Regarding build table weight and skipping teeth - this is a mass independent issue. A heavier table will require more force to make it skip, but you also end up with higher acceleration forces trying to make it skip. Increasing or decreasing the table mass has no affect on the peak accelerations you can achieve. The primary factor in determining skipping is the pressure angle of the gear teeth (we use 20 deg. pressure angle pinion gears). With a 20 deg. pressure angle the maximum theoretical acceleration is well over 1G (9807 mm/s^2).

Regarding backlash, small errors in the teeth profile will create some lost motion (not actually backlash because the teeth are always in contact). In practice, though, these errors are small compared to all of the other mechanical errors of bearing play, mechanical stiffness, motor positioning errors, etc, etc. Most discussions of accuracy often miss the major sources of errors. For instance, people talk about using 16x or 32x microstepping and think they are getting 0.01mm or 0.005mm position accuracy, where in fact steppers running at speed or under a load are still only accurate to +/- 1 full step -- the equivalent of 0.15mm accuracy. Structural deformations under acceleration loading further degrade the accuracy, but these sorts of dynamic positioning errors are very difficult to measure directly. At the end of the day, what you really have to look at is the quality of the print.