skeptical:
You get conservation of energy in classical mechanics because a gravitational force field is a conservative vector field. It takes energy to move upwards in a gravitational field--or, in an accelerative model, it takes energy to resist the acceleration due to this "Universal Accelerator". In any case, a la Einstein, because a gravitational acceleration is indistinguishable physically from a linear acceleration, the physical rules of potential energy and conservation of energy must hold in both cases.
The amount of energy it takes to raise an object up [d] meters in Earth's gravity field at sea level, assuming a small value of [d], is [mgd]. Thus, an object m1 at height d1 has [m1gd1] - [m2gd2] more energy than an object m2 at height d2. The difference in energies between an object and a given rest object is referred to as potential energy.
And, believe it or not, potential energy exists in other aspects of science, too. Any linear harmonic oscillator works because of the sinusodial conversion of potential energy into actual energy, and back.
Consider this: You have a mass hanging vertically on a spring. This mass is oscillating up and down with SHO motion. At the point of equallibrium, the object is moving the fastest, with a kinetic energy of (mv^2)/2. If there is no such thing as gravitational potential energy, then where is the energy coming from when the object is at the node of its oscillation? Where does the energy go when it's at an antipode?
To answer the second part of your question, if there is matter accelerating downward elsewhere, then what is the means by which that matter is being accelerated? Where is it? Are there giant thrusters under the Earth's surface emitting huge quantities of matter?
Lastly, we can MEASURE the existence of the gravitational force. As a matter of fact, Newton himself did it when he proposed its existence. It is extremly difficult because the force is so very weak, but here is the general idea: Take a large, massive, balanced armature and suspend it by a thin strand. Take another larger object and put it near one end of the armature. Ground both elements and hermetically seal the whole apparatus. Wait a few days, and measure the angular deflection.