Hrm. Alternative question: Has anyone re-evaluated / re-derived special relativity using Terrell-Penrose rotation as the basis?

Or is there a reason that the assumption underlying SR - that an inertial observer will see light moving further by a direct proportion of a moving observer's velocity and the speed of light than that moving observer - being incorrect doesn't matter?

All the centrifuge and orbital / circular experiments don't need SR to work as advertised since GR and the equivalence principle mean they'd work the same anyways. And as far as I know that takes care of empirical observation.

...actually, given Einstein's ongoing frustration with how SR was handled, and his insistence it was acceleration and not velocity which caused time dilation, it wouldn't surprise me if the original version of SR didn't have that particular issue.

There is a limit to how much observers can disagree about distance without a privileged reference frame, and in the particular case of distance between them when they are moving directly towards or away from each other, that limit is not at all. This applies to an observer at the front of a ship and an observer at the back of another ship, and every position in between.

(I think they might be unable to ever disagree about the distance between them, but I might be missing some edge cases there.)

(Noting that they can disagree with the observer at the front of the other ship about the distance to the observer in the back of the ship.)

And if that seems perfectly obvious, consider two observer ships at rest with each other, both docked at either end of a one light-minute long tube, and consider what Lorentz contraction might lead you to expect to happen to the distance between them when one undocks and starts accelerating towards the other.

And if it now seems obvious that they can disagree on distance, that implies that two ships can tell which is really moving just by flashing lights at each other to communicate distance, and whoever measures the shorter distance is the one who is really moving; indeed, a docking contraption such as just described in that observer's reference frame would appear to extend, rather than contract, when that observer moved to the reference frame of the observer measuring the longer distance.

(Well, it would do both, as the metric extended but the tube contracted. But without rotation, once synchronized, they would disagree on the final length of the tube.)

Wait, no, last two were incorrect. Forgot to account for simultaneity. It contracts fine. Still, privileged reference frame.

Ah! The original version lacked simultaneity. I need to stop using other people's thought experiments. The tube can't contract in the first place, just rotate, since the speed of light is finite.

Hrm... of course, the curvature of the path rotation implies acceleration from other reference frames. So we might end up right back where we started.

That would be more convenient for my crackpot hypothesis that Lorentz Contraction is a type of curvature, and causes motion rather than the other way around.

Having it avoid causing time dilation was a serious headache.