During the course of play, I discovered that actually planning plot and organization relationships outside of session time was, more often than not, a waste of time. I discovered that players wouldn't pick up plot hints if they were too subtle, and would feel like they were being led around by the nose if the hints were highly visible. Character motivations often led to the PCs behaving in a manner opposite the planned storyline, and they subtly flipped back and forth between good and evil. (Sure, they'll save the community from the evil bad boss. But if a mook in a group they were fighting tries to flee, he'd be killed while he was running away. Prisoners were for interrogation or to be sent back as messengers.) Worst of all, I discovered my players were smarter than I was; I'd think I had all of a character's motivations and connections planned out, and somebody would say at some point, "Wait. That doesn't make sense. If that's the case, then why didn't x do y?" And they'd have a very good point, and I didn't always have an answer in mind.
Eventually, I stopped planning outside the sessions. Heck, for the most part, I stopped planning altogether; The world became whatever the players saw in it. I might listen to the players talk to each other about the NPCs they'd encountered and their suspicions of NPC and in-game organizational activities, and I'd take their concepts, apply them, twist them or subvert them, and have pieces of them slowly appear as the characters investigated further. Or I might listen to a player mutter to themselves about an NPC or scenario, and take a piece of what they're saying and use it. Or I might just watch the characters play out their current situation and think, "Hey, it'd be pretty cool if that [otherwise mundane element] was important," and give it a little extra flavor text; The players' subsequent debate over the element would paint a rough picture of what I could do with it. Within the first couple months of that campaign, I'd largely exhausted the resources in my binder and was laying plot, NPCs and encounters out on an as-needed basis in-session. And it was a lot of fun.
My current campaign, started six weeks ago (and running on an alternating-weekend schedule), began with a puzzle dungeon. I had a vague idea of some of the puzzles I wanted to use in it, including some flavor and concepts for the entrance. The entrance had a rune on it that I had intended as a non-lingual hint to how to solve the first puzzle they would face once they got in the door. Instead, one of the characters saw it, and cast a very expensive comprehend languages spell to understand its meaning. Well, crud; If I told him what it might mean in its original language, it would completely throw their attention in the wrong direction for the puzzle it was supposed to help with. I wanted to get them to use their brains, not frustrate them to boredom and tears.
So I threw out the planned puzzles, gave the rune a meaning, and applied it to opening the front door (Which, while entirely non-magical, and entirely non-mechanical, did require them to think before they got it open. Though one of the PCs nearly got himself crushed in the process.). As for the rest of the puzzles, I made them up on the fly, using the PC's comprehend languages to give one-word clues to how to solve them. And it significantly added to the dungeon, because I was able to tie together the religious themes of the PCs and the dungeon's background, while using the runes to give an atmosphere of religious test.
With these experiences, I've found I prefer this way of creating a campaign; The world and plot are almost as much a mystery to me as to my players, but it's continually revealed to be a rich one built from the fantasies of my players.