I suppose none of these conjectures can be ruled out a priori, they don't obviously violate any known physics, but that doesn't mean they are likely, or even possible. I'm not convinced by the whole artilect/Jupiter brain/Matrioshka concept. I don't think its impossible, I just think its pointless.

I've corresponded with Dr Bradbury, who seems a very well informed and likeable gentleman, but he can't give any good reasons (to me, at least) why this kind of computational power would be necessary, or even desirable. Sure, we can postulate potential applications, like simulating entire alternate universes full of intelligences, or modeling every particle in the cosmos, but why would anyone want to do that? The whole purpose of science is to find simplifications of natural phenomena so they can be described and predicted with elegant and compact equations. We rely on numerical modeling when we don't understand the science completely, it's an engineering technique, not a scientific goal. Advanced societies won't need to model hurricanes or economies, they'll have an equation that will fit on a t-shirt. They won't need to know all there is to know because they will already know all they want to know.

I would hope that truly advanced societies would find better things to do with their time than calculating pi and e to a googolplex of decimal places, or predicting the future history of every fundamental particle in the galaxy. Even the creation of artificial intelligences will get old after a while. Not to mention they sound downright dangerous, especially if we give them access to the sensors and the servos. I really think that this supposed desire to increase computational power to such exaggerated extremes appeals to us only because we are so bad at it, we've only been doing it for half a century or so, and it is the primary hobby of computer literate people. Eventually, we'll be able to do what we need to so easily that we'll be able to concentrate on what really matters to us, things we probably can't even conceive of yet. And I suspect that may require no additional logical operations at all.