However, when it comes to a socially complex game where apposing factions have entirely their own motivations, then taking the time to work out what exactly each are capable of and who the major personalities are, what they can do, is probably a good idea. The rest you can improvise.
In that sort of game, it's a lot harder to improvise the intricate hierarchy and interplay between half a dozen noble families, without at first having a foundation in which to work from, than it is to improvise a 1 hour encounter with monsters in the cellar dungeon.
And you're still better off going with the MOAR because the players will completely ignore the social complexities and go off in the direction of SQUIRREL!!!
Yes, I'm still bitter about all the work I put into coming up with a complex web of political alliances and opponents and agendas for the last Vampire campaign I ran... only to have the PCs utterly ignore it all in favor of trying to become YouTube pop stars and using Tinder as a meal delivery app. **headdesk headdesk headdesk**
Things ran so much better in my Mage campaign where I stopped doing prep work years ago because I learned that there is no way in a sandbox to predict which bit of flavor is going to send the PCs off on a tangent... in one case a book that was intended to be an Easter Egg referencing a past campaign turned into the players dropping the entire line of investigation they were exploring and running off from campaign central to Mexico City, then the Middle East, then Vienna and Prague... all of it on the spur of the moment as they chased one random world detail they interpreted as a clue after another.
Over the decades, I've gotten VERY good at improvisation. So much so that when I gave advice to one of my players looking to GM something that they should start really small and use the MOAR approach to running games they literally told me "I know you don't do it that way because I've seen the adventures you run."
Now, I keep copious notes on what has happened IN the campaign so whatever I set up remains consistent, but I actually let that player take a peak inside my game binder where all that was there was notes of the previous adventures, a list of names (because I'll invariably need to come up with a speaking-role NPC on the fly) and blank sheets of paper for taking new notes.
Not prepping beyond my immediate needs is the only thing that has allowed me to not just give up the GM screen permanently... and EVERY time I forget that lesson and think, "this time it'll be different," things like my current Vampire campaign happen and I'm left disheartened about even running that game because all that hard work I put into creating an intricate setting would have been better served watching YouTube videos on Boston's history and nightlife. All my work is NEVER going to be seen by anyone but me because the players just don't care.
So my honest to God advice to the newbie GM to not end up completely disheartened by all your prep work going to waste is... take an Improv class or two and learn deductive reasoning as it relates to cause and effect (as in, the effect of players doing X will be Y). Both will serve you MUCH better than any "here's what you need to prep in your world before running" advice ever will.