Yes certainly this is true.
But on the other hand we get a little too caught up with extreme variance. Most of the times large numbers are quite boring. Look at a roulette display, most of the time its just randomly distributed black and red numbers. Or the columns indicate Column 1 came up around 33%, column 2 another 33%, column 3 around 33% time (and the sample size on this is usually only 100).
Pollsters of course sample only hundreds of people at most to estimate the opinions of millions, and they do a reasonably good job at predicting elections (rarely are they off by, lets say 30%).
Heck if we take concerns about extreme variance too far, we would have to say Babe Ruth was not a great hitter. Lets say he had 500 ABs for 12 years (
only 6000 ABs) and he hit about 10% higher than an average baseball player (just rough numbers). Certainly well within variance! We as APs should form a mob carrying torches and pitchforks down to Cooperstown and demand they remove Babe Ruth from the Hall of Fame, because he hit well within statistical variance! He could have just been lucky!
Guess what I'm saying is we ain't that special... usually