This plot comes from Komossa & Merritt (arXiv:0811.1037). It's about gravitational recoil of supermassive black holes in mergers, and the implications for unified models of AGN. There's a hint of a rumour of a possibility, I gather, that the fraction of 'type 2' quasars (ones where the central region of the quasar is obscured by a surrounding torus) doesn't match up with the fraction of type 2 Seyferts. We don't see as many type 2 quasars as we'd expect, which is odd if the obscuration in both cases is just a result of the angle at which we're viewing the object.

The authors looked at some N-body simulations and noticed that in major mergers gravitational wave emission can kick the central SMBH in a quasar a significant distance away from the nucleus of the galaxy. Far enough, perhaps, that the formerly obscured central engine will pop up above the obscuring torus, and make us think we're looking at a type 1, when we're really looking at a type 2.

Weird.