The source doesn't break out EECS from the rest of MIT. To be fair, though, this looks to be really long term data. If you dig into the numbers (http://gradeinflation.com/MIT.html) things look a lot more steady over the last 30 years. It looks like there was significant inflation through the 1960s but it flattened out since then.
It's very important to look at the time series, for the '60s were a special case of giving (male) students high enough grades so that they wouldn't flunk out and become eligible for the draft and service in Vietnam. I don't know about MIT's history of grading in that period, but I wouldn't be surprised if that happened.
In the early '80s EECS drastically changed it's core curriculum (or at least the CS part of it, SICP/6.001 was the start), and it definitely got better at teaching it.
If that also gave the students a better foundation for higher level courses, then we'd expect significant grade inflation based on MIT's policy of grading on mastery instead of the curve.
Also during that period the EE to CS ratio was 3:1 and flipped to a mirror 1:3. If CS is better or markedly better taught ... and be sure to factor in higher availability of computers, before MIT and at MIT ... that probably also is a factor.
Assume also during this period that only half the majors had a "calling" to EECS; therefore the quality of teaching would make the most difference to the other half to whom it doesn't come so naturally.