We extended our haptic paddle to use three motors. The code now times the pulses between the three and randomly picks one of five points to hit. If it hits on the extreme of one side, the first motor fires, then the middle, than the one on the opposite end. The intensity reduces, too. If it's only slightly to one side, the side and middle fire at once, with reduced intensity, and then the far side fires. If the middle is the target, it fires at full intensity, then the two ends fire.

Cursory user testing has revealed some interesting results:

Users don't really have a reliable feel for which side is being hit at first. It takes anywhere from 30 seconds to a minute before their responses start to be consistently accurate. Also, placement of the thumbs and/or finger relative to the side motors has an effect on how accurately they feel the pulses.

All in all, it's starting to work pretty well, though, so I'm happy.