I am hoping to do the #Dungeon23 challenge, originally proposed here by Sean McCoy. (Short version: build a megadungeon by generating one room a day throughout 2023.) I'm planning to use my approach to this challenge as an opportunity to develop a big asset for my homegrown system in progress, The Meanings. 

I immediately tacked toward emmy verte's Alternative Schedule for Dungeon 23, which alternates the production of rooms with the production of other important dungeon elements. You end up with "about 150 [rooms], but in turn you get 50 NPCs, 50 rumours, 50 items, and plenty of faction descriptions." emmy adds: "[f]or me this method brings the room count to a bit more manageable size, and adds a bit more depth to the dungeon ecology/environment." I agree, but their exact breakdown didn't feel like a perfect fit for me either for various reasons.

I took a peek at some of the other #Dungeon23 resources helpfully compiled by Andy Duvall, and took a special interest in GoblinPunch's "Dungeon Checklist." Hacking only slightly at these resources yielded my own alternative schedule, which I think will be the best fit for my needs. 

Each day, roll a d8:

1: Location (important area, key building, room, architectural feature, connecting corridor)
2: Something to be found (treasure item, equipment, piece of knowledge, clue)
3: Something the players probably won't find (same as above, only specify a more complex hiding place)
4: Someone to talk to (an NPC who begins with a neutral or friendly stance toward the PCs)
5: An enemy (monster, other antagonist, an NPC who begins with a hostile stance toward the PCs) 
6: Something else dangerous (a non-sentient element that can damage or kill the PCs)
7: Something to experiment with (interesting machine, inscrutable object, magical pool, strange altar, mystery to solve, rumor to substantiate, etc.)
8: Location (or, alternately, add 1-2 more sentences of description, detail, or lore to any previously-generated element)

Rather than commit the elements this system produces to a journal, I am writing them on index cards, which can easily be shuffled around and reconfigured. That will allow me to create things like hazards or NPCs without initially locking them in to a location. They can be tied specifically to locations later on (like, "yes, it makes sense to put the Fire Imp enemy in the Foundry location") or left free-floating (like, "yes, it makes sense that the Withered Guard enemy might patrol an interlinked set of areas rather than stay tethered to a specific location"). I will probably group these things into 12 levels/districts/whatever, as Sean McCoy proposed in his original post, but probably won't decide on the exact division until the year ends. 

I am hopeful that this system will work well not only for people doing #Dungeon23 megadungeons but also should accommodate people doing one of the variants on the challenge--#City23, or the science-fiction themed #Facility23. I will probably do one of those variants myself! The Meanings is a flexible enough that a megadungeon wouldn't be totally out of place, but I think one of the variants on the challenge is likely to produce an asset that will be of greater use to people trying my system.