All of this reminds me of a "Psi" experiment that I once read about (and I can't seem to find it online right now, though I'm sure it's still there). In summary, the experiment demonstrated the validity and reality of "retrokinesis" -- i.e. being able to change the past with one's mind. I can't recall the precise details of the experiment, but I believe that it went something like this:
A computer produced a series of monotones in a completely isolated room which were recorded to tape. The computer could either produce a tone in the left speaker channel or the right speaker channel. A program was used to ensure that the series of monotones was completely randomised, so that one would expect approximately 50% of the tones to be recorded as LEFT, and 50% to be recorded as RIGHT. Upon completion of the randomised recording, the tape was sealed in a container, put in the bottom of a drawer somewhere and forgotten about.
The next stage of the experiment involved a group of people (known to have "psychic" abilities) whose task it was to "influence" with their mind the outcome of the "randomised" tonal recordings which had already taken place. These people had no idea of when the recordings took place, or where they took place. In fact I don't even know if they knew about the recordings to begin with, not until after they'd already occurred. Anyway, these subjects were asked to cause more LEFT tones to appear on the tape, for example.
After this "retrokinesis", the tape was played back. The results amazingly showed a statistically significant bias towards the LEFT channel. The experiment was of course repeated many times, with different people, and the results were nearly always the same. Both RIGHT and LEFT bias was induced according to the thoughts of the participants.
Another strange finding was that if someone who didn't have any particular "psychic" abilities tried to change the results, and then a more psychically inclined individual attempted the same thing on the same tape, the results were neutral -- that is, no bias was found. In effect, the mere act of trying to influence the past forfeited the success of any subsequent attempt to influence the past!
The only conclusion is that until something is observed, it's not concrete or real. Perhaps the experiment should be repeated with a dog or cat in the recording room.