I'm going to attempt to refine my thinking out loud here, so if some of this turns contradictory, don't fret.

"Experiencing a well-engineered story is more fun than exploring an open-ended world." Point. The big problem with open-ended worlds, as Spector notes, is that players frequently have little conception of the background rules or of their motivations and goals. For this reason, almost all video games impose goals on the player rather than letting them choose their own, because "defeat the evil spirit Chaos who has corrupted the Four Elemental Orbs and brought darkness to the land" is more compelling than "wander around".

But what makes this story more compelling as a video game than as a novel is that the method for reaching the goal is left to the player. There may be one optimal path that the designers have built in, but it's by no means obvious, and the "holy crap" moments tend to come from discovering it. Example: if you've ever played chess, think of your first checkmate. Of all the paths through the game, you have discovered one that lets you reach the goal.

That thrill of discovery is also what makes more open-ended games fun -- you discover something new about the world. But we humans are naturally goal-directed creatures, so mere exploration is usually not enough to excite us. In the real world, our goals tend to emerge naturally; in virtual worlds, in the absence of the physical drives, we need to have them defined for us.

The ideal open-ended game, then, would be one with enough complexity that goals would emerge naturally without drowning the player in a sea of confusing options. This is not an easy balancing act, and games that get it wrong are boring, confusing, or both. Most games instead focus on diversifying the available paths, rather than the available goals. This means 'richly simulated environments' and lots of focus on 'emergent narrative'.

I agree that these games are very effective. I went back last night and put an extra hour into Deus Ex just to see one of the other plot threads, the one where my brother Paul doesn't get killed by the Men in Black. People do recognize and respond to stories, and that's what sells.

I just want to see game design go somewhere new and different, somewhere that other media can't go at all. And that means open-ended worlds. Someday, after the Singularity, we'll be able to design free-form systems that do have compelling narratives, that arise from the interaction of the player and the program instead of the designer's pen. Computer games have the potential to be works of art unparalleled in other media, and it seems at times that story (that is, pre-planned, designer-written story) is just holding them back.

More later.

One point I forgot to address: Spector said that, as long as we're going to have narrative in our games, let's make it good. I heartily agree. To use a movie analogy: too many games are Underworld, not enough are Mystic River. And what's with all these World War II games that keep coming out? I mean, haven't those five years of history been pretty much played out by now? It's almost as bad as the ice/earth/fire/wind/water levels...

Or you could take the game design approach of this guy and bring about the coming of the Messiah with a massively multiplayer game, a blimp, and a giant hologram. It's like the bizarre love child of Left Behind and Snow Crash.