Re: Running Sandbox Games
I have a personal dislike of games that are divided into reusable concurrent locations. Having a thread titled 'The Pig and Whistle' that includes every event at the local pub over the past five years is fine while you're playing along, but it's almost impossible for a new player to read and have any idea of what's going on, and it's equally difficult for archiving purposes or GM reference.
I try to start a new thread for each scene, with scenes titled something like 'Castle Bram, dusk, 13th January 1885'. If your game has five (or twenty five) scenes at Castle Bram, anyone needing to check back will be able to find the right scene by looking at the date and/or time.
This means you'll generate more threads long term, but you'll probably have fewer open threads than the 'reuseable' game types do. If someone wants to go somewhere, just open a new scene for the action.
If there are concurrent scenes in different locations, then time can become a problem with different posting rates. Often, these can be rectified by making a small time jump and starting a new scene a little ahead of all the others. This is easier to do if the scenes are fairly short, so you're only jumping a few hours. It's far more difficult to ask what Frodo and Bilbo were doing at Bag End while Gandalf was travelling to Minas Tirith...
Which highlights what others have said above. I make it clear to players that their characters should act together whenever possible. Yes, theoretically a sandbox means everyone can do as s/he pleases, but I tend to insist that they come to a consensus about what they intend to do and then do it together! Otherwise, the game falls apart. I don't have the time or inclination to run seven or so solo games.
Sometimes a split off is inevitable, and the story will dictate whether that should be hidden or open, but ideally such splits should be short and simple, bringing everyone back together as soon as practicable.
Abstract time? Hmm, maybe, if it's not obviously distorted or broken. If the splits are short enough, it shouldn't be too obvious or even necessary.
I'm not at all sure about action over description. I think this is a matter of taste. Some GMs and players consider description to be a waste of dice-rolling time, whereas for others interactive writing is their purpose for gaming.
If you have both player types in one game, you're going to have a hard time! It's probably something you should iron out before you start.
One last thing I would say: make sure you have an overarching plot. It's not enough to create a world and expect the players to come up with a story by interacting with it. If players wanted to create a plot, they'd be GMing their own game. They come to your game to take part in a story. They want to affect the story, and their actions will often create new story arcs along the way, but players have a tendency to stagnate if there is no obvious direction to move in. You need to give them pushes and pulls and to know where those pushes and pulls are likely to lead.
And yes, I've made every mistake I've highlighted, and will probably do so again, but the main thing is to have fun. Try something out, if you don't like it, either change it or do it differently in the next game. There's always a next game. :)