This is a fascinating post, in part because I wrestled with some of these same issues when writing "The Whispering Cairn," the Age of Worms Adventure Path kick-off module from Dungeon #124. The Cairn is a much smaller dungeon than many of the ones analyzed in Melan's thought-provoking post, but in order to make it more than just a simple dungeon crawl, I put a lockout mechanism in the dungeon and forced the players to leave the cairn midway through and go on a seemingly unrelated mini-quest. I'm very curious what Melan thinks of that type of encounter setup, and how he might incorporate it into his visual model.
One of the most important points in this thread, I think, is that there is not "one true way" of designing a fun adventure. Melan, for example, hates the moathouse section of "Village of Hommlet," but it is one of my favorite dungeons of all time, both as a player and as a DM (I have run it, one one form or another four times).*
Not everyone comes to D&D for the same reason. In my own Age of Worms campaign I have players who love roleplaying and players who seem utterly bored with it. Others are less enamored with fighting and would prefer to spend the whole session shopping. As a member of the RPGA, I played at nearly 500 tables with literally thousands of D&D players. You would be surprised at the range of interests and stylistic preferences out there in D&D land. Some people love mapping, some love puzzles. Some gamers enjoy anachronistic puns, and some cringe the slightest sign of silliness.
Accordingly, players and Dungeon Masters have a wide range of tastes when it comes to dungeon complexity and "linearity." As some have mentioned in this thread, a wide-open approach means planning for lots of encounters the players might not find. With some groups, you can change "might not" to "probably won't." That's fine for a published adventure, because it's not you doing the work. But a lot of DMs, including (I'd wager) some who have posted to this thread, prefer to design things themselves. Creating whole wings of dungeons hidden by secret doors or off the main path to the treasure room is an investment of time they might not be able to afford.
So some context is in order.
I actually agree with you, to a point, that the dungeons that look more complex on your charts are most likely to be more interesting from a mapping perspective, but it cannot be said loudly enough that an interesting map is just the first of many steps involved in creating a great adventure. Monster choice is certainly important, and for my money a certain degree of "plot" is also of paramount importance. There has been a great deal of development for the good in the years since 1974, and it is best to incorporate some of that into new designs rather than conforming to a rigid orthodoxy.
The visual model Melan has developed is useful, and I have really enjoyed reading this thread. I think it might also be worth looking at the models in terms of value. Does a more complex model suggest a larger investment in play time? Does that therefore result in an adventure that provides more bang for the buck? Melan suggested that much of "The Village of Hommlet" was wasted by a boring** village background. Had it been 95% dungeon would it have been a "better" adventure? A better value?
Very thought provoking.
--Erik
* As an aside, Melan, how dare you call the moathouse dungeon crayfish a more interesting encounter than Lareth the Beautiful? The cleric's staff of striking alone is enough to fell one character in a single round. The encounter has always been very fun when I've run it.
** The town itself is no more boring than places like Orlane or Restenford, but the map is absolutely great as a utility player when you need a village on the fly. Were the adventure redesigned today, I'd expect a lot more development of the village to make it a truly useful backdrop, but all of a sudden we're talking about substantially more than the module's original 24 pages.