To me, it's not D&D unless there's a possibility of the princess ending up dead. Or issuing a vendetta against the party.
To each their own. But I've had too much experience with those games becoming really unfocused and boring:
Me: "Let me get this straight...the princess was dead because we didn't get there in time. Fine...so now the King wants us dead for failing in the mission to save her. So, we need to hide in this attic and avoid leaving because his guards might see us. We have no food and we'll be dead before they stop looking for us. This was fun. Shall we roll up new characters and start another game?"
DM: "Hey, don't look at me, I just let the dice fall as they may. You rolled low on your Spot check to see the kidnappers going into that building, so you didn't find them in time. Then you rolled low on that Diplomacy check when talking to the King. This was the only logical conclusion. What would you have me do, have the King forgive you even though you failed the roll?"
Me: "Yes, it would have been more fun."
Or at least, that tends to be the result of "letting things happen". Another time we spent nearly 2 sessions in a bar hitting on barmaids because one player REALLY wanted to get one to come up to his room and she didn't want to. But the DM didn't want to railroad us into leaving the bar...so we stayed there. Any action the DM took to change our minds would have been railroading(she already offered us 2 or 3 reasons to leave the bar, we rejected them all). It was great fun for the two players who wanted to be there. Not so much for the rest of us.
I use the existence of boxed text to show that railroading was ALWAYS an option that was intended. And that railroading has gotten a bad name over the years, but it is necessary to run a good game. I think extreme railroading is bad. I hate even using the word "railroading" except in the worst cases. I figure as long as I have at least 2 choices of what to do, it isn't railroading. Even if both choices lead to the same result.