njharman asked how to avoid railroading. For the purpose of this post, I assume reader does not want to railroad and has a good reason for it, such as not enjoying railroading or wanting to try new things. (Bad reason would be because some internet person told that railroading is inherently evil.) Any examples will be drawn from D&D because njharman used DM as a phrase, which kinda implies D&D or similar. I further a prep-heavy game (that is, not Wushu or similar).
Railroading can be avoided, and can happen, on several levels. The easiest and lest painful change is to change the macro level. The simplest method is to always prepare at least two adventures and let players essentially select which they pick. After one has been finished, the other should be altered as appropriate due to PCs ignoring it. The undead gains more minions or the orcish horde sacks more towns or the rakshasa infiltrates a position of power. Even if the adventures are on rails, players still have some choices: To engage this adventure or the other one, with the neglected situation often growing worse. Or maybe some other adventurers solve the other case. Something concrete that the players will notice.
Slightly less simple method is to ask players what their characters will next pursue and to build the next adventure around this. Pacing is important: The direction that PCs take should be clear at the start of each adventure, otherwise boring play and all symptons of that may start manifesting. Big twists and reveals should happen near the end of the session and the next adventure at the very end, preferably roleplayed to not make it boring, if people are in the mood for that. Email or simple face-to-face meetings between games may be an option based on the social environment of participants.
The key in both of the above methods is to clearly communicate with players that their choice matters. This may be out-of-game or in-game, whichever works more smoothly and reliable. Another important factor is to always start each game on the run. Either the plot hooks hit them on the head (more or less literally) by someone asking for help, some PC hearing rumours, the party being attacked, et cetera or the preselected adventure gets rolling right away, e.g. “You are in the sewers and have discovered the body. Who has the light source?”.
One way to get rid of the rails is sandbox gaming. It takes a heavy up-fron investment, though, and is not trivial to get correct. Do this only if you like building or memorising settings and have the time for it.
Not surprisingly, the method is to create a dynamic setting and then let players create characters and do as they will within the confines of the setting. The tricky part is “dynamic”. The setting must have interesting things going on on the scope that players can concretely affect. If they start with power and prestige, politics and wars are good default solutions. If they start as random farmboys, which is probably the better way to introduce elaborate settings for long-term play, slowly giving them power and prestige opens up many possibilities. The “slowly” part is to avoid player freesing at the terror of having the negotiate politics or an unknown setting with a fairly unfamiliar character.
There are some common pitfalls hidden sandboxes. First is players who don’t get interested. It is probably a good idea to start with a bang to avoid this. Burning farms is a long-time tradition among orcs and evil empires. Second is falling in love with the setting. Good gaming is the point. Setting is at best secondary. Players probably are not that interested in random setting bits (but if someone is, let him build parts of it or get involved in other ways). Also: Player characters are the most unreliable portion of the setting. They will blow it to pieces and reassemble them. Or not. Allow this to happen. Maybe they dethroned the emperor. Good for them. Play on. The setting is not sacred. Third is to make a totally impenetrable or alien setting. Resist the temptation. Players will start interacting with the setting only after they understand how it works, generally speaking. As clash bowley did when designing Book of Jalan, steal liberally from real world cultures, but feel free to mix and match cultural tropes, religion, environment, etc. Monotheistic roman dwarves living in jungle or something. It will look and feel exotic but also familiar, which is exactly the point.
The method of gamemastering in a sandbox is to have the aforementioned dynamic forces, player characters who are involved with them, and the rest of the setting for context and ideas. When preparing, think about the motivation and goals of the dynamic parts. What will they do to accomplish them? What do they want player characters to do? What will they do to PCs? How will they be involved? There’s the adventure. Some fairly static but interesting locations and events are good to keep the game changing. A dragon sleeping in the cavern, some random ruins there, an enchanted island here. Just don’t assume that players will go after them. Restricting PC travel is useful, because one can always spring random encounters (that hopefully do have a point or serve a purpose besides depleting hit points) on hapless travelers.
More on the next post. These are the easiest to adopt from railroading background, I’d say. Also: remember to inform the players that you won’t railroad as much anymore. You can’t change a group’s style by yourself.