This is going to be a little weird so please bear with me for a moment. Based on the results of experiments like the delayed choice quantum eraser it would seem that you could use one of these experiments to determine the results of a future random event. Say for example you ran a series of two slit experiments with a detector on one of the slits to determine which path is followed. The information from the detector is recorded to a hard drive. Meanwhile the photons or electrons or whatever create the appropriate pattern on the screen at the end of the experiment. After running say a half dozen of these experiments, a program on the computer would randomly delete (or zero-write the bits, just so there is no chance at all that the information can be recovered) half of the data sets so that the which-path information was unrecoverable. So theoretically speaking, the presence or lack of interference patterns should correctly predict which data sets will be randomly erased and which will not, prior to the data actually being deleted. I suppose it is also possible that just the mere recording of the data to a hard drive will prevent any interference patterns but it would seem that as long as you devised a way to randomly “lose” that information after the fact, you could still use the presence or lack of an interference pattern to determine the results of the future random data erasing event. If this is true, then you could in turn develop an experiment to predict other random events, such as a lottery drawing. If you could setup the experiment such that there was one data set for each possible outcome, and all data sets that did not correspond with the one that is randomly drawn (the winning lottery combination) get deleted, while only that data set is preserved, then the presence or absence of the interference pattern should correctly predict the random lottery drawing. The difficulty would be in designing the experiment but as long as the data can be recorded (or even held) and then “lost” it seems like it should be possible. Because you can see the presence or lack of the interference pattern prior to the data actually being lost it should provide predictive value as to whether or not the data actually will get lost. So as long as you can tie whether or not the data gets lost to the random event, the interference pattern should predict the results of the event, despite it being random.