The standard resistive touchsreen (as in N900) with uniform resistive foils (usually indium-tin oxide, or ITO in short) does NOT allow multitouch in any sane and economically mass-producible way. The resistive multitouch that has been so much advertised lately is owned by a single company, Stantum. Their implementation splits the resistive foil into thin, cross-arranged slivers of material allowing a very high number of touch detection points - the maximum number of touches is only limited by the width and density of the slivers (and of course, the amount of reading pins that the poor controller chip has to provide).

The N900 uses a so called 4-wire resistive touchscreen with a TI-made controller, according to most sources the TSC2005. It's a very simple way to make a touchscreen with only a few wires while maintaining very good accuracy. Someone might be interested enough to look into the possibility to provide a Stantum panel and a controller for N900 (or even a captouch panel and its controller, Atmel MAXTouch might even get me excited), but interfacing it will be a problem as TSC2005 is an I2C bus device and most likely in a very non-DIY friendly package - there's no I2C socket that you can just plug a chip into. The driver on the OS side will naturally have to be rewritten as well to support the new approach.

Resistive multitouch is a hardware limited thing, it can't be just conjured out of thin air on any random controllers just by adding a few lines of code. On capacitive screens, the difference between single-point and multitouch solutions is not as simple as both multitouch and single-point capacitive touch solutions use very similar panels, the main differences being in how well the controller can process the data it reads AND what the host OS can do with it.