September 1, 2026
Dear Leland and Everett,
Today I want to talk about higher education.
College education has become something of a flashpoint in American politics. In the previous administration (when Democrats were in power) the federal government undertook to forgive college debt. In Democratic presidential primaries candidates often talk about how we need to make college education free for everyone. Now that Republicans are in power, they have managed to shift the tone of the discussion, slinging accusations at universities ranging from discrimination to incompetence to ideological capture. I’m less interested in taking sides in the debate, and more unpacking how I think we got to where we are today.
My original assumption (and I think the default American assumption) was that attending university taught us skills directly applicable to our future careers. However, I was frequently advised not to worry too much about picking my future career too early – and, by extension, worrying about picking the right major to prepare me for said career. Examples abounded of people who found meaningful careers in fields completely unrelated to their degrees; the common thread seemed to be that their degrees taught them valuable skills regarding how to think or write, which proved to be enough to figure the rest out on the job. So I went into college assuming that the primary objective was to get a degree directly related to my future career, but understanding that folks often navigated meaningful careers thanks to the indirect skills developed while getting their degree.
My current understanding of higher education is broader. I still believe people get degrees in order to learn information or skills relevant to future careers; the difference is that I now understand skill acquisition to be merely one of three reasons for going to college. And I kinda suspect skill acquisition is the least important of the three.
When I was in business school, a professor introduced a hiring theory of higher education. This theory suggested that universities provide something of a filtering function for prospective employers, making it easier for employers to identify a pool of qualified applicants. The extensive work universities do to screen highly qualified applicants becomes something employers leverage.
As a hiring manager, I came to experience the value of university filtering firsthand. Most paths toward identifying qualified applicants proved time intensive. When my team posted jobs, lots of people applied, but the overwhelming majority of these applicants proved unqualified, or at least a poor fit. Internal recommendations proved the most reliable way to hire: when someone working for the company understands the job requirements and recommends someone from their professional network, these recommendations succeed at a remarkably high rate. But the second most efficient method for hiring qualified candidates came from identifying select university programs where we experienced track records of hiring success. In our specific case, we discovered MBAs (and particularly MBAs with backgrounds in finance, consulting, or engineering) generated the highest yield. With some schools I venture we had hiring rates approaching 10% of applicants, which was orders of magnitude better than online applications. So, hiring at these programs saved us enormous amounts of time and energy in identifying and hiring quality candidates.
The third reason people go to college is for what I call the “country club benefits”. Nominally, people join country clubs for access to the amenities; really, people join country clubs for access to a society of high status individuals. Similarly, people nominally go to college to learn relevant jobs skills, but really go to college for access to high society. There are multiple layers to the country club benefits. The first layer represents the relationships one develops at college: these tend to be lifetime friendships with people who help us navigate the world the rest of our lives. Forming particularly close relationships with high caliber people is enormously valuable. The second layer represents access to polite society more broadly. Lots of jobs require a college degree – even when, again, the degree isn’t directly applicable to the scope of work. College degrees play such a filtering function in our society that lots of college graduates hardly interact with non-graduates in any meaningful way, and vice versa. This social segmentation between graduates and non-graduates represents a huge change within the scope of my lifetime: relatively few people in my parents’ and grandparents’ generations went to college, so when I was growing up graduates interacted with non-graduates regularly. Nowadays this segmentation means college graduates have access to better information, more impactful social networks, and more high-paying job opportunities by the mere fact of having earned a college degree.
On some level, I think colleges figured out how critical they were in terms of bestowing access to high society, and undertook to extract as much value in the exchange as possible.
When I was an undergrad at a small liberal arts school in Texas, I was struck by the caliber of student the university recruited from all over the country. I later came to realize that most of these high caliber students were on scholarship. I also observed a substantial population of private school kids from big cities, particularly nearby Houston and Dallas. I later came to understand these students were often less impressive, and their parents were usually paying the full cost of tuition. At our most elite universities, what I have come to understand is that the same dynamic plays out in a more concentrated form. Elite universities compete fiercely to attract the nation’s best and brightest across a litany of fields. Elite universities also allow our nation’s most connected to compete even more fiercely for the opportunity to attend the university at full cost (or sometimes even well above full cost – with extra donations required to secure a space for less qualified legacies). While this arrangement is not commonly understood, and would likely be unpopular if it were understood, I do find some merit in the arrangement: the nation’s best and brightest benefit by gaining access to the nation’s most connected, and the nation’s most connected benefit by gaining access to the nation’s most capable and talented. Together, the talented cohort and the connected cohort help the university graduates achieve more success than either cohort could achieve independently.
We’ve seen the nominal cost of tuition skyrocket over the last 40 years. The real cost of tuition has increased faster than inflation, but more slowly than the nominal cost. That’s because most students don’t pay full tuition. Most students qualify for scholarships or financial aid; universities keep raising prices in part to maximize how much they are able to charge the richest prospective students.
Even still, real tuition costs rose faster than GDP continuously for decades. Interestingly, those increased costs didn’t go to teaching: teaching costs remained roughly fixed in real terms. Instead, administrative costs exploded. Universities now employ many, many more administrators than they did when I attended. What value these administrators bring is unclear to me. My hunch is that they are meant to justify the increased tuition costs charged by providing extra services.
By definition, no industry can grow faster than the economy forever; if it did, the industry would eventually become the whole economy. Thus, higher education costs eventually have to stop growing, and some signs suggest they may have already peaked. Indeed, universities now face some interesting headwinds. For one thing, the population of college age students is shrinking, thus shrinking the pool of American candidates for college. Interestingly, the percentage of American kids attending college appears to be shrinking as well, and a shrinking number of young people combined with a shrinking percentage of young people attending college implies a shrinking overall demand for college education. One way colleges appear to be responding is by recruiting more international students: whereas there were roughly half a million international university students (undergraduate + graduate) when I graduated around the turn of the century, that number has roughly doubled since. International populations provide universities an opportunity to attract more of the world’s best and brightest, and also more of the world’s willing to pay full tuition costs.
Even still, university attendance is down from an all time peak in the early 2010s, and the implied dynamics will be interesting to watch. Will universities be able to convince more American’s to go to college? Will they continue to increase international enrollment? Will they be able to offset declining enrollment by increasing tuition costs? I don’t know the answers, but the implications could be profound.
Of course, like with so many industries, AI will have a huge impact on higher education. I anticipate AI will impact not only what we learn, but how we learn. Sometimes techies anticipate the impending doom of higher education because they anticipate learning becoming much more efficient. Personally, I find these mental models oversimplified, for the reasons outlined in this letter: tech optimist / higher-ed doomers are focusing on the skill acquisition element of higher education, but ignoring the hiring benefits and country club aspects. Before anyone can truly disrupt higher education, they must first figure out how to either help employers surface high quality candidates efficiently or disrupt the country club benefits of higher education.
The most likely path forward, were I an entrepreneur looking to disrupt the field, would be to focus on the highest caliber students. These should be the easiest to empower, the easiest to teach new skills, the easiest to help demonstrate their potential worth to prospective employers, and the easiest to help find other ways to gain access to connected society (high society is always on the lookout for fresh talent). Once the highest potential candidates begin eschewing higher education, the whole edifice begins to break down. The second tier of talent will begin to follow the path laid out by the most talented; eventually, the wealthy and connected students will begin to understand they are no longer paying for access to the best and brightest, and will presumably reduce their willingness to pay accordingly. I won’t make any precise predictions, only that I do expect some disruption to higher education in the coming years.
Which is probably a good thing. I don’t like the degree to which a college degree has become a dividing ground between the haves and the have nots in America. I don’t like how the division has been, at least in part, self-reinforcing: lack of a degree starves an individual of access to other talented people and potential jobs, thus permanently reducing job prospects. Once upon a time I saw college as a great equalizer, granting folks from different socio-economic backgrounds equal opportunity to earn their way in the world. I’m not sure whether that mental model was ever true, but these days I see college as a tool that appears to be solidifying a caste system of sorts, with graduates and non-graduates occupying separate and distinct places on the social hierarchy.
One thing that will be interesting to watch is whether AI increases or decreases the value of intelligence. Intelligence appears to be a huge differentiator in career outcome: it strongly influence which students get into elite institutions, but also appears to influence how much those who graduate from elite institutions earn (a business school professor once told the class he ran regressions to understand what factors influenced future income, and the strongest predictor by far was the score on the entrance exam). On the one hand, AI will be smarter than all of us, so perhaps will dramatically reduce the value of raw intelligence by leveling the playing field. On the other hand, perhaps AI will turbocharge the capacity of the intelligent while the less intelligent struggle to take full advantage of AI’s potential. I honestly don’t know. Either way I expect AI will shake up higher education, and to the extent that means disrupting the formation of castes, I think that’s a good thing.
Of course, the two of you are only a few years away from making your own decisions regarding higher education, and by that time the disruptions will become apparent. A hiring manager at Google recently told me he’s increasingly using AI to write the code he used to assign to entry-level engineers. What I realized from talking to him: our educational system is designed to teach students how to *do* things: write code, draft a document, summarize research, or build a model to simulate an outcome, among others. As AI overtakes our capacity to do these activities, I think we need to start leveling up our students such that they think less in terms of producing something, and more in terms of how to edit the thing. Typically in a workforce the entry level staff produces the first version of the thing, and then their manager or supervisor helps modify or edit the thing into its final shape. Executives then help the team shepherd the output through to adoption. My suspicion is that all levels of the stack will need to “level up”. Entry level workers will need to develop a capacity to assign work to an AI engine and then refine the outputs until it is ready for public consumption, and then managers will become responsible for shepherding the output through to adoption. That shift will require a dramatic rethink in terms of how we teach. We need to teach you how to work with AI to create things that are ready for public consumption. Interestingly, the best way to do this will be with real world experiences: classroom simulations are likely to decline in value while we increasingly test out real world applications in monitored settings. As we demonstrate competence for producing work of real value, students will find their own paths away from the learning environment and out into the professional world. The lines between classroom and real world are thus likely to blur. All of this seems interesting and promising, at least to me.
What this means for higher education, I don’t know. I anticipate there will always be institutions dedicated to cutting edge research, and I anticipate there will always be students interested in learning from those doing cutting edge research. But I think the future of education is far less lecture-based, and far less everyone-does-the-same-homework driven. I think people will learn at their own pace, directed by their own interests and abilities, and classes will increasingly become geared toward solving real problems (or, in primary and secondary education, at least simulations of real problems).
The two of you are likely to be caught somewhere in between: as of today you are mostly learning using legacy methods. By the time you finish high school and (presumably) enter college I anticipate AI job disruption will be rather fierce. So you are unlikely to have grown up in AI-native classrooms, but are likely to encounter a job market significantly disrupted by AI. We have some time to gauge which way the winds are blowing and try to react proactively. We’ll see how we do.
In the meantime, I anticipate higher education will remain something of a political and cultural flashpoint for the foreseeable future, for the very reason that the tool we’ve long relied upon to train us for the future may longer be able to do so.
I love you.
Love,
Dad
Letters to Leland 