Funny, I just voted on the troop increase on Vote.com. 30,000 people have voted and 77% approve the increase. If 30,000 people do not constitute a valid sample I don't know how many it takes.
Remember, we used these same arguments on this forum before the election, about how polls are false, etc. Look where that got us. In fact, the ABC-Washington Post poll was one of the most accurate generic predictors of the 2006 elections, prediciting a 6-point win in the generic for the Democrats in their final poll (the actual number was 6.5).