Black holes are a mistake of general relativity (GR), specifically their absolute event horizon is incompatible with GR's equivalence principle postulate, as the blog (which is not mine) shows. In software terms, black holes are a bug in the theory.
Let's remember that Einstein disbelieved in black holes and for many years tried to prove they couldn't exist. Also there is no definitive evidence that they actually exist in nature. By their definition, such evidence can't be detected.
The argument there looks wrong to me; in particular combining the equivalence principle with the proposition called (K) there seems wrong. EP in the form given there says that no test within the small region in question can distinguish it from a similar small region in a gravitation-free universe, but "seeing whether the small region can be extended to a larger region" is not a test you can perform within the small region and it isn't an EP violation if you can do the extension in the gravityless universe but not in the with-gravity one.
Other articles on that blog claim to solve other major problems in physics, in every case giving an answer that disagrees with (what I understand to be) the consensus of working physicists.
So I could trust the author of the blog, Just Because (which doesn't sound like a great strategy). Or I could trust my own opinion (which is that on the one point I looked at it looks as if there's a substantial conceptual mistake). Or I could trust the current mainstream consensus.
I'm having trouble seeing reasons why I should believe what the author of that blog says. Do you happen to know of any?
I'm not concerned with trust or consensus, those aren't scientific concepts. If anyone shows a logical inconsistency with GR, then it's invalid regardless.
> but "seeing whether the small region can be extended to a larger region" is not a test you can perform within the small region
That law K doesn't apply in a frame falling through a horizon (a logical test within that frame, which is defined to be an inertial frame) indicates that physical tests can also show that the laws of physics differ between that frame and other inertial frames. The picture in the blog is an example of that. It's impossible to conduct a test where the cloud described there isn't splitting in two, unlike in some other inertial frame.
I agree that trust and consensus are not scientific concepts (well, except maybe in psychology or sociology, but that's not relevant here). But they are things it's reasonable for those of us who aren't expert theoretical physicists to take into account, when Some Guy On The Internet says all the professional physicists are wrong and he's found an internal inconsistency in general relativity.
The "larger region" you need to look at to determine whether K applies is not small enough for the defining characteristic of the small region -- i.e., that spacetime curvature is negligible there -- to hold. The equivalence principle stops applying once you start considering regions large enough for spacetime curvature to be non-negligible within them.
Go ahead and take that into account, but hinge your conclusion on whether the logic's valid, if you wish to stay fully scientific.
Frame X, in which law K is tested, is defined to be small enough that the spacetime curvature is negligible there. So it's definitely small enough. The test of law K is conducted wholly within frame X.
That shows that you have good intuition. It is wrong; it relies on incorrect assumptions about how a local inertial frame straddling the horizon works. See previous discussion on HN in these threads:
Black holes are a mistake of general relativity (GR), specifically their absolute event horizon is incompatible with GR's equivalence principle postulate, as the blog (which is not mine) shows. In software terms, black holes are a bug in the theory.
Let's remember that Einstein disbelieved in black holes and for many years tried to prove they couldn't exist. Also there is no definitive evidence that they actually exist in nature. By their definition, such evidence can't be detected.