I started the campaign with the second group.
Despite being more people I hadn't known previously, they were also really nice, and also had really nice character ideas. We had more tiny machiavellian mastermind pcs, and more stabby stabby PCs, and more giant nerd PCs, and was generally fun :)
The session benefited from all the hooks I'd been strewing about in keeping up with group 1, and in bringing in the backstories from group 2 into the world, so other than the immediate goal of "delve for treasure", they had some reasonable suggestions for the them to actively choose between, following the bounties issued for the fungus creatures who invaded the mines, or following up more clues to a lost civilisation suggested by things group 1 found, or several other possible hooks.
And I managed to bring their characters together, first them, and then a couple of recurring NPCs, to have a chance to chat and get familiar with their characters, so stuff started off with more momentum than group 1 had.
Over-prep
But I'm still suffering from ending up with much too much prep. Partly because I'm figuring out the system, not the rules, but the way I want to run the campaign, so lots of notes about "this PC ties into this detail" etc, get muddled together, until I found a sensible way of organising everything. And partly because I just get over-invested in doing all the cool things I've thought of.
I *hope* that this does in fact help produce a cool campaign. I think it will, but now the PCs have caught up where I hoped in terms of interacting with the world, I need to step back and relax a bit more, and let stuff I've already thought of play out without constantly rejigging things.
The Labyrinth concept
I still really love the idea of the "undeground labyrinth", and after some false starts I really like the exploring mechanic I have for having the PCs able to explore unknown parts, map locations and follow routes, without physically mapping an actual world-wide labyrinth, but feeling like it varies from "reasonably straightforward" in the routes they're familiar with, to "dangerous" exploring unknown areas where you might legitimately get lost.
And I love that it lets me mix together two different groups in the same world, and rely on its magical weirdness to cover over "hey, wouldn't we have been here BEFORE group 2" moments, if things get out of sync.
And I love a lot of the worldbuilding I've come up with, and the players reactions when they uncover it.
But I think, it hasn't worked as well as I'd hoped as an old-school dungeon delve. Partly because the character and worldbuilding stuff was just more grabbing so that got a lot more focus, and partly because it's hard to balance encounters for a good learning experience where the PCs get to know what they can handle and not, when you don't want to hand-wave away accidental deaths. I think that will come more into focus as we play a little more, it's just that third level characters are already quite complicated!
But I think, I am much more excited by games with lots of npcs, plots, worldbuilding, etc, so if I do another big setting it will probably be a city.
I would still enjoy a small proportion of mechanics focused games -- basically the feeling of excitement when everything goes right and you win a fight. But I think they need to be structured slightly differently. You need more, "you fight a goblin, you fight some more goblins, you cross a chasm, you fight a goblin on a bridge across a chasm", where you experience elements which are easy individually but interesting in combination. That was how this campaign was SUPPOSED to work, but we ended up without enough of the easy stuff.
And the idea that "I wouldn't have to do much prep" hasn't been true, as I've been sucked into expanding... almost everything :) I hope it's a lot quieter from now on, now I'm familiar with all the characters. The plan was, most of the planning would be done once, and each session would be PCs just thrown into it and told "here's a top three hooks, go and explore", where everything already tied in interesting ways. But I ended up doing a lot to tie ongoing PC plots into events, and what I'd originally envisaged as a fairly slow growing background plot got more so, so keeping it all straight might have been easier if I'd just planned each session separately as "what are you doing this session" and not tried to maintain all the worldbuilding up to date so much.
But I am really enjoying it, I hope the games with these groups go well, and I hope I do run more variety of games in future.