This is a twisting of what Krauss says. Krauss merely shows that the current known laws of physics are not inconsistent with this scenario. He does not claim there is evidence that proves this is what happened. It's not much different than seeing that some solutions of General Relatively permit warp travel. It doesn't prove it can work, merely that the current known theories allow it.

Also, the comments he makes on determinism are nonsense, and resemble the claims of creationists. A computer can't arise from the laws of physics because it had a designer? By transitivity, this implies human beings could not have been the result of bottom up physical processes either. Is what way is this claim scientific or testable?

"there is a group of people out there writing papers based on the idea that physics is a computational process. But a physical law is not an algorithm. So who chooses the computational strategy and the algorithms that realise a specific physical law? (Finite elements perhaps?) What language is it written in? (Does Nature use Java or C++? What machine code is used?) Where is the CPU? What is used for memory, and in what way are read and write commands executed? Additionally if it’s a computation, how does Nature avoid the halting problem? It’s all a very bad analogy that does not work."

This just seems really poorly reasoned out. What does the halting problem have to do with it? Total non-sequitur. Why can't the physical laws merely be a kind of cellular automata? No need to solve any halting problems. Things interact iteratively, new states arise, some of those states lead to more complex forms, etc.

Since cellular automata have been proven to be universal, that means in principle, any thing that is computable can be carried out, and therefore, all of classical computer science fits perfectly in this view without having to invoke some kind of Penrose-like "the universe solves non-computable functions" magic.

If you asked Ellis how planets or galaxies formed, he would not say "What algorithm caused star/galaxy/solar system formation", or "how does the computer solve N-body problems of trillions of elements". There's no need to posit a designed algorithm for these things, merely that the system interacts with itself in ways that evolve towards these structures, that human beings produce higher level mathematical models to predict and manipulate more easily.

I don't think any scientist working today works under the delusion that our theories are anything but models.