An Experiment in communicating with the past

Edit note: Okay, this has been draft for a couple days and I'm done caring about making it pretty. Just a quick edit through for logical cohesion and grammar then done. 

Just listened to the description of the dual slit delayed choice quantum eraser experiment. It occurs to me that this is a fascinating effect. Consider this: Take the down converters as described in in the book and send the idle photons down a loop or path long enough to delay their arrival at the subsequent series of detectors by a day or just a couple of seconds but spatially separated such that the two parts of the experiment cannot observe each other to make the experiment more technically feasible.

Rig three idle photon detectors. One for each path and one to accumulate both (I think this minor alteration of the experiment as described would still work since you wouldn't be able to glean path information from the accumulated photons though clearly this wrinkle can be ironed out to exactly match the delayed choice experiment)(a tangential question is would the accumulative detector display an interference pattern)

Set up a splitter on each incoming stream such that you can control the path of both streams to simultaneously either be detected with path information or detected indeterminately.

Now send you photons down the experiment in sets. However many photons are required to conclusively display an interference pattern or lack of one, send them in sets of that many with enough rest time between to easily differentiate the sets when analyzing the data. Presumably there doesn't need to be much separation between individual photons as we aren't concerned with confirming that they act like waves individually. We just want to see if they're acting like waves or like particles for any given set of events.

Analyze each set for a series of binary information based on the presence or absence of an interference pattern (wave or particle interactions).

The next day, or however long your delay loop is, or in the next room depending on the technical level of your experiment your associate who has not been compromised with the data acquired previously comes in and using the splitter control coerces the idle photons to create the binary data you received yesterday.

Would that really work? Can I send myself the lotto numbers? And, would that be illegal? >.>

This also all depends on my understanding that these experiments do not require vacuum. Which is to say that the probability wave doesn't collapse until a human observer looks at it (if any particle interaction counts as an observation then the whole quantum eraser experiment ceases to make sense to me since both the 'eraser' and the initial polarizing are clearly interactions with the particles).

This is also an interesting consideration when remembering that light, due to its speed through space, is not moving through time. Technically the photons that reach each experimenter are the same age and so trying to decide if we would actually send information into the past makes my head hurt.