I must store some references to good recent talks, in part so I don't forget them:

Two talks by Tsuyoshi Ito:  an IQC theory lunch talk about how the complexity class defined by interactive proof systems with two provers who may share *arbitrary non-signaling correlations* (rather than just quantum ones) with each other, turns out to be PSPACE.  Communication between verifier and provers is classical.  This is the same class as that defined by quantum interactive proof systems (for any fixed number of provers, but *without* shared correlations, quantum or otherwise)---a result due (at least the hard direction, that PSPACE is contained in QIP) to Jain, Ji, Upadhyay, and Watrous).  Very nice stuff.  (Note that as far as I know it doesn't imply that two-prover proof systems with classical communication between verifier and provers, but verifiers sharing quantum entanglement, is in PSPACE.  Although this latter model gives the verifiers, collectively, strictly less power, the reduction in their collective power could increase the power of the proof system:  crudely speaking, the provers might have less power to attempt to coordinate in bamboozling the verifier into wrongly accepting.

The really kooky, like me, might ask, what if we allow the communication between verifiers and provers to involve exchanging the kinds of systems involved in the provers' shared non-signaling correlations.  "Popescu-Rohrlich bits", or generalizations involving larger numbers of outcomes per measurement, and larger numbers of measurements.  (States of "semiclassical test spaces", in a different lingo.)

Tsuyoshi's other talk was an informal one at the quantum information group meeting at PI, about XOR games with more than two players.  More on it later.

A few weeks ago Stephanie Wehner gave us a lovely informal quantum foundations seminar on a recent paper she wrote with Jonathan Oppenheim, whose thesis is that if quantum theory were more nonlocal it would violate uncertainty relations.  The uncertainty relations involved are a new type she (and Andreas Winter?) introduced, called "finegrained uncertainty relations".  I'll try to summarize this talk in a later post, too.