Citations should primarily serve to mention relevant work, which often includes authors' earlier works. The problem really is the abuse of citation metrics and

Citations should primarily serve to mention relevant work, which often includes authors' earlier works. The problem really is the abuse of citation metrics and journal brand names (and especially journal-based metrics) as a means of evaluating researchers. What we really need is a different method of evaluating researchers that does not rely on where they publish or what they cite. (But I would say that, given that I work on one such a system.)