Stop me if you've heard this one before.
You're sitting around the table playing in your weekly (bi-weekly, monthly, whatever) game, and you've just watched as an 80 year old skin and bones man, using only one hand, lifts a fully loaded wagon out of the mud and then casually holds it upright while the wheel is quickly replaced. Then, not five seconds later you're told something can't happen because "it's just not realistic", or using real world logic your idea is explained to not work.
It happens all the time in gaming, and it goes back to what I was talking about in the world design entry about internal consistency. That being said, what seems logical for you, might not be for others, so I figured I'd bring up a few things to talk about on the subject.
Again, Keep it Consistent
So, once again just to go over things, it is ok for your world to not make sense by real world standards, as long as it makes sense within its own framework. What this means though, is that explanations for how things work needs to remain consistent. Let's look at Legend of the Five Ring's world of Rokugan for a second.
In Rokugan, the world is made up of 5 magical, sentient, elements represented by various spirits called 'kami' (yes, lower case k is important here). A mixture of all the 5 elements (earth, air, fire, water, void) makes up everything. Including people. This is a known fact of the world, Shugenja can speak to these kami, and that is how magic happens and works. Lots of things in the world are explained using this system, the awakening of metals, how some people can take significantly stronger hits than others, why fire consumes whatever it can, even why steel acts like steel is explained by this.
However, in taking this system into account, you throw out a lot of real world restrictions in things. Rokugan doesn't work like our world, we have 118 elements in our world (that we know about right now or have made) Rokugan has 5. If anything, reactions between elements would work closer to how Aristotle thought the 5 elements functioned to make the world work than our own real experiences and what we know can happen in the real world. Especially when you get to more technical things, like the fact that in the real world you can make a stone canoe float, I doubt this can be done in Rokugan since the earth element's place is not floating atop water and stone is raw earth.
What I am essentially saying here is that you need to expand on ideas you have in the world beyond just what you are trying to explain with it. If magic works because the world is based on 5 elements and not 118, that's fine. However, you need to understand you have just given a fundamental fact about your world, it functions on 5 interwoven elements, which means it is incredibly different from the real world. Expanding and extrapolating on this to its logical conclusion will effect everything around it, including what is possible and what is not, and brings your world a sense of consistency that makes it seem like a living and breathing place.
Listen to Others
This one is a bit harder to do, especially if you're like me and don't work too well with other people on creative projects, but listen to other people. If you show a friend some of your design and he has impressions on things, listen to them. Everyone thinks differently, and a lot of times it's the ideas a friend gives you, or sparks with their own thoughts, that really helps bring the world together while maintaining its consistency. It may feel a little dorky to be discussing your fantasy world like a real thing, but treating it with that respect is really only going to help it to grow and develop further as a world.
This is especially important when running a game, as your players with a feel for the world will help you bring it more to life than you can on your own. Their interpretations can help you if you don't mind working with it. The end result? A significantly better game, provided you keep things consistent while listening.
Is it really so bad?
Finally, this comes more to the game master themself, when a player wants to do something that doesn't agree with your view of the world, but they seem to think is logical for it, is what they want to do really so bad as to deserve a 'no' or "that can't happen"? This goes back to what some view the primary rule of GMing is (never say no to a character, a player maybe, but not a character) but really is a bit more than that, or perhaps an adaptation of it.
Look at the idea given to you, and ask yourself why you are rejecting it? Is it fundamentally wrong against something in your world? Does it open the door for something that is mechanically broken? Is it some technical piece of information (from the real world) that is making you deny it? Or do you just not want to deal with it? If it's the real world reason, and you aren't in the real world, you have pretty much shot yourself int he foot sadly. We've already established that your world isn't the real world, so the real world rules don't apply. If it is mechanically broken, talk to the player about it, and maybe find a compromise or a way to restrict it to not make the game unbalanced for everyone.
For the other reasons though, as yourself 'is it really so bad?' if the player wants it, others don't mind, and seems like it could be fun or cool. Why not allow it? If it violates fundamentals of the world, then maybe those need to be clearer cut, but if it is only a small infraction things could possibly be changed around it to make it happen couldn't they? I mean, you built this world, surely you can adapt for something small like this to be possible right? Especially when it brings in fun, and probably is a cool idea.
Now, if you're still against it, that's fine and you should talk to the player, explain why the idea won't work. But make it make sense to the player in the confines of the world. "Horses can't speak ebonics because their mouth doesn't work that way in the real world" doesn't make sense if you've already established that horses can speak in your world, even if it is only in their own language. A bad example perhaps, but something to think on.
So I guess the conclusion is, keep things internally consistent, but don't forget to allow fun, especially if it can still remain internally consistent. Doing the two of them together helps bring your world to life, and you never know where a compromise you make may spark an idea that brings the next extremely cool thing into your world.