Hi Philip, you have successfully proven that it is possible to link some unknown but assumed to be existent and exclusive abstractions like information and probability to derive the ultimate unknown abstraction, nothing = something. However, the question remains, do we really know what 'something' is in-itself and do we really know what 'nothing'is in-itself? Surely not, since we even do not know what the term 'information' should mean to discriminate between 'nothing' and 'something'. According to the zero-information approach of Marc, nothing must be something, it merely cannot be fully formalized in bits or other formal systems, since it has zero bits of 'information', zero bits of formalizable content. This tells me that the whole menue of 'nothing' as well as that for something cannot be completely understood by human beings with only the menue card at hand.
Every shortest description of 'nothing' or 'something' must remain incomplete, since it neither can determine the essence of either of them, nor their relationship other than concatenating two unknowns to come to a third unknown. Otherwise one could say that the shortest computer program that is able to emulate 'nothing' has exactly zero bits and is complete - and that therefore an infinity of such programs run unnoticed permanently on our computers, non-existing programs that emulate, well, 'nothing'.
I think the failure here is to assume that undefinable, unknowable things must necessarily be equal to non-existent things and that non-existent things must necessarily be equal to unknowable, undefinable things. But this would mean that the non-existence of a real elephant in my room is an unknowable thing and that there could well be a real elephant in my room, albeit in a rather undefinable manner... so, just a moment... where are you...elephant...at least there is the potential for such an elephant to be here, since I have enough space in my room... and you are invited to guess whether or not there is indeed an elephant in my room at the moment - and state some probabilities for either case. The big question is, I think, which things we should reasonably consider as non-existent and which things we should reasonably consider as existent, but undefinable.