One of the bigger themes to be developed in recent years is the concept of saying "Yes" to your players. This topic, naturally enough, has stirred a fair amount of debate amongst the D&D crowd with many on both sides of the debate crying foul. So, at the risk of starting a flame war, while at the same time trying to help out, I'll try to tackle this issue, at least a little.
The Big Idea
The idea behind "Saying Yes", is that when a player asks "Can I . . . ?" your answer, as DM, should, more or less, be "Yes." That's simplifying it a bit of course, but I'll go into the complexity of it below. At any rate, the theory here is that the game will be a lot more fun for your group if you are saying "Yes" as opposed to saying "No." As a general rule, its a fairly good one in my opinion. When the player says "Can I jump off the balcony and grab the chandelier to swing to the other side?" the answer shouldn't be "No, there's no rules for that," but rather "Sure, you'll need to make and Acrobatics (or possibly Athletics) check." True, there's no real rule stating how to adjudicate that, but its better to let your player try, and then consult the DC difficulty table (as revised, and revised again :P).
The above situation though is not where the controversy stems from. Rather, it tends to be more of a worldbuilding or campaign building issue. The problem comes when the DM has a well formulated idea about what his world is like. He knows who the movers and shakers of the world are. He knows the theme of his world. Most importantly, he knows who the adventurers (not necessarily the PCs) are. As he has all of this set out, a player comes up and says "Hey, I want to play an ogre. I have this cool concept for an ogre shaman embarassed by his race's history of violence." The knee jerk reaction for the DM in this case is sometimes to say "No, an ogre would never be accepted in the civilized areas."
"Saying Yes" though suggests that the DM should find a way to make this work because the player will have more fun if he can play the character that he really wants to, as opposed to playing a character he has to. At the same time, this character threatens to challenge the DM's concept for his world. The result is that the DM might feel as though he is in a lose-lose situation.
The standard argument against "Say Yes" is Rule 0, i.e. the DM is always right, or rather, the DM is the final arbiter of the rules. While this is true to an extent, I do not believe this is sufficient. Admittedly, if the DM wants to create a world wherein all the PCs are Shardmind Cunning Bards, he can. That's his right. Of course, its also the right of his players to choose not to play in that game. Obviously this is an extreme example but you get the point. The DM is, in my opinion, perfectly within his right to say that the player cannot play an ogre. The question though is should he?
The Eberron Effect
Let me take a couple of Eberron examples. Personally, I love the Eberron world. I love a lot of the unique things that came with it like warforged and dragonmarks. However, I also feel that warforged and dragon marks are very much products of the Eberron world. The dragonmarks in particular are tied to a prophecy of Eberron. To me, these elements just don't lend themselves well to other worlds.
That being said, in my current homebrew game, we are not playing in Eberron. The world is much more the typical fantasy setting. No lightning rails, no air elemental powered airships, certainly no House Cannith creation forges. Well, needless to say, one of my players wanted to play a warforged. My initial reaction was to say no (in fact I had already posted a campaign document that basically said nothing specific to a particular campaign setting was allowed -- though technically warforged are no longer Eberron-specific. However, rather than immediately saying "No", I thought about how I could make it work. The answer I came up with was to turn warforged into fairly advanced golems. Golems gone wrong, or really right depending on your viewpoint. As a result, they were really rare, but not unheard of. Most people would assume that they were golems, and warforged were created by wizards attempting to create a golem, etc. Its not quite an "only one of its kind" thing, but it still kept true to the feel of the world.
Dragonmarks though were a stickier situation for me. Again, I really feel that Dragon marks belong in Eberron (at least with their flavor). When another player wanted to pick up a dragon mark, I struggled with the idea at first. However, I again found a happy medium. Turns out the player was really just looking for the mark to set up a later player option that required the mark as a pre-req. From a mechanics standpoint, I had no problem with marks either, just the flavor. So the solution was simply to create a feat that gave the player the benefits of the mark, without calling it a dragon mark. While the player didn't ask for an actual visual mark, I later realized that I could even have the feat show up as a tattoo if the player had really wanted to. So the mark was no longer tied to the prophecies, but it did give the player the mechanical options he was looking for.
Know when to say when
All of the above aside, I do still feel that there can come a time where the DM really should say no. Take an extreme example wherein the player wants to play Master Chief, the Space Marine with laser weapons, a com-link and a warthog, etc. etc. There's really no way to fit that into a typical fantasy world if the player is not willing to compromise on the flavor and mechanics, etc. Of course, I have also yet to have this request happen.
Other times though, there may be valid reasons to say no. I once ran a campaign where there were no drow. Part of this was to avoid any Drizzt clones, but the main reason was campaign background story. In this world, the elves had largely retreated and rarely left their forest home. The reason, as it would eventually come out in the campaign, was that during a war against the drow that spanned the entire world, the elves had gone somewhat battle-insane and literally killed every single drow (even the babies and children). Afterwards, they realized the horror of what they had done, and had thus retreated to avoid anything like that again. This back story would become a major plot element of the campaign and ultimately made for a great time. However, it did require both an investment in story and a payoff in the campaign in order to justify excluding drow as an option (in my opinion of course).
In sum, I do agree with the concept that if you can find any way in which to make the idea work, you should try to do so. Of course, this is a two way street and the player should also be willing to work with the DM to make it fit into your campaign world as well.