The two educations, 1983–1992
Berkeley engineering enforced a 2.7 mean and taught that nothing is fair. Stanford GSB suppressed grades entirely and handed out a T-shirt telling everyone to stop competing. The same person went through both.
Hare went through both regimes personally, four years apart, and the contrast is the cheapest available demonstration of the argument the Governance Premiums rest on.
Berkeley engineering enforced a 2.7 mean, gave him two electives in four years, and taught him that nothing is fair. Stanford suppressed grades entirely, barred recruiters who asked, and issued a T-shirt telling the class to stop competing.
He was the same person in both. The behavior around him was not, and nothing about the students had changed — only what each institution chose to count and to show.
It is the cheapest available demonstration of the argument the Governance Premiums rest on: behavior under pressure is a property of what a system measures and shows, not of the character of the people inside it.
Two institutions took similar people and produced different behavior by changing what they measured and what they showed. Berkeley enforced a 2.7 mean and taught that nothing is fair. Stanford suppressed grades entirely and handed out a T-shirt telling everyone to stop competing. Both worked. Neither outcome was about the students.
The trail exists so that people who were there can reconstruct their own memories, recognize what they helped incubate, and add to or correct the record — or forward the few pages that belong to someone they know.
Two institutions took similar people and produced opposite behavior by changing what they measured.
Stanford's non-disclosure was explicit institutional design — recruiters who asked were barred from campus.
One signal leaked anyway, and firms spent real effort inferring it, which is Spence's problem appearing in the wild.
Hare lost the top-graduate prize by dialling into Microsoft's network instead of attending class.
Engineering ran a required mean of 2.7. Two electives in four years, and only then if advanced placement let you skip the usual freshman requirements. Brutal, and deliberately so.
The funding was extraordinary, and so was the faculty: engineering professors were paid at the average of the top universities in the country, so Berkeley paid the most. The consequence was that they mostly did research. Undergraduate classes of five hundred to a thousand were normal, because the professor was off winning a Nobel Prize. There were around twenty-nine laureates on campus in my day.
My summary, delivered widely and not entirely in jest: Berkeley is the top research university on the planet, and Stanford is a trade school. A very nice one. Like a country club.
Grades ran H, P+, P, P− and a fail. Students could not report them. Recruiters who asked in an on-campus interview were barred from campus recruiting. First-year honors were not published. Nothing was released until after graduation, and a recommended curve in each course capped what any professor could award.
During intro week the class distributed a T-shirt carrying a mock dictionary entry: Stanford Graduate School of Business, a two-year vacation. A non-aggression pact, issued as clothing. The message was that you had clawed your way in, this was the end of the academic train, and it was time to stop backstabbing.
One signal escaped: Arjay Miller Scholar, the academically highest ten percent of the class, named for the school's fourth dean — formerly president of Ford. Firms spent real effort inferring what they were forbidden to ask.
I was one. I was also leading for the Henry Ford II Scholar — the single top graduate — going into my final semester, and lost it by blowing off my classes to dial into Microsoft's network on one of the first copies of Remote Access Server, taking P-pluses instead of As. I rescheduled a flight to Europe by two weeks to attend a graduation where somebody else received it.
He came from Berkeley engineering and was intimidated: Stanford was hard to get into, everyone was from somewhere impressive, and everyone seemed rich.
Then he worked, in the way Berkeley had taught him to work, and discovered the gap was enormous.
Remove the ranking and people cooperate. Publish it and cut the bottom decile and they compete. The population is the same in both cases. This is the Governance Premium argument in its cheapest and clearest form, observed from inside both regimes — and it is why the architecture treats behavior under pressure as a property of what a system counts, rather than of the character of the people counted.
Episodes are building blocks. The same material appears in more than one where it belongs in more than one, and every claim traces back to a primary source.
1990–92 — the faculty, the classmates, and the prize he gave up
Open →In the graph“Imaginary engineers” — third floor, Etcheverry Hall
Open →In the graphStanford engineered collaboration. Harvard engineered competition. Both worked.
Open →In the graphFord's president, Stanford's fourth dean, and the name on the award
Open →In the graphDeferred Stanford to command a tank in the Gulf
Open →In the graphHarvard then Stanford, Tesseract, Benchmark — and a game of chicken with a city bus
Open →The account this was distilled from, preserved as dictated — verbatim and immutable.
Read the source →In his own words · S-155The account this was distilled from, preserved as dictated — verbatim and immutable.
Read the source →