Layra-chan
I wouldn't say gravity is weak as much as I'd say that protons and electrons don't have a lot of mass.
You're completely right, but then there is nothing unkosher in making such comparisons through stable elementary particles. One usually says that a force is strong (weak) in that the corresponding coupling constant is large (small), which definitely makes a lot of difference when one employs perturbative methods. So, for EM, you have a Planck charge Qp = sqrt(4πℏcε₀), and the corresponding coupling constant as α = (e/Qp)². Doing the same thing with the Planck mass Mp and an elementary mass (say electron, for consistency), you have a gravitational coupling that's absurdly tiny by comparison.
Then with the understood dependence on elementary particles, these are all equivalent questions:
-- "Why is gravity so much weaker than other forces?"
-- "Why is the gravitational coupling so much smaller than that of other forces?"
-- "Why is the Planck mass so freakin' big compared to elementary particles?"
Layra-chan
Besides which, if gravity were escaping along other dimensions than the 3+1 we normally observe, then intuitively (naively?) we'd get something other than an inverse square law; we'd get an inverse (n-1)-th power law, where n is the number of spatial dimensions we have.
Yes, but it's modified over large distances when the extra dimensions is compact. Intuitively, a mass smeared over a small compact dimension won't look too different from a point-mass at large distances, so we can pretend to spread it throughout the volume of the extra dimensions, which reproduce inverse-squared behavior with an altered gravitational constant (divided by the volume of the extra dimensions).
Gauss's law is correct in all dimensions; writing it in terms of a potential gives the standard Poisson's equation ∇²V = 4πGρ. In Cartesian coordinates (w,x,y,z), let R=x²+y²+z² and r²=w²+R², the potential of a point-mass m should be
[1] V = -Gm/(πr²)
Compactifying the w-direction into a circle of radius a, the identification w~(w+2πa) wraps around the potential, repeating it over w+2πan, producing a summation over all integers n:
[2] V = -(Gm/π) Sum[ 1/(R²+(w+2πan)²),
Now, if I let w = 0, then the summation is (R/2a) coth(R/2a), so for very large R, this is essentially.
[4] V₀ = -G'm/R,
where G' = G/(2πa). Nonzero w might also has a closed form, but I'm not sure how to get it. It's probably something hellish involving digamma functions.
Edit: I don't know what the hell I was thinking here... that's it; I'm going to sleep instead.