Unambiguously? Nothing; no hypothesis ever does, else it’d be a fact. But it provides, at least in my eyes, probable explanations for a number of issues that are, to date, pretty large open questions:
First of all, it provides a background framework for a dual inheritance theory, i.e. a link between cultural and biological evolution, which has been proposed independently of and I believe also prior to memetics. By identifying the unit of selection on the cultural side of this link, mutual influences are likely to be easier quantifiable than they are at present.
Second, if memes undergo a selection process in an individual’s brain, this sheds light at the question of how exactly we arrive at a decision, and, since decision is the fundamental process of (intelligent) action, how our intelligence works.
Third, and to me, that’s the biggie, because it’s so easily overlooked, yet rather fundamental: if memetics is correct, it might just give us a handle to answer the question of why we are conscious. See, the way it seems to me, we wouldn’t need to be – each and every human action can just as well be undertaken by an unconscious agent. In theory, one can easily envision a table that lists all possible inputs and correlates them with their appropriate outputs; this design can be enhanced, but for the purposes of argument, its mere existence is enough. One could then devise a mechanism that merely looks up the input and gives the appropriate output; I don’t think anybody would argue that this system would be conscious, and yet, it’d be completely able to reproduce any given action a human being can conceivably take.
So, that’s the question: Why are we conscious if it doesn’t seem we need to be?
Now, how does memetics enter the issue? Truth be told, I’m not entirely sure; my reasoning isn’t absolutely sound. Nevertheless, my hunch is that consciousness might be to memes similar as life is to genes – it provides a powerful vector of reproduction. But how, if, as I just pointed out, having or not having consciousness doesn’t affect our actions? Well, it might not affect atomic actions; but it might affect their frequency and grouping, i.e. certain actions become more likely to occur with greater frequency, or in conjunction with other actions, depending on memetic relationships.
This is all very vague, I know, and I don’t really have time right now to elaborate, nor am I even certain my reasoning leads to anything fruitful, but as I said, I believe it’s enough to continue memetics as, as jackdavinci put it, an area of inquiry.
As for ‘not producing any predictions’ – if memetics is indeed correct, then its predictions are basically our whole sphere of experience, which makes it kinda hard to determine any specific – and presently unknown – effects; if those indeed exist, they’d have to be pretty subtle to not be known by now.
And again to address the criticism of being too vaguely defined, take, for instance, a table – everyone knows what it is, but still, a general definition, based on its physical properties, is elusive: the number of legs seems to vary between 0 and n, it can be made out of nearly every material, comes in every imaginable colour and with every decorative pattern imaginable, varies in shape from triangular to circular, and so on and on. It doesn’t seem like a simple, all-encompassing definition of ‘table’ exists, since for nearly every definition I can think of, I can immediately find a counter-example; yet nobody would argue that the label ‘table’ is, therefore, useless.