Beso's point about the billions of stars and only lottery winners getting to observe the draw is a good one and refers to something called the Antropic principle.
However it is limited.
There are indeed billions of stars but there is only one Universe - Only one speed of light in a vacuum, only one Gravitational constant, electronic charge, Plancks constant. Yet all these values come together to form a Universe capable of supporting stable matter, of stars of chemistry that can support life.
This is why some people have postulated that there must be other Universes where these are different but we have no other evidence for this.
I am suspicious of empirical laws like Zipfs law. The reason is that where you see it it becomes remarkable. Where you don't see it it is ignored.
Statistics are very good at tricking us. Some things are much more probable than we expect. Like the probability of two people in a group having the same birthday. Or the famous Monty Hall problem where it's better to switch doors (
http://en.wikipedia.org/wiki/Monty_hall_problem )
I suspect that Zipfs law is much like this, that really it only seems remarkable to us because we lack a good enough understanding of the statistics involved.
A similar principle is Benfords law
http://en.wikipedia.org/wiki/Benford's_law which says that in any list of data (most commonly from the real world) the number 1 appears most commonly