There is no reason to believe that the deviations from the curve for each draft pick reflects anything more than randomness. Look at the "upside surprises" in the data for picks 11, 20 and 47. Do those results convince you that those picks are going to be more valuable than their neighboring picks, let alone some of the higher picks (like the 3rd) that display "downside surprises."
I'm arguing that picks 3-10 or so aren't all that different in value, and it's all a matter of perspective and scouting, so no, they wouldn't convince me.