Worlds qua truthmakers
I have the vague impression that Lewis' paper 'Things qua truthmakers', and in particular the appendix by Lewis and Rosen, proves something important. But I'm not sure what it is. Maybe it's that the request for truthmakers was thoroughly misguided in the first place.
The problem is that the truthmaker principle is saisfied so eaily: Let 'w' be a name for our world that does not apply to any qualitatively different world, nor to anything inside any world. (That is, 'w' denotes our world under a rather strict counterpart relation.) Let T be any qualitative truth. Necessarily, if w exists, then T, since otherwise 'w' would be applicable to a world in which not-T, even though T holds at our world, contrary to the rule just stated. Hence w is a truthmaker for T, that is, for any truth whatsoever.
This really looks like a cheap trick: Want a complete description of fundamental reality upon which everything else supervenes? Here you are: 'w exists'. Note that this time, unlike in the trick mentioned in 'New work for a theory of universals', where the complete description was 'everything is F', no exceedingly gruesome property is involved. The fundamental fact doesn't even contain a (non-logical) predicate. So what's wrong this time?
Nothing, I think. Complete descriptions of fundamental reality are really much easier to come by than I thought. However, not every complete description is metaphysically interesting. An interesting description tells you something about how every truth is made true by it. And here is where the request for truthmakers went wrong: If someone comes up with an ideologically problematic truth, and we ask what it is that makes this thing true, we won't be satisfied with the answer 'the world as it is'. What we want is an analysis, an explation of how the world makes that truth true.
By the way, for any truth T, 'if w exists then T' is necessary, but hardly a priori, otherwise T itself would be a priori. So the fundamental facts from which every truth is a priori deducible can't just be any facts that necessitate every truth. They also must be 'metaphysically interesting'. Maybe what I always thought of as the fundamental facts can best be described as those facts from which every truth is a priori (or, I'd prefer, analytically) deducible. The deducibility thesis would of course then be true by definition.