Elections bring up the interesting question of the psychology of no-hope parties and candidates. When they choose to run in areas where there is a real contest between two other candidates with differing views, they generally risk harming whichever candidate has views most similar to their own.
Perhaps the most dramatic example ongoing is Ralph Nader’s campaign in Pennsylvania. It is difficult to understand why he would choose to run there, and perhaps even more difficult to understand why people would choose to vote for him. Doing so can only increase the probability of an outcome these people would probably find less preferable (it is difficult to imagine someone who prefers Nader to Obama finding McCain more attractive than Obama). It definitely challenges the notion that the aim of voting is to rationally advance whatever program of government you would most like to see implemented.
I suppose saying “don’t run serious candidates until you are popular enough to win” is a strategy that would forever prevent the emergence of new parties. In a sense, it is a bit like the strategy of not negotiating with kidnappers: it may contribute to a bad outcome in a specific instance, but establishes a more desirable long-term pattern.