I'd like to talk about forking paths.
Although it's starting to fade, forking paths are still very popular in game design. Bioware is the biggest example of this, in that every RPG they've created in the past decade or so has been entirely based around the idea of "choosing between good and bad".
There are some problems with this kind of design, the big one being that when you give your players some choice, it changes how they view that element of the game. It goes from being a simple story to being a part of the gameplay. And bad gameplay, like Dragon's Lair or the quick time events that are killing otherwise interesting games these days.
I've already talked about the fact that these choices are typically only ONE choice repeated over and over and over. This is obviously not terribly freeing, but it does offer some advantages. The first is that it's relatively easy to program, since you're essentially programming two plots and letting the player pretend that it's an infinite number of plots. The second is that it gives the player the illusion of choice - every time he re-selects his chosen path, he feels a bit like he's actually choosing a path. So, psychologically, it has some benefits.
Personally, I feel that choices like these are actually bad. I wouldn't complain if the plot made my avatar good or evil and made him do all those plot choices, but when you give me choices, I always think "all these choices are so... inapplicable. None of them is what I want my avatar to do." Whereas if you never brought it up, I'd basically just say "my avatar's a wuss/a dick. Okay, let's get on with it."
It's obvious that many players don't feel that way, and I don't see any reason why game designers should be avoiding a psychological trick just because I don't personally like it. If that was the case, there would be no MMORPGs at all, because I hate those psychological tricks, too.
But if you're looking to design a game that features some level of actual choice, you have to consider how you're going to do it.
One way, and the way most would-be designers seem to immediately pick, is to simply make a wider variety of choices. Perhaps your first few choices are good/evil, but then they start to turn into things like honor vs life, for example.
There are some problems with this, such as the fact that it's still on rails and players like me will still fail to find a choice that really matches what we want to do. But more importantly, it takes a TON of scripting. Would-be designers who think in this manner are setting themselves up for the let-down of a lifetime when they realize the literally thousands of pages of scripting they'll need to do to support this. That's one of the fundamental problems with these kinds of scripted choices.
If we think about it in terms of first person shooters, we've got an easy way to recognize these things.
Rails shooters such as House of the Dead are quite fun. However, you do not control your motion, you only control the shooting. This is like a linear story RPG. This has some serious advantages: you're not only able to focus only on the parts of the level that the players will see, but you can also insure that the players will move and interact with the scenery in a fun and interesting way. For example, diving down a well.
Some rails shooters allow players to go alternate routes - route A or route B? This is like a Bioware RPG. It doesn't ACTUALLY matter which route you choose, because they're programmed to be roughly equivalent in terms of difficulty and they both lead to the same boss. But it gives the player a sense of control and is a useful psychological trick. Sure, you may think both choices are dumb, but at least you can choose the dumb.
Now, we can imagine such a thing being taken to an extreme. At every door the game pauses and asks whether you want to enter. At every corner, it stops and asks whether to turn right or left.
This doesn't actually give you a better game! It gives you a worse game!
It is much, much better to make the player control his position at an infinitely higher level of granularity. By which I mean letting him walk around and face whatever direction he wants.
He still gets to explore all the cool things you want him to explore, but he also gets to do it fluidly and in complete control at all times. There is no sense of being railroaded into bad choices (assuming a relatively competent designer).
It's true that building a game that allows the player to explore rather than simply putting him on rails is more difficult both in terms of mechanics and level design. However, once you start considering how many more rails you need to lay in order to give the player the "freedom" to go more places, you realize that the price of rails is exponential while the price of the walking mechanic is linear.
It's probably cheaper to lay two or three main rail lines than to create movement mechanics and carefully program out all the nooks and crannies of the level. But when you start talking about branching those rails out to hundreds of possible destinations... the price skyrockets and the railway map becomes hideous.
I hope this explains where I'm coming from.