companies! They don't need the conformity of the classroom, they can go out and make their way in Silicon Valley!"

An AI’s depiction of silicon valley. Something about how goofy the text is really just fits.

I'm being a bit glib, but only because I related to the above sentiment a lot in a past life. Leaving college, I felt like most of my Computer Engineering degree coursework was useless. Almost everything that was important, I learned in my free time. The one exception was a 6mo study abroad I did at Oxford, where I basically had 1:1 tutoring on nothing but deep learning. That was truly a fantastic experience, but, still. Even though I do stand by the Oxbridge system — they have really figured out this whole education thing over the last 800 years — I felt like I would have been much better served if I just skipped the school thing, pocketed the $250k or whatever my parents spent, and made my way to San Francisco.

Part of the reason I felt this way is because I happened to meet a cohort of folks who did exactly that. For a variety of reasons they didn't go to college, and ended up in the Bay. By the time I met them, they were all incredibly accomplished people. Meanwhile, I with my fresh newly minted degree felt like I was miles behind. And I lamented the four years that I wasted in university.

I'm a bit older now. And I've met a lot of the folks who have moved to the Bay explicitly to start companies after choosing not to go to college. And let me be very blunt: all of them are arrogant little shits.

When I was 18 I was full of confidently incorrect opinions. And upon arriving at the gates of my school I basically immediately ran head-first into a whole bunch of other 18 year-olds with confidently incorrect opinions that were totally different from mine.² That experience polished off a lot of my rougher edges. It taught me epistemic humility; it forced me to be more of a team player; it made me tolerant.

The average 18 year old who is going to the Valley today is getting the opposite experience. They are being rewarded for their youth and opinionated ideas. VCs and billionaires are pouring money into college dropouts. And as a result, these already-too-self-assured kids have their egos blow up like a Macy's Day Parade float. If some billionaire gave me money when I was 18, I would a) immediately become the most insufferable person alive, and b) would do anything that billionaire wanted forever. This is not a healthy environment for these kids. Almost everyone I know who has come out the other side of this experience has become extremely jaded.

I bring this up as a separate call out because I know people like Elon love to exalt first principle thinking, and openly denigrate the college experience. I think that position is getting more and more popular as the university system is under open attack from the right. And, look, there's definitely a class of person who can make it out to Silicon Valley at 18 and thrive. But if you ask me, the average person is simply going to fall through the cracks and be significantly worse off for it. The Thiel fellowship works because of selection bias. It's always selection bias.

Can we fix it?

At the end of all this pontificating, it's worth at least trying to grasp at some solutions.

One popular camp says that we should just get rid of universities.

Right well. Trump is doing that. Funny how when you see your ideas actually implemented you realize how terrible they are.

As it turns out, no, we shouldn't just wholesale get rid of universities. In the wake of the wanton destruction caused by the Trump research grant cuts and funding withholding, some of the most ardent anti-university crusaders are realizing the level to which the government-university research and leadership pipelines are a critical load bearing part of American supremacy. That's in addition to the real benefits to having systems for higher education, credentialing, and "personality polishing".

Another camp demands free college for everyone. This is significantly less stupid than the first idea, but it's still a pretty bad one all things considered. Contrary to the average liberal arts major, I don't think endless education is an unalloyed good. Besides the obvious costs to broader society, there is a real cost to students in the form of time spent in a classroom during what should otherwise be incredibly productive years. Many of the supporters of free college argue that it will help everyone get a job, which only betrays that their liberal arts degrees were not in economics.

The results of free college for everyone means everyone will have to go to college and the status and credentialing game will get even more costly. It's just another tulip bubble (and you should really read that article if you haven't, it's Scott Alexander at his best).

Ok, now that we've dismissed other people's bad ideas, let's talk about my bad ideas.

The first proposal is actually not my idea at all, it's Scott's from the article I just linked:

I’d propose a different strategy. Make “college degree” a protected characteristic, like race and religion and sexuality. If you’re not allowed to ask a job candidate whether they’re gay, you’re not allowed to ask them whether they’re a college graduate or not. You can give them all sorts of examinations, you can ask them their high school grades and SAT scores, you can ask their work history, but if you ask them if they have a degree then that’s illegal class-based discrimination and you’re going to jail.

I actually really like this idea! It's extremely elegant, because it restructures our incentives to neatly cleave the two universities in half without any additional complication or carve out. The existing institutions remain as places of higher learning and research. And the market creates an entirely separate category of credentialing, mostly dominated by standardized tests like the GRE or the LSAT.

To flesh this world out, imagine a suite of exams that test every proficiency level from highschool to advanced PhD lifetime of research. Like modern accreditation exams for engineers, or the SAT / LSAT / GRE / MCAT for general competency, these exams would all be fairly cheap because there are a lot of people who can provide such exams and because they have low personnel cost once created. They act as independent credentials. Someone who passes the extremely-difficult-theoretical-physics exam likely knows a fair bit about extremely-difficult-theoretical-physics. We don't need to know how that person studied for the exam; for the average quant fund it doesn't really matter because they are going to retrain the physicist anyway. And of course you could create exams for a wide range of industries. The only reason we don't have something like this already is because a college degree is simply a good enough signal that the market demand doesn't exist. But if college degrees became a protected class, you'd essentially force the market to shift its criteria around.

There are a few downsides. The first is that a lot of people will just not go to college and so lose the formative benefits of being around a cohort of similarly minded peers. The second is that our research institutions depend on bodies, and part of the incentive of doing research is getting a job of some kind; less incentive to get a degree means less research, which seems bad. The last is that it is unlikely the political left will go for this. College experience is an exalted part of the average lefty's world view, in large part because many left wing leaders in the modern world are representing the professional managerial class more than they are the working class. That means that any attack on colleges as an institution will be at best frowned upon.

And the universities themselves are pretty well equipped to defend against attack, often quite literally.

The second proposal is inspired by the Israeli startup ecosystem. Israel has mandatory military service. It's roughly about 3 years. Because it is mandatory, you are put into an environment with a bunch of peers. It's not strictly an educational environment, but you end up learning a lot. Many of the folks end up skipping college entirely because they more or less have a practical education in some actually useful skills by the time they leave. The reason Israel's startup ecosystem punches way above its weight is because some these kids go into the information arms of the military, learn a ton about security and software, and come out with a super focused and determined attitude in addition to the obvious ability.

The US should definitely increase the range of options for incoming recruits to its own military, such that a student could reasonably go into the armed forces and come out with experience equivalent to a degree. Current programs like ROTC or the various academies still depend on a school-like system, which seems wasteful when you can clearly throw these kids directly into useful roles while still teaching them.

But I also don't think the military angle is a necessity here. You could imagine an equivalent program that is entirely domestic and focused on internals — a civil corps. In the ideal setting, a civil corps would be able to give degree-equivalent experience in fields like civil / mechanical / electrical engineering through its infrastructure and construction arms; health and bio through work with the HHS and NIH; software through a sorely needed government-funded open-source division; and, of course, a raft of public policy experience. The benefit of a civil corps is that it can easily double as a trade school or apprenticeship program for folks who are uninterested or unable to work in white collar areas. Instead of taking a 4 year degree in communications that will put them into thousands of tax-payer-funded debt, those people could easily just start building roads or running tolls or whatever else.

There is, surprisingly, some history for this. During the Great Depression, FDR created the Civil Conservation Corps (or the CCC). This was a work relief program for men aged 17-28. Among the New Deal programs, the CCC was one of the most popular — in addition to the obvious morale boost that came from being able to check notes eat food, the program led to increased long term employability among the folks who participated. So it was a win-win: the government got cheap useful labor to actually execute projects; and the kids got paid in a bit of cash and a lot of experience.

There are two practical impediments to this. The first is that you need to make the civil corps alternative compelling, which means it needs to start with taking extremely talented students and giving them legitimately extraordinary opportunities. That branding will need to build up over a few years before the program as a whole can develop into a meaningful alternative for the university. The second is that there is unlikely to be any political appetite for this from the American right. MAGA has staked the position that universities must die, but ALSO government spending must be cut. This just…leaves a vacuum. In the MAGA ideal world, there are no credentialing institutions AND no education institutions. You could imagine some sort of halfway watered down compromise, but it is almost worse to half-ass a program like this since you are partially taking stewardship of the future careers of the students involved. That in turn makes it difficult to imagine this taking off the ground — it is just too nuanced a position.

This man is famously incapable of nuance.

A third proposal is to try and separate higher education and research from career preparation. In this model, you create a crop of new programs that look similar to universities but are much more practical in focus, may involve explicit partnerships with various employers, and may have income-share agreements with students conditioned on placing them in high paying jobs.

The original attempt at this was a startup called Make School, which was specifically focused on the software industry. I happen to know the founders and some of the early students who went through the program, and though the universal opinion was that the startup struggled due to the inexperience of the founders in running a school, the idea was fantastic. Make School helped hundreds of kids land high paying jobs while providing a decent and practical crash course in CS. The students got to learn for free, and only paid back part of their salary after the fact. No salary, no payment. It aligned incentives really well.

Unfortunately, Make School ended up shutting down because it was just hard to run. But it also started back in 2012, which is insanely ahead of the curve. I imagine that any equivalent that started today would have a significantly easier time of it, since the university system is already under attack from so many different angles.

Two problems with something like Make School. First, there is a chicken-egg signalling issue. The point of these career prep programs is that they implicitly act as a credential of their own. But it is hard for someone at Google to evaluate what it means to graduate from your new program. Harvard benefits from literal centuries of name recognition, what is MySmartSchoolAcademy.com going to do to overcome that? Second, for-profit universities get a bad name for a reason. There is just a lot of scamminess with these kinds of programs. I think the ability to align incentives by doing income-sharing helps a lot. Still, there must be something about the space that attracts scams for there to be so many of them, in much the same way that there’s something about crypto that attracts scams.

---

I admit that all of these suggestions have weaknesses. But we shouldn't let perfect be the enemy of the good; the more nuanced suggestions seem both a lot more practicable and a lot less dangerous than the more extreme proposals. And the upside of all of these suggestions is that the market will naturally bring prices down, allowing more people to earnestly pursue higher education as it was intended.

In the immediate short term, I would love to see an ambitious startup take a crack at the credentialing problem using exams. I think it's extremely doable to make some exams; the hard part is the same chicken-egg advertising problem. No recruiter ever got fired from hiring a Harvard grad; they may get fired if they hire someone who's only credential is your random test. But if you can get some big names to vouch for your exams — like Terrance Tao for a suite of math proficiency exams — and get some big names like Google to openly consider your exams as qualifications, you may be able to make some meaningful headway. Leetcode does something similar, and while they are obviously aiming for a different market niche it is still a useful comparison to see how far you can go.

All things being equal, as of right now I think I am still sending my kids to college. I believe in the value of education independent of its effects on salary, and as a result am willing to pay huge amounts for my kids to access education even if they can't pay off the related cost. For me, education is very much an end in itself. But…look, if the price of a degree quadruples again in the next 15 years, my kids are still going to get an education, but they're getting shipped to the UK instead.

They have some very good sticky toffee pudding in Oxford.

1. As an aside, the boundaries on this are almost purposely fuzzy and kafkaesque. CS is proudly open source, but also you are not allowed to talk to anyone about what you are working on, but also you all live in dorms together and study in the same place and are encouraged to talk about what you learn, but also if you talk about this particular assignment with anyone we are sending you to university jail.

2. I joined the debate team, so I was like really exposed to this. But even my daily interactions with friends in the lounge had a lot of clash.