In standard delayed choice quantum erasure experiment ( http://en.wikipedia.org/wiki/Delayed_choice_quantum_eraser ), the which-way information is randomly erased in one path, what chooses between classical and quantum behavior for the second path (if there is interference pattern). The "delayed" means that changing lengths of optical ways, we can make that the erasure (reason) is made after the result. As the erasure is made randomly, we couldn't use this time reversed reason-result relation. But what if there would be a configuration where we can control if the erasure is made? And it turns out that there is such configuration - proposed and tested by group of Walborn. Here is a decade old http://grad.physics.sunysb.edu/~amarch/Walborn.pdf [Broken], http://grad.physics.sunysb.edu/~amarch/ [Broken]: http://grad.physics.sunysb.edu/~amarch/PHY5656.gif [Broken] BBO produces two entangled photons - first polarization x, second y or oppositely. Photon s comes through double slit before which there are installed two different quarter wave plates(QWP): changing linear polarization to circular in two different ways. Finally there are two possibilities: x y R L y x L R where succeeding columns are: linear polarization of p, initial linear polarization of s, circular polarization of s after going through slit 1, circular polarization of s after going through slit 2. So if we know only the final circular polarization of s, we still don't know which slit was chosen, so we should get interference on the second path. But if we would additionally know if p is x or y, we would know which slit was chosen for s and so interference pattern would disappear. So let us add linear polarizer on p path - depending on its rotation, we can or cannot get required information - it can erase the which-path information, choosing between quantum and classical behavior for s. If s detector is in the first minimum of interference, we control statistics it observes by rotating polarizer on path p. Changing lengths of optical paths, we would get delayed version, which seems to allow to control statistics in the past, by rotating polarizer in the future. It suggests that something is wrong with such setup, but I don't see what - does anybody see where is the problem?