# why l1 norm can result in variable selection but not l2 [duplicate]

In applying a penalty term with either $l_1$ or $l_2$ norm, why would the former result in variable selection but not the latter?