I like the system overall, although that's not a big surprise since I'm a fan of strong, central character classes in D&D. Even in 3e, I never really saw the "problem" with multiclassed spellcasters having a significant knock in power. If you don't focus, why should you expect to be near the level of character who *do* focus? I actually disliked the fact that classes like rogue and fighter got relatively little at higher levels, so there was little reason to stick with them. This made other people happy, since you could multiclass these "easily", because you were giving up so little. I always wished that not gaining those high level fighter or barbarian levels was actually a somewhat painful trade-off, rather than "Meh, I'm not missing much and gaining a whole raftload of novel class abilities that stack great! Multiclassing martial classes is great! They don't get anything cool at high levels!" OK, that's an exaggeration, but this *is* the internet.
Given that the DMG is going to be apparently focused on how DMs can customize and make changes that suit their own campaigns, I'd guess that there will be a discussion of altering the multiclassing rules. Mearls has already noted that 4e could be changed into a classless game much more easily than previous editions and I think we can certainly see how. For those campaigns that don't mind quadruple classed character concepts and easy access to class features, I think that the house rules to break those barriers down should be pretty straightforward. Just allowing anyone taking a Class Training feat to select that classes' powers as a straight swap (will to will, encounter to encounter) doesn't seem too crazy. I do think it might lead to lots of poorly conceived multiclassed characters because it does seem too much to me.