Celedam wrote:The takeaway seems to be that no matter what you do to change events in the past, those events will realign themselves to continue in the same direction and maintain causality.
"Will realign" is, I think, too strong. From my vague impression of reading the non-mathy parts of the paper in question, what I think they're claiming is that if you go back in time and try to kill gramps, there's a series of dynamical processes, however improbable, for you to exercise your free will and commit the murder, and yet for the present to still exist unchanged. That series of dynamical processes might be that your grandfather's cells spontaneously un-decompose and he mysteriously comes back to life (and also that everybody somehow forgets that he died in the first place!) The point is that they have shown there exists SOME path back to your starting conditions. A "closed time-like curve" can in theory be shown to exist on paper. Not that the path has any likelihood of ever happening in a thermodynamic sense, or that it must or shall inevitably happen. Merely that its existence is compatible with reality.
We've known for a very long time that dynamic processes are generally time reversible. An single atom moves from point A to point B. It can just as easily move from point B to point A. What stops eggs from unbreaking and ice cubes from unmelting is the 2nd Law of Thermodynamics, which is really more of a probabilistic principle than a law. But in theory those unlikely events could happen. Similarly any action that you undertake by going to the past ought to, in theory be able to be undone by a series of finite interactions. What they've done in this paper is basically prove that that is the case.