The undergrads are more involved with research than at most other schools (huge range in quality/difficulty/etc, but the option is there). The teaching quality doesn't really suffer from the focus on research. Blackboard style classes probably aren't much better than any other school, but lab classes were quite good.
To be clear, I'm not being critical here, it was an impression from my visit. Lots of people I know (like gumby) and others have gone to MIT as undergraduates and really felt it was worthwhile. I really wanted to go to Caltech but only landed on the wait list and settled for USC (who offered better financial aid so there was that)
My kids all ended up at various small private colleges depending on what they found as a fit for their interests and I got to see through them the difference between going to a school with 20,000 students (USC) vs one with 1,500 students (Reed). In hindsight I suspect I would have done better at a more intimate school (harder to have one's slacking off ignored by the administration right?)
No worries, that's definitely a reasonable impression since there are way more postgrads than undergrads. I just wanted to point out that undergrads are allowed to participate in the 'action'. But of course there is the occasional bad class (Peter Shor is not that good at teaching Shor's algorithm).
For what it's worth, the lab classes I attended at the public university in Texas I attended after one quarter at MIT was essentially... "here's the key to the lab, knock yourself out. come find me if you have questions." The only reason we didn't make an initiator in the lab that year is we couldn't find the polonium.
