Everything that happens, should happen for a reason. Every decision can be used to say something, to make your players feel something, to encourage them to do something. Part of why I love roleplaying games so much is that everything is completely mailable, and can happen anyway people want it to. These games are no more than a shared hallucination, a reality built upon conversations and understandings, the archetypal social construction. By changing how the stories we create are told, by adjusting the rules and tone of the systems we use, we can adjust the stories in whatever way we want. With this in mind we can look at everything about our games and ask how we should change them to better suit our goals. Now I'm going to talk about some of the things I think I have learned by looking at game design through this lens, and maybe some of my words can help you.
The first thing I learned was that its easy to just use the rules which are already in the book. The less questions you are forced to ask about the basic and not so basic assumptions which a system makes the less work you have to do. Its much easier to assume that all the decisions which the system's designer made are the right ones for you and to just take them, but this isn't always the case. HP in D&D assumes you want heroes who can take risks and get hit without permanent consequences, the skill system in WoD assumes that each skill and stat will be equally useful in your game, and Champions assumes that you are a robot who derives pleasure from balancing your hero's checkbook and meticulously simulating 12 second of combat over the course of 8 hours. For many groups and games these assumptions may be correct, but its worth taking the time to stop and think if the game which is presented in the rules is really what you want to play. For me personally, the answer "because thats what the book says" is never a sufficient reason to keep a rule the way it is.
That being said, its also easy to not realize the value in a group's continuity of rules. For the most part, everyone will assume that things are the way they were before or how they have always been, and changing these assumptions can come as a shock. If people have been playing with the same understanding of elves in their games for decades then suddenly changing them will come as a surprise which may work against your goals for the game. It may be worth it to not change rules because the shift itself would be too difficult and take away from what one is really trying to convey in their game. A game about political intrigue is not the right situation to introduce new duel wielding rules, even if they are way better than the old ones. In the same vein, changing something can be used as a tool to create confusion, alienation, or surprise, regardless of what the change actually is. Violations of continuity can be used to break up monotony or show something which couldn't be shown without a explicit change.
In the long run, however, I think my most important realization is that it is easy to make decisions for the wrong reasons. This is the main reason why I believe knowing what you want to achieve with your design is so important. Once you have guiding principles you can look at every decision and have a metric to judge what the right and wrong answers are. There is nothing better or worse about playing a game which is fun because of its evocative storytelling or engaging mechanics, but the decisions which need to be made for each of them are drastically different.
For me, the only thing more frustrating than finding a flaw in a design is finding a flaw which I have been making over and over again. Right now that flaw is failing to have solid goals and failing to make decisions which match up with those goals. In the recent Donjon game at THE CABIN, I failed to identify that one of my goals was to have fun by providing a sandbox where people could come up with their own interesting and wacky solutions to the problems presented. Because I didn't realize this I made decisions which failed to allow a diversity of solutions. In my Colors game I didn't think about how my desire for unique, special, and long lasting characters clashed with the DCC mortality system which ended up causing a really unsatisfying encounter the first time someone went to 0 HP. I'm sure that I will make some mistake along these lines again in my upcoming Hell game, but hopefully by explicitly thinking about and laying out these ideas I can avoid some of my mistakes, and help others to do the same.
