The examples of games that made a comeback were No Man's Sky, a sandbox game missing features where, development-wise, it's very feasible to add in missing promised features; and Cyberpunk, a game with good bones that didn't function a lot of the time. Starfield's problems are deeper than that, at least from my perspective.

The tech tree and leveling system is "improve by doing", which runs into the same problems those systems always run into, which is why no one else does them anymore. It incentivizes me to get shot in combat on purpose so that I can improve my healing, and other stupid behaviors like that. So many of the quests are thoughtless fetch quests with nothing interesting along the way, and the game would actually be better with their omission than their inclusion. The endgame mechanic is an interesting one on paper, but seeing as the major quest lines only really play out one or two slightly different ways, there's not much that's interesting about going back to them, and you can also do all of them in a single playthrough, so there's no need to engage in the endgame mechanic to see it. These are some of the problems that can be fixed but will likely be so costly and time consuming when there are Elder Scrolls and Fallout games to be made that I doubt it'll ever happen.

The more fundamental flaws are that you can't spec your character to interact with the world in wildly different ways and get clever with its systems; the universe doesn't flow together the way that one of their terrestrial open worlds from before do, and fast travel is now mandatory; and the story walks right up to an interesting sci-fi story and stops just short of being good. To change these things sounds a lot like making an entirely different game.