Discussion of "Physical Limits of Inference" article http://arxiv.org/abs/0708.1362 I am part-way through this article and so far it's quite exciting for me to read it. I have many comments and discussion points, as well as questions of my own to throw into the venue here to see what people's answers might be. While the title of the article might seem like I completely misplaced this thread, the key words are: Turing machine, automata, observation, prediction, multiverse, Kolmogorov Complexity. Consequently, I wasn't sure where exactly to put it but after a quick skim, I think you'd agree it belongs somewhere in the math section. I'd like to jump in and post my questions and remarks about this article but I need time to organize my thoughts (this will also give you a change to look the article over if you haven't already). However, I do want to telegraph the fact that I want this thread, if possible, to start with a focus on section 9 on self aware devices. Let D be a self-aware device. (This has a formal mathematical definition and formulation in the article.) Now, to somewhat abuse language, pose the following question to D: Is D a self-aware device? You can see that I don't quite understand these formalizations yet as I am not sure what the result of posing to D, "is D a self-aware device?" Now if D can correctly answer the question, "is D a self aware device," then I would say that D really is self aware because it at least thinks it knows what it is, or at least, what it's essential nature is, that it is a self-aware device. It could be interesting if the result were something like :no self aware Device can know that it is a self aware device: or :all self aware Devices can know that it is a self aware device:. However, I think the most "interesting" truth would be that some self aware devices can correctly answer the question "am I a self aware device" while the rest can not. If so, what is so special about the some self aware devices that think they are self aware devices? btw, for D to be self-aware in this context means that it is aware of what the question is. I think self there is, then, a misnomer, since D is not a question, D is a machine that answers questions (ie, a device).