FWIW, I personally consider MIT, Mudd, Caltech and Olin to be the “top tier” undergrad schools in the US; Harvard/Stanford are one step down. If I were to make a case for Harvard, the case would be “well, you can cross-register for classes at MIT”.
MIT, Mudd, Caltech and Olin are basically the only undergrad schools at which ~100% of the students are top-tier in STEM specifically, and are required to take the full set of core STEM courses. Therefore they are basically the only schools where all the courses can be designed on the assumption that ~100% of the students are top-tier in STEM and have all of the core background knowledge. Courses will generally not need to review basics of calculus, courses past freshman year will generally not need to review the basics of linear algebra or differential equations or mechanics, because basically everybody knows all that. The result is that most courses can spend less time reviewing basics, and more time covering the main substance—and that advantage further compounds over time.
(Source of my models here: I did my undergrad at Mudd, have audited courses at Stanford, and have used lecture videos from MIT pretty heavily and Harvard once or twice. I’ve also worked at the sort of tiny high-frequency trading firm which needs really top-tier technical people; the person who handled recruiting there told me that they only bother with MIT, Mudd, Caltech and Olin. You can find top-tier people elsewhere, but it’s generally a much smaller proportion of the student body.)
(Also there’s more to my models than just this—e.g. MIT, Mudd, Caltech, and Olin all share a kind of culture that I think is pretty important. But that part is less legible.)
FWIW, I personally consider MIT, Mudd, Caltech and Olin to be the “top tier” undergrad schools in the US; Harvard/Stanford are one step down. If I were to make a case for Harvard, the case would be “well, you can cross-register for classes at MIT”.
MIT, Mudd, Caltech and Olin are basically the only undergrad schools at which ~100% of the students are top-tier in STEM specifically, and are required to take the full set of core STEM courses. Therefore they are basically the only schools where all the courses can be designed on the assumption that ~100% of the students are top-tier in STEM and have all of the core background knowledge. Courses will generally not need to review basics of calculus, courses past freshman year will generally not need to review the basics of linear algebra or differential equations or mechanics, because basically everybody knows all that. The result is that most courses can spend less time reviewing basics, and more time covering the main substance—and that advantage further compounds over time.
(Source of my models here: I did my undergrad at Mudd, have audited courses at Stanford, and have used lecture videos from MIT pretty heavily and Harvard once or twice. I’ve also worked at the sort of tiny high-frequency trading firm which needs really top-tier technical people; the person who handled recruiting there told me that they only bother with MIT, Mudd, Caltech and Olin. You can find top-tier people elsewhere, but it’s generally a much smaller proportion of the student body.)
(Also there’s more to my models than just this—e.g. MIT, Mudd, Caltech, and Olin all share a kind of culture that I think is pretty important. But that part is less legible.)