The operation principle is based on clever probing on the properties of the resistive touchscreen, particularly the reduced resistivity of one plane when parts of it get "shorted" via a "bypass" through the two touchpoints to the other plane. The principle is known since ~6 or more years. No capacitive technology involved.
(actually the connects "touchpoint1" and "touchpoint2" also have some resistive component depending on pressure and area/diameter of the touchpoint - I dropped that for clarity of the schematics) Think of plane1 as upper and plane2 as lower plane. On touchpoints they contact each other.

Simplified explanation:
You got 4 "wires" X-Left, X-Right, Y-Upper, Y-Lower, and you can probe 6 distinctive resistivities from that 4 terminals. I leave it up to the reader to do the permutations. 6 input variables to your clever algorithm are good for delivering a solution for X, Y, pressure - for two touchpoints. It all depends on the math you do. Some sets of input variable values may yield more than one valid solution, in that case you can't say for sure which is the right one. This clean mathematical analysis also shows why you can't do 3-touch - your max 6 input variables are not supporting any algo that yields 9 resulting values for 3 touchpoints (for 5-wire r-ts this looks different).