I've done a lot of this, mostly running but also recently playing as a DM adopted
My standard for running is a hybrid, of making a lot of paths available and then following on player interest. I have no idea where my campaigns will end up when I start, including what the end will be. My last campaign started with exploring a newly discovered continent and ended with a frontal assault from the moon on the regent forced on the child-empress by the council of nobles. Yeah, nothing I expected. The campaign before that started with no plan at all in place from me, and ended up with foiling a decades-long plan by the queen of the elves to start a war to use the blood and deaths to summon a horror to destroy a floating city that she believed was the phylactery of the lich lord of the undead, with other things along the way such as helping the high druid sacrifice herself to grow a new world tree to balance out the three worlds. It was a far ranging campaign, and the plots that got put in place were all because of player interest and character actions.
When I say hybrid, "full" player driven is that the players are proactive and the world reacts but in my campaign both those reactions as well as things the PCs didn't deal with may come up with situations they want to react to.
I create a lot of hooks that I don't use, and I improv a lot as players decide to do something novel or even something I predicted but not as likely and didn't flesh out, which is pretty common. My prep is often broad strokes of different things they might do, unless they have locked themselves into something. An example of locking themselves in was when they infiltrated the Imperial Catacombs full of undead ancestors who took blood oaths to support the line, I could prep a few sessions of a dungeon.
I did have feedback from that at the beginning of the campaign they weren't sure what to do, because I had given them lots of freedom and too many things of interest to do so they couldn't come to a consensus or judge what they were more interested in persuing. From my perspective it was just information so they could chose what to do, from their perspective it was a bunch of hooks they were looking at traditionally and they had problems evaluating which to do. Once they got a clear idea what was important to their characters though they found it much easier to direct and a better experience.
One group that I play with the GM ran basing on The Gamemaster's Book of Proactive Roleplaying, which recommends all PCs have three goals of varying lengths. But the group was very inconsistant about giving him these. He worked in character arcs for all of us based on backstory to help fill that void, which to me is also player-guided even if not the method from that book. We also had a bunch of "filler" sessions, because a lot of the goals that were stated were diplomacy/intrigue/investigation based, but when he would meta-ask players what they were interested in at the end of a session so he could prep, we often had players requesting combat if we hadn't seen any in the session, so we'd often get pulled (PCs reacting, not proactive) into situations and sidequests of combat. In the end, I think it was poor group goal setting, goals that didn't weave together with other character goals, and player meta-interest in combat that the GM didn't weave into our goals except as speed bumps that caused issues. Also there were places were we didn't have enough information to choose a path to pursue a goal, nor a path on how to get that information.
From seeing that side of it, it gave me a lot of perspective for how to run my next campaign.
