Time to reveal the solution to the cat killing mystery introduced in “If Schrödinger's Cats All Die, Do the Alive ones go to Hell?” and further explained in “Rotating Schrödinger's Cat to Death”. The solution is a huge letdown for all cat haters. Instead of being send to hell or at least into a parallel universe where they won’t bother no more, the cats stay right here in our lab! (See Nina – I am not as bad as you think.)
Remember, the question was: If we interact with the Schrödinger cat inside the box so that we will only have dead cats result, what do the alive cats see? There must be a place into which those cats can jump. However, it cannot be the outside of the box that they went into, because that outside includes the experimenter, and the experimenter only observes dead cats.
The answer to the question is surprisingly simple. In short: To do what was discussed, we need a very elaborate setup in order to prepare the photons that are directed into the box. In order to definitively retrieve a dead cat from the experimental box while the alive cat does not notice a thing, the |Alive> cat must quantum teleport out of the experimental box and into the photon-preparation setup. Where do the alive cats jump out; what do they see if the box opens? They see the photon-preparation room; that is where they jump out!
You may object: “Wait a minute – now there are at least two cats every time, namely a dead one in the experimental box, and the same one, dead or alive, in the photon-preparation room.”
This is indeed the moral of the story. If we want to achieve what we set out to do, there is no other way. It is a lesson about the in popularity rising quantum teleportation, which is widely misunderstood and which I plan to demystify further in future posts.
Let us now get into the meat of the explanation – there is no shortcut to the lazy ones shorter than this:
The kill-mechanism involved a single initial photon (different from the many photons send in later by the photon-preparation setup) that was put into a superposition state. If we want to change the state inside the box so that only dead cats result, we must effectively let the dead and alive cats re-interfere. In other words, we must effectively resurrect the state as it was before the dead/alive decision occurred inside the box. We must effectively resurrect the initial photon. “Effectively” is again meant in the sense explained in “Rotating Schrödinger's Cat to Death”.
In order to let the states effectively re-interfere, we must take care of, or in some sense know of all the so called decoherence terms, that is all the entanglement that the two photon states have undergone with the cat. In other words, we would need to “know” the complete state of the cat in its entirety, because only then is there any chance to revert all the entanglement. In order to know the cat completely however, we need an exact copy.
We assumed linear quantum evolution as it has been observed by experiment, which means the no-cloning theorem as proved by Wootters and Zurek  holds. We cannot copy any quantum state. There is only one method to have two quantum states that are exactly the same: Prepare (make) the whole state completely twice.
Recall that the problem was clearly stated in a way so that the |Alive> cat does not notice something intrusive is happening. The whole problem is only mysterious if the cat after three minutes still really sees the box starting to open and expects that the laboratory is still out there. If your method to kill the cat is noticed by the |Alive> cat state, then it is neither the |Alive> cat state as discussed by Schrödinger, nor is it at all interesting: If the cat notices being killed in any way, you might as well just shoot it – no quantum physics required.
If the cat is not to notice the diminishing of the probability of her ever leaving the experimental box, the photon-preparation setup must already have something inside that knows the complete state of the cat in the experimental box, something that is effectively a copy. When the photons interact with the inside of the experimental box, the box and the photon-preparation setup become entangled, exactly like it is known from quantum teleportation.
If you want the alive cat to not notice anything, you must effectively teleport the superposition state out of the experimental box and into the photon-preparation set-up.
 W. K. Wootters and W. H. Zurek, Nature299, 802 (1982)