That can't be entirely right, can it? A medium's early adopters are really the most likely to have interesting things to say? How can the two possibly be correlated? I agree that there's an obvious pattern between signal:noise ratio and community size, but my gut tells me the reason is likely more than "the early adopters were more interesting."

I can't find the source now, but I recall reading an article that posed another explanation: communities start out small, and the members know the rules/mores/norms and abide them. When a newb steps out of line, the graybeard:newb ratio is small enough that it's not hard for the community's graybeards to reinforce the rules with the him. As the community grows, the definition of what's interesting to it gets more vague and distorted, like a drawing on a balloon that's continually inflated. And if it grows quickly, the graybeard:newb ratio drops. Both factors make it harder to enforce the interesting-ness norms.