It shouldn't take 2 to 3 hours to resolve that situation. You simply tell them the guards have arrested their friend.
Except the rogue player may claim you just railroaded him by not playing out his capture.
You're better off not rewarding the forker in the first place. leave him hanging the moment he starts heading away from the party.
I'm all for the DM moving things around, adding clues to keep the game moving forward. Forward being defined as the pursuit of the players goal with fair or reasonable obstacles and setbacks.
Railroading to me is when the DM not only assumes a certain path through the adventure, he thwarts any other attempt to choose differently, arbitrarily, rather than as an outcome of gameplay or reasonableness.
I don't think I like the idea of a wide-open sandbox. I don't want to have to write that much material. I also want the retelling of the game's event to make a decent story.
So, I plan my game with presenting the players with an opportunity or threat to their interests that I'm certain they'll pursue. If its a little weak, or iffy, I'll do mulitiple. I avoid "The king asks you to..." type plots. I generally make mine based directly on what the players want to do, or threats to what the PC cares about.
From there, as a DM, I assume the PCs will be successful, barring critical failure. Meaning I expect they'll confront the big challenge. But I don't know how they'll get there (though I can guess). So I'll document the kinds of things that are in the way of getting the end goal.
At that point, I suppose the game becomes a partial sandbox. Once the players opt to pursue the goal, I relay them information that lets them determine their tactical options, and they start trying stuff.
If they go the wrong direction to their chosen goal (presumably a misunderstanding of the facts), I'll either reveal it or re-arrange things to recycle content and still keep things interesting. Presumably, going the wrong way would have some setback effect, but once overcome, they'd be able to continue with their goal.
It is possible that through luck or mistake, the party could fail. Just because I assume they'll succeed, that only means i've figured out stuff that was between them and the goal. It doesn't literally mean they get a free pass. So in each encounter, I'm winging what happens next, and shifting my content for the repercussions if there are any.
What I think the OP's podcast point was, that in an extremely dull game with no compelling hooks, the party sits around the bar causing trouble. And thats it. By my definition of Railroad, Railroading isn't the solution.
The solution to a boring game is to present opportunities and threats the PCs/Players are interested in pursuing. And ideally, there should be more opportunities, than threats. Because threats make the player feel he has no choice (I have to stop the goblin raid or they'll wipe out my mining operation).
Once the players choose a goal (wipe out the goblins in the local dungeon, so we can expand our mining operations to there), the GM may still have to take a pro-active hand at getting the players to the end.
I think the key to that is recognizing what the players goal is, and presenting clues and information so they know which ways are viable, and which are dead-ends. Dead-ends being unintential wrong-turns to meeting the goal, not unconventional solutions to meeting the goal.