Since I have been playing .hack//G.U. Volume 3 a bit lately, I have been thinking a bit about the idea of games which come out in a series of volumes, each of which is purchased as an individual game. Right now it is an unusual scheme that has its pitfalls, but it has a few clear advantages. Its largest advantage is that it allows a company to create a long game with high production values, even if the game would not otherwise be able to sell enough copies to justify the large cost. These days, rising production costs have worked to make games a lot shorter than they once were, and long games with an epic story are becomming less common, so the movement to multi-volume games does make sense. Still, it is rather difficult to make a multi-volume game that makes the best advantage of the structure.
Because I have played them, I am going to use the original ,hack games, the Xenosaga trilogy, and the .hack//G.U. trilogy as my examples. From what I hear, recent episodic PC games like Half-life Episodes 1 and 2 might also work as examples, but I don't know enough about those games in particular to commenton them.
As far as I am aware, the four volumes of .hack were the first games to use this set-up. All four games use the same save data, have the same engine, and all tell different chapters of the same plot. As such, it really feels like one game that has been split into four parts. In terms of gameplay, this works well, because this removes any learning curve for the later volumes, and makes the games all feel more like a coherent whole, but at the same time squanders some potential to make use of the volume structure. A larger problem is that the story and gameplay were not really well suited to the legnth and pacing of four volumes. The fist volume works well, providing needed exposition and setting up the premise of the story, and ends at a great place: the first battle against the main enemy, and a dramatic sequence which sets up the goal for the rest of the four games. Unfortunatly, the next three volumes do not distinguish themsleves from each other very well, and the whole experience ends up being overly repetative, with too little plot and character development for such a long story. Another significant problem is that the game does not have the variety of scenery, good visuals, and intricate story that would justify buying it in four volumes. The whole thing feels like it would fit on a single game disc quite easily, so it can be hard to justify the purchase.
The successor games to .hack, the three volumes of .hack//G.U., vastly surpass the original games in every way. While there are a few storytelling issues here and there, the .hack//G.U. games tell a much more complicated and interesting story that has enough variations and character development to flesh out a long game. The story works better as a set of volumes as well, because each volume focuses on particular charcters and subplots, and each volume ends with a dramatic sequence that wraps up many subplots and has a revelation that triggers a clear shift in the story and the characters' goals. Also, the beginning of each volume introduces new plot elements and complications, which works quite well. The game keeps the same system between each volume, but adds minor changes and improvements, so there are new things to try and experience in each volume. Finally, the game has fairly high production values, which helps justify the purchase. As a whole, it makes much better use of the multi-volume structure than its predeccesor.
I think I need to write about .hack//Liminality and Online Jack at some later time.
I can probably write a whole essay on the mistkaes and lost potential of the Xenosaga games, but certainy one of the biggest problems with the story of the games is the flawed divisions between the different volumes. The story of the first volume is left incomplete, the second volume only serves as the second half of the first volume, and the third volume completely skips a large section of the story, and wraps up too many plot threads from the first two games too quickly, while ignoring others. In terms of both gameplay and graphics, each volume is highly inconsistent, which makes the experience somewhat jarring. The fact that you can't directly carry save data between volumes because of the gameplay changes is a very severe problem. If nothing else, this series is proof that it is nearly impossible to use later volumes of a game series to fix problems with previous volumes.
As a whole, there are a few things I can conclude about multi-volume games. First, the story of each game needs to both tell part of the greater story, and stand on its own as a story-arc. Transitions between volumes should be memorable and distinct, and involve major changes in the story. Also, having a lot of consistency in graphics and gameplay is very important, but it is also important to keep adding minor new things for each volume so the player does not feel like he is just playing the same game over again. In that regard, .hack//G.U. serves as the best example with the way the main character's abilities are expanded once each volume. Also, it seems that having more than three long volumes might drag things out a bit, so it might be best to stick to just three (though this is probably very different for games in which each volume is very short). Finally, it is probably best to use the same game engine between each volume, to keep the expereince consistent and avoid the costs of making each volume be a full project.
Other than those conclusions, I am a bit uncertain about two important details: I can't decide how introducing new characters should work across multiple volumes and I am not sure whther it should be possible to miss something in one volume and lose the chance to get that thing in a later volume. The first comes up because I disliked waiting until the third volume to recruit some characters in.hack//G.U., but having the dsame cast across all three volumes of Xenosaga became too boring. The second problem is because both alternatives have benefits and drawbacks. Losing the opportunity to get something is painful (I missed a lot in .hack//G.U. becuase I didn't do a few things in the first volume), but it rewards the players who went through previous volumes, which adds to the experience. I suppose some compromise needs to be made for these issues.