In the case of winning ballgames, there is an increasing marginal utility. Ten games from 85 to 95 wins are a lot more important than ten games from 42 to 52 wins. Therefore, a player must show, demonstrably, that his last-place team would have been, by far, the worst team in the history of baseball but is instead only a sorry team that will soon ship its superstar to a contender for minor leaguers. This, I don't think, can happen since a team can win 40-60 games just by showing up every day. So, really, I don't think there's ever a case where a superstar is more valuable to his sucky team than the guy who got his team over the playoff hump. It could be argued that Chone Figgins was more valuable to the Angels than Randy Johnson was to the D-Backs this year. Without Figgins filling in for all the injuries, the Angels might not have made the playoffs. Without Johnson, the D-Backs would have...um...finished last. Shoot. Value.
The problem with this argument is that you've just defined value as a team dependent metric. By this definition, anything a great player on a lousy team does is, essentially, worth nothing.
MVP, in my opinion, ought to be defined solely by an individual's contribution, not by his teammates, efforts. It should NEVER be rational to argue that Chone Figgins is more valuable than RJ simply because Chone's teammates played so well, that his marginal contribution put the team over the hump.
That's not value, that's luck.