phenomenenal post, Melan!
Melan---
A hearty eye-opening "wowza!" for your phenomenal mapping analysis essay. Thank you!!

I put a link to this thread (and the original Quasqueston question about the Greyhawk Castle maps @
http://www.enworld.org/showthread.php?t=165693) over in the dungeon design 101 thread @ the Knights & Knaves site.
One of the assumptions that you make, and that only one or two of the first responders mentioned, is that in the more complex maps, the players themselves must be interested in being challenged by the dungeon environment (vs. the more-standard 3.x paradigm of challenge the PCs instead of the players). In order for that challenge to be effective---to fully leverage the less-linear, more freeform, discovery-laden map models---I believe that the players need to map as they explore: otherwise the big blank areas/hidden sublevels/etc. won't be revealed, and the more complex environment loses a lot of its intrinsic appeal. (Aside: it would be interesting to see how linear some of the more complex adventures and maps appear to be if the players failed to find any secret doors/sublevels/etc.: the "first glance" map vs. the "mapped the whole entire dungeon using Divination, Find the Path, Wand of Secret Door Detection, etc., etc." map).
As has also been discussed here recently, player mapping in D&D seems to be a lost art at best, and a very loathed experience at worst. The trends expressed in a few threads seem to be that mapping wastes too much playing time and/or is too much of a hassle to mess with, regardless whether the map is relatively linear or very complex and specifically designed to mess with mappers (like original Castle Greyhawk and El Raja Key maps were). Based on the mapping threads here, that opinion seems to cut across editions too: many older AD&D players said that they never mapped then and don't map now either. If the players aren't interested in mapping any kind of dungeon environment, is it even possible to employ more complex dungeon building techiniques successfully?
Taking mapping opinions into account, I'm curious to hear what people think about Melan's map creation ideas relative to the idea of needing to map them: not the actual act of mapping itself (which is not likely to reveal much useful to this thread), but whether or not building a more complex, freeform map does in fact necessitate mapping on the part of players in order to get the most out of the level.
(BTW Melan, the Greyhawk Castle maps just arrived on Thursday: I'd be curious to see your flow chart analysis of the top map image from the GH dungeon, or the big Judges Guild maps we linked to in Quasqueton's GC thread, or Wheggi's fabulous Quilt Dungeon map @
http://www.knights-n-knaves.com/phpbb/viewtopic.php?t=1153). I'm not sure that the really complex, larger maps would be very easy to flow, which is why I imagine you stuck to the 1 sheet maps from modules?).