Come on game industry. Let's get back to basics. A game just set in our solar system with colonised planets and moons would still be absolutely massive.
Kerbal Space Program. 
OK, it's not real scale (mainly for gameplay reasons), but can be made so with mods. What's interesting, it didn't start with a huge idea. It started as a very rudimentary physics sim that simply kept growing. Sure, it has its flaws and the graphics are medicore at best, but compared to how things look for other projects operating on such a scale, well...
Another thing to note that it has never involved any sort of multiplayer. Sure, the idea has been floating about, but it's always been a sideshow at most. So yeah. Start small, work your way up and stay away from grand multiplayer schemes (at least until the point you really don't have anything better to work on). To think of it, FSO (and many other open-source projects) also worked out that way. On the other hand, when ArmA3's devs tried focusing on multiplayer in the latest expansion, it was a complete disaster. Make of that what you will.
Scope is consistently a source of problems for pretty much anything these days. I don't know how long the actual development cycle was for NMS, but for a team of 12 people and what it appears they wanted to develop, they way over-scoped for whatever production time they had.
The last project I worked on, my senior game design project had a 6 month production cycle and 7 people all working part time on it. In retrospect given the uneven distribution of talent, the project was slightly over-scoped, though it had all the features (save for camera perspective switching) originally outlined, it's very rough around the edges.
Networking anything is a massive can of worms to open up if you're a small team and in hindsight, I think believing that a team of 12 would be able to deliver the scale of game they were promising as well as having Dark Souls esque multiplayer was optimistic, not necessarily completely foolish, just extremely optimistic.
