I liked the way 'The speaker in dreams' was structured. I didn't like the adventure itself very much, though. Something about the idea of a mindflayer creating a portal to summon devils struck me as inapproriate.It's really funny how different my experience is from everyone here. I started out with 3e. I think the first adventure I ever bought (outside of Dungeon mags and "The Red Mask of Red Death" Ravenloft box, thinking it was a D&D game) was "The Speaker Of Dreams". I really liked it; never ran it, but I still dig it a whole lot, and wish I could run it sometime.
But I've plundered the adventure's better encounters, restructured them, added a few of my own and turned it into a key adventure for my campaign.
Another 3E adventure that worked really well for me was 'Standing Stone'. I've translated it into my setting, changed the locations and redid every single monster and npc in the adventure but the kept the structure and order of scenes.
But, anyway, I like the flowchart-based approach for creating adventures from scenes. Almost all of the 2E Darksun adventures used that style, as well.
When I create an adventure from scratch I also start with a flowchart. I rarely stick to it, but it's still an excellent way to start.
I also seem to belong to the minority who likes the Delve style of the new adventures. It makes it pretty easy to steal neat encounters and huge set piece battles for my own adventures. I've rarely if ever run an adventure I bought exactly as written. I usually customize pretty much everything about them. I don't save much work that way, but it's worth it for me AND has the advantage that players that bought and read the adventure can still be surprised by events.
My next adventure will be a heavily customized version of 'Last Breath of Ashenport'. I've turned the underground area into a former mindflayer base that was overrun by aboleth minions in their ongoing power-struggles about who gets to control the region's inhabitants.