As both Dick and Gib Z have pointed out, first, you have to define "bigger" for functions!
There are three definitions commonly used: max |f(x)| where the maximum is taken for x in a specific set, [itex]\int |f(x)dx[/itex] where the integral is over a specific measurable set, and [itex]\sqrt{\int f(x)^2 dx}[/itex] where, again, the integral is over a specific measurable set.
Technically, those are referred to as the "uniform", L1, and L2 norms, respectively.
In any case, when you are talking about how "big" a function is, you are talking about the y value. You appear to be looking at the "width" of graph- the distance horizontally, from the y-axis. That is exactly the opposite of what we would normally think of as one function being bigger than the other. x2 looks "bigger" horizontally than x3 precisely because for a given y, x must be larger in x2. That, of course, is because, for a given x, y is larger in x3.