Playing to Lose
One of the hardest things a GM has to do is lose. We could say, "Oh, the point of running a game is to make sure everyone as fun, so as long as everyone has fun, that's a victory!" But, I mean, come on. You spend hours, days, years, building a world, designing cities, cultivating NPCs and villains, and then you have to hand it over to the players and watch them CRUSH it. Your plots - ruined. Your dramatic moments - turned to comedy. Your favorite NPC - the laughing stock of the party. Being a GM means letting your players break your shit. That's why I suggest a Dr. Who approach to running.
Your players are the Doctor and his companion(s). They are going to bounce around all over the place and try to save people when it's the wrong thing to do, and do nothing when you need them to do something. They are going to stick their noses into things you didn't plan anything for and ignore the nicely put together plot that you spent all week prepping. To deal with this, you have to be flexible. Go with the flow! Make shit up as you go! If you are a planner, and you want things to go a certain way, then you need a party that is going to follow along - good luck. And some GMs try to control the party by adding a DMPC, meaning a player character that is played by the DM. Unfortunately, I do not believe this method is beneficial for the players, GM, or the game. No one is "fooled" by the presence of a nanny character. (The Gamers: Dorkness Rising perfectly captures this issue.) The players know why the DMPC is there, and they roll their eyes and get on the train so the GM can railroad them to the plot.
Blah. Do better. Be better.
The GM has to let the players go off book, and the best way to do this is to focus on fixed points in time (going back to the Dr. Who method). Instead of trying to get the players to go to town X, have them decide where to go, and wherever they end up, put in your plots for town X. The illusion of choice is as good as choice itself for the players. Have large events that are absolutely going to happen whether the players are there or not - then you can "script" the narrative.
As the players engage in adventures, let them alter outcomes. Save prince A and he will defend town B, but let prince A die and the players will have to defend town B. Either way, town B is going to be attacked, so the GM can plan that out and decide what happens if town B makes it and what happens if town B falls into enemy control. Having multiple endings to each encounter allows a GM to be more flexible! That flexibility will make the game more fun. And then the GM never wins or loses, because all that hard work and planning still comes into play.
Also, move things. If you want the characters to go into a cave at the bottom of a mountain and they choose to head to the coast, then move the cave to the coast. When you're GMing, you can move anything in the world. Think of it as setting up a chessboard by placing pieces that you want to be captured. Instead of defending your king, you are setting up the order in which you want your pieces to be captured leading up to the capture of the king. And whenever you want, you can just pick up a pawn and move it into range of the players.
Next time I think I'll write about consequences and rewards, but to touch on that note, playing to lose doesn't mean you should make it easy. Players and characters should work, suffer, and overcome. If the victory is handed to them, it has no value. On the other hand, if the players constantly lose and fail to achieve goals, the game is likely going to be unsatisfactory. At the end of the day, most people want to be amazing. As a GM, you get the opportunity to give that gift to your players.