I don't know the grandparents reasoning for the statement but if I where to assume, it would be that many of the popular implementations support much of the functionality natively. If which it may be that the grandparent is not considering the case that many use underscore as a polyfill to support such functionality on a broader array of clients that may or may not support the functionality of underscore natively. The grandparent may have other reasoning but that would be my first guess.
