Dismissing most of the philosophy that came before, as a prelude to announcing the one correct way to do philosophy, is nothing new. Kant is a prominent example: in Critique of Pure Reason, he dismissed most systematic metaphysics (of his time) as epistemologically dreadful, and then set up what was meant to be a decisive delineation of what pure reason can and cannot accomplish, as a new foundation for philosophy. And when we get to the 20th century, you have whole movements like pragmatism and positivism which want (in a very 20th century way) to dismiss everything that came before; the counterparts in philosophy, of behaviorism in psychology, and formalism in mathematics. Paul Graham's emphasis on "usefulness" seems to mark him as belonging to the pragmatist school. Someone should ask a scholar of pragmatism if he has actually said anything new. 

I don't see another link post to Graham's essay, but similar sentiments are expressed in this post by MIRI's former executive director Luke Muehlhauser, and it even contains the same quote by Bertrand Russell. Look at Luke's profile and you'll see a whole sequence developing his idea of what philosophy should be. 

Now personally, I am in the pro-metaphysics camp. Far from being a meaningless desert, all that old philosophy and academic philosophy is actually extremely rich in meaning, so much so that it defies easy summary. Paul Graham is actually somewhat correct when he says that philosophical metaphysics seems to be short on practical consequences; but something doesn't have to be useful to be real. To use a physics analogy, you can explain most phenomena, and certainly most technologically useful phenomena, by supposing up and down quarks, and electrons and electron-neutrinos, are the only fundamental matter particles. Nonetheless, in reality, there's actually two more "generations" of particles beyond those, and they are a window onto a whole realm of facts that can be studied. The same is true of metaphysics.