What really happened is that when enrollment also crashed, the students still taking the major were serious about it per se, not just as a good and high paying career (hah!).
So nowadays almost all the students are making As and Bs, with the notable exceptions being those going through some sort of personal crisis unrelated to academics.
And thinking about it, to the extent the 1/5 of MIT's students who are no longer taking EECS are instead in majors that they are a better match for, I'd expect their grades to increase as well.
Or look at the cancellation of the Superconducting Supercollider. Some departments had lines of students switching their majors out of physics; to the extent those students found their new majors easier (physics is hard) I would expect a bit of "grade inflation" as well.
A final note on MIT: this is not the first time this site has been forwarded to HN, and I'll note again that its grade inflation ended in 1970. It was pretty much a phenomena of the turbulent cultural '60s (which didn't quite start in 1960), which is not surprising for a whole bunch of reasons, e.g. I'm pretty sure MIT decided to make itself a bit less difficult to succeed in (e.g. when I attended in 1979-80, the entire freshman year was pass/fail, now it's only the first term, and I'm sure before the '60s they didn't do this).
Oh, that rule of thumb predicts 3.00, while MIT including the freshmen is 3.17. Given the tremendous self-selection in the applicants, that's interesting.
That does explain quite a lot about economics!
"As a rough rule of thumb, the average GPA of a school today can be estimated by the rejection percentage of its applicant pool: GPA = 2.8 + Rejection Percentage /200 + (if the school is private add 0.2)"
Interesting the more students a school rejects the higher its GPA seems to be. To say it another way, the more selective an institution is the higher the GPAs. ie. If you get smarter students, they tend to have higher GPAs. Example my econ professor has been looking at the actual distribution of performance in his class. He does not find a bell curve. Additionally he asserts there is no reason to expect a bell curve in a college course, as students tend take courses they are both interested in, and good at. The consequence, since you don't have an even distribution of stupid students to smart students, you won't have an even distribution of grades.