This is nice because there's no three-sided Platonic solid, but this way you can have a die with an equal likelihood of each of three outcomes. The same technique could be used to made an N-sided die for any N.
There is a chance (albeit a small one) that the die could land on one of the two ends. So it's really a five-sided die. But you could just make a rule that that's a "push" and the die should be rolled again. Of course the same kind of rule permits three possible results from a regular old cubic die, though a person who points that out to the inventor of this clever shape would have to be a poor sport. (Or a mathematician.)