Table of Contents
For nearly a decade and a half between the 2008 Financial Crisis and the end of COVID-19, an unprecedented bargain was presented to college graduates: learn to code, become reasonably technical, and you could out-earn investment bankers and consultants while working for cushy companies that would provide you with everything from a massage room to indoor go-karting.
You didn’t even have to be particularly good. Mediocre programmers and boot camp graduates could claim a spot in an industry where ever-increasing demand for technical labor always outpaced supply.
The scalable economics of software created a historical aberration where fresh college graduates could make six-figure salaries and become fantastically wealthy in a relatively short period of time, through stock options and equity, without working eighty hours per week and running through the competitive gauntlet of firms like McKinsey or Morgan Stanley. The tech boom even led one tech mogul to compare California to Saudi Arabia: an incredibly lucrative sector underwrites corporate and ideological excess, whether Woke 1.0 or Wahhabism.
Before this, learning a narrow, technical vocation (even if highly skilled) was merely a gateway to a stable, upper to upper-middle class life, instead of a shortcut to becoming a multi or deca-millionaire.
Technology – and especially software – powered Stanford’s rise from regional to national prominence; we even displaced Harvard as the most sought-after “dream school” in 2013 precisely because Computer Science (CS) replaced finance as the “golden ticket” career. Goldman was cast aside for Google, as CS became the default high-status choice for Stanford’s strivers.
Until interest rates began rising in 2022, it was difficult to argue against pursuing this extraordinary path.
But the golden ladder for the CS major has been largely pulled up. After the Fed hiked rates, venture funding dried up, future growth mattered less, and companies that had hired years ahead of demand suddenly had to care about profits. And even the lucky ones who break into tech post-COVID are now expected to work “996” rather than 30 hours per week from an Airbnb in Lake Tahoe.
As CS has become maximally popular, the supply of technical graduates has finally outstripped demand, leaving the Stanford striver with a terrifying conclusion: his major has become the victim of its own success.
A technological and sociological tragedy
Part of this is a technological tragedy. Much of the progress made by computer scientists directly precipitated the decline in the value of ordinary software engineering, and AI tools are now becoming increasingly competent at the job. Even at Meta, employees' keystrokes and mouse movements are tracked to directly train models that will replace them.
This means that the best CS graduates, who can leverage AI to vastly increase their productivity, will bifurcate upwards away from the merely acceptable ones whose skill was pushing PRs. The latter probably won’t disappear overnight, but they’ll face wage compression, job insecurity, and a steadily rising bar before their work is eventually automated. Unfortunately, the sociological evolution of Stanford’s CS major will produce far more of these mediocre engineers.
When systems become sufficiently popular, they attract students who optimize for the system itself. This creates a paradox where the system becomes more efficient and more competitive than ever, while simultaneously producing more interchangeable people. When everyone optimizes towards the same legible path, you lose the natural spikiness that produces outlier outcomes.
When CS was previously just another STEM major, or even viewed by some as a backup option for failed mathematicians and electrical engineers, it was only studied by those genuinely interested in computers. Once companies needed enormous numbers of software engineers, CS attracted strivers who wanted the highest-return credential available and optimized accordingly. They take the right classes, grind LeetCode, get the right internship, and follow an increasingly standardized path into a prestigious technology company.
Stanford CS majors become more polished, but also more concentrated around a median. The system selects increasingly well for people who know how to pass through the system.
This is the problem afflicting Y Combinator; it is more selective than ever while simultaneously becoming lower-signal. Once entrepreneurship itself becomes prestigious and legible, the same risk-averse ladder-climbers who once would have gone to Goldman learn how to cosplay as founders. When phenomena become more popular, the influx of newcomers does not necessarily make it higher-quality.
The Stanford CS major is no different. Classes once intended for a relatively small group of technical specialists now have to accommodate enormous numbers of students with much wider distributions of ability and interest.
Stanford deliberately chose to become a “pump, not a filter,” scaling the major through recorded lectures, armies of TAs and automated grading rather than restricting access. Since ChatGPT was released, it has become even easier to skate through the major without actually becoming a competent programmer. Students can outsource much of the cognitive work on problem sets and code to AI, and unless they are especially careless, only the most egregious cases are ever likely to get honor coded. The curriculum may not have formally gotten easier, but it has become much easier for a mediocre student to cheese his way through it and emerge with the same credential.
And when his golden ticket stops redeeming into a cushy job, the response of the Stanford man isn’t introspection. Rather, it’s more credentialism. He pursues a coterm to extend adolescence and indecision, postponing the labor market and adding another line to his résumé rather than asking whether CS was ever actually his comparative advantage. This is how Stanford can graduate more credentialed, mediocre, and unemployed CS majors than ever before.
Stanford consensus realizes it too little, too late
The Stanford student body isn’t hopelessly incognizant. Anecdotally, I’ve noticed an appreciable drop in underclassmen majoring only in CS, with many adding humanistic minors or double majors, or choosing math, electrical engineering, or economics (for the less technically-inclined.) Though CS still dominates Stanford, the number of CS majors declined in the class of 2026, the first where ChatGPT was available to students, just as CS coursework occupies a smaller share of the most-enrolled classes. At the UCs too, CS enrollment is dipping for the first time since the dot-com crash, as even their best students struggle to gain employment.
But does majoring in economics really future-proof your career? The more intangible skills required for consulting or investment banking, like making compelling PowerPoints or competent Excel models, may be harder to optimize for than producing good code, but they are not immune from the spectre of automation. Multiple friends in consulting, even at MBB firms, have admitted to me that they feel increasingly like “physical AI wrappers.”
Though a Stanford degree may extend a striver's employability for a while, that just means unexceptional Stanford students are simply up later on the chopping block. The idea of any credential serving as a golden ticket is vanishing, and the old social contract – optimize hard enough, acquire the right credentials, and receive a high-status job at the end – is now disappearing for the highest echelons of college graduates first.
The fifteen-cent degree
Imagine it’s July of 2027. A student who entered Stanford thinking he’d become the next Mark Zuckerberg in 2022 has received his MS in Computer Science. He doesn’t have a job or any sense of what he wants to do with his life. What did that degree get him?
Clarence Thomas thought much the same after finishing his education, and attached a 15 cent sticker from a box of cigars to his diploma from Yale Law School. Major law firms, who assumed his JD and grades were earned through affirmative action, declined to employ him, leaving him unemployed and directionless until he joined the Missouri attorney general’s office. He “never did change [his] mind about its value.”
Unjust racism motivated law firms to discount Thomas' ultra-prestigious degree. The Stanford CS degree has a much more sensible reason to be discounted: its supply has exploded, its signaling value has weakened, and AI is commoditizing the underlying skill. The wrong instinct is to flee to another golden ticket. Stanford is not a trade school, or a factory for high-end vocations. If all you want is to become a technology employee, go to a16z Academy.
Stanford was created to form a future leadership class for the American West. For fifteen years, the CS boom created a historical anomaly where a student could join a handful of clubs run by the “boba mafia,” get a cushy internship in Menlo Park or Mountain View, master one narrow technical skill, and become a multimillionaire without learning much about how the real world actually works. He did not need to write especially well, develop commercial judgment, understand institutions, know much history or politics, or learn how to lead people. The market rewarded him handsomely because, for just over a decade, his labor was extraordinarily valuable.
Many of my smartest and most successful friends still major in computer science. But their degree no longer engenders their success; they succeed because they are exceptional beyond it. If you want to avoid ending up with a fifteen-cent degree, stop searching for the next golden ticket. Find what you are unusually good at, learn enough about the world to exercise judgment beyond a narrow vocation, and become valuable in a way that can’t be reduced to the letters on your diploma.
