And the natural question is, what modules do I think are better models?
B2 is brilliant because it presents a sandbox where every choice is possibly lethal. Simply fighting it out is rarely the right choice; PCs need to learn to explore, negotiate, trick and improvise. It's just big enough for the neophyte referee to not be overwhelmed, while giving the party truly free rein. There are several potential plots, or new ones can emerge through gameplay, but the module would work if you didn't pursue any of them, or added new plot lines in that were totally irrelevant to the existing rumor table.
BTAM is dark, and disturbing, and has nasty stuff about sex. It also has great ideas, like an infinite repeating tower and several other interesting mini-dungeons. There are really only two that lead directly to the climax, but getting there is neither assumed nor a very good idea, to be honest. Instead it offers a number of very inventive locations to explore before the PCs should get out of
The weakness of B1 is that the room descriptions are really long and overly mundane, and that it's not given over to randomness.
What's really interesting is that Dungeon of the Unknown is much more of a dungeon creation kit with a filled-in sketch in the back. I think that approach is something more modules should follow.
What I love about Tomb of Horrors is how few monsters it has. It's one of the deadliest modules ever, but aside from a gargoyle and a demilich there just isn't much to fight. And that's brilliant. As I've said before, LotFP modules read as if James Raggi took this as an artistic manifesto.
Each of these is something I'd love to see more of in modules: the sandbox aspect of B2, the time pressure in BTAM, the modularity of B1, the kit format of DotU, and the non-monster threats of S1. Even those of us doing OSR modules could stand to go back through these and pick up a few of their ideas.