It all depends on what you include in the model.
If you sealed both ends of the tunnel, pumped out all the air, made sure the rock would never touch the side of the walls (maybe you drill a 1 km wide tunnel and the rock is 10 cm across) then the rock would fall down the hole accelerate until it hits the center of the Earth, then it would keep going but instead will slow down and come to a rest on the other end of the tunnel - where you could just grab it.
Then, if it's not grabbed it would fall once again, repeating the whole thing the other way around.
The reason is that once you're inside a spherical shell the pull of the shell cancels itself out. It's a bit complicated in both mathematics and writing the equations, but it comes out that even your position within the shell doesn't matter.
So if you go 30 kilometers below the surface that means that a 30 km sick shell no longer pulls you down.
So if he could make a pocket of the deep Earth (through magic!) inhabitable since the gravity would be "effectively lower" there all kinds of giant creatures could live. there.
Whee! Found it:
http://www.merlyn.demon.co.uk/gravity1.htm#GoSSh
Field Inside a Spherical Shell
It has been said that : "the field inside a uniform spherical shell is zero (which is indeed so for any 1/r2 type force); a non-obvious, non-trivial result needing calculus for proof."
This is not so; the cancellation of field is correct, but there is no need for calculus. This is relatively well known in the case of gravitation, but not for electrostatics. In the following, "charge" stands for "electric charge", "mass", or any other source of an inverse square field; and "small" means "infinitesimal".
Consider an arbitrary point X inside a uniformly-charged spherical surface, and consider both an arbitrary small element of "charged" surface area A and that second small element of area B which is marked out by straight lines from the edge of the first area through the point to meet the sphere again on the opposite side of the point.
The solid angles subtended by A and B at the point X are equal, and the directions XA and XB are opposite. Simple geometry makes it clear that the areas A and B (which have the same "tilt") and hence the charges are proportional to the squares of their distances from the point X. Their fields therefore cancel at that point; so there will be no net field from the whole set of such areas which covers the whole sphere, for any point X.
The argument is correct; I think I have expressed it correctly. Ramsey (s.3.2) says that Newton used it, in Principia Proposition LXX; indeed, it is in Hawking's cited book, p.880.