But is this sort of sandbox the only alternative to railroading? As I indicated in my post above I don't think so, but I'm curious about what anyone else thinks.
"Railroading," IMXP, occurs when the PC's don't get to do what they want because the DM either (a) forces them to make a choice ("Stupidity/Failure/Success/Capture is the only option!") or (b) makes their choices irrelevant ("Oh. I see you chose to try and ally with the Demon King. He attacks you anyway."). The idea comes from a train hurtling down a track: you can't control where it goes, it just goes where there is already a plan for it to go. You are passive. This is bad, as a player, because it usually means the DM might as well go play with himself rather than have you at the table as his captive audience.
Your description and my thoughts for FFZ are pretty much on the same page, it seems: Big Choices. The PC's can do whatever they want, and the DM gets to react to it. It is relevant to the action. They get to decide what their characters would do, and what their characters would do matters.
Putting in good motivations (desires and fears) helps the DM get a handle on predicting the unpredictable to a certain extent, and helps them set up plots where they can present a handful of options and watch the PC's choose between them.
Making the PC's react at the beginning ("justify my set-up!") helps the players feel empowered, but still limits it to what the DM has prepared.
"Sandbox" is a good option for Exploration/Adventure type gameplay (the kind of stuff I love to do in D&D). But this kind of Points of Conflict design works really well in a strong narrative gameplay (the kind of thing FFZ does fabulously).