Let's be honest here, the main goal was undoubtedly to drive sales. While part of me resents this overt manipulation, I've also heard that the total annual sales on all RPG products is less than the total sales for live opera--so I can't really blame them for trying to boost income. Game companies justified this tactic, claiming it provided both the GM and the players with a dynamic world that they could play within.
As I understand it, some feel that metaplots reduce PCs to secondary characters. They feel the main story gets resolved off-scene, and the characters end up playing Wedge Antilles to the metaplot's Luke Skywalker. Others complain that the characters must be railroaded for the story to be resolved as planned. Critics then argue that metaplots seriously harm the creation and maintenance of story in a roleplaying game (please correct me if I have any of this wrong). This argument, however, seems odd to me.
Here's the problem. Metaplots exist in other medias. They are often very successful. One of the most obvious examples is historical fiction. Take "Saving Private Ryan" or "Titanic". In both cases, we know how the meta-plot will end. The allies win; the boat sinks. This, however, does not matter. The metaplot acts as a frame--the stories themselves explore issues within that frame.
Sure, Captain Miller doesn't put a bullet in Hitler's brain, but that doesn't make him any less of a hero. It doesn't take away the meaning or emotional impact of his decisions or his death. Private Ryan is not the story of WWII, it is the story of Miller and his men--their actions, reactions and choices against an unquestionably horrible backdrop.
Admittedly, there is one difference between historical fiction and serial metaplots--the players and GM may not know how the metaplot will develop as they go along. However, I don't see this as an important difference. Whether you know the metaplot's story arc, or whether you expose it a bite at a time, its role in the story should not change. You're not telling the story of the metaplot, you're telling the character's story within the metaplot frame. The players have their own conflicts, their own goals, their own hopes and fears. These conflicts should evolve as they rub up against the metaplot, just as the conflicts would evolve from any interactions with NPCs or other PCs. But the conflicts remain, and the players still need to resolve them. In an ideal world, the metaplot deepens these conflicts by throwing additional complications in their paths. Still, the character's story--just like Captain Miller's story--must focus on the personal decisions and their results.
Arguably, some modern story-focused games (e.g. Polaris) are revolve around a single metaplot. From the reviews I've read, knowing how the story will end actually heightens the emotional content of the journey. Sure, there may have been problems with the designs of particular metalplots, or with the implementation of a metaplot in a particular game. But, clearly, we can use metaplots to improve the story content of an RPG.