Ibix said:
Inside the shell we have Minkowski spacetime, with a particular simultaneity convention picked out by the symmetry of the shell.
Yes, this part is fine.
Ibix said:
Outside we have Schwarzschild spacetime.
Yes, but bounded on the "inside" by the shell, whose areal radius ##r## is decreasing.
Ibix said:
So I have two regions with well defined notions of simultaneity.
Not quite. For the region occupied by simultaneity surfaces that cross the shell before the shell crosses the horizon, the Schwarzschild region is entirely exterior to the horizon, so yes, it will have a well-defined set of simultaneity surfaces picked out by orthogonality to the timelike Killing vector field in the exterior region. But you still have to match the surfaces, and there is an issue there that you do not appear to have recognized; see below.
But once the shell crosses the horizon, the Schwarzschild region includes the horizon and a portion inside the horizon, and Schwarzschild spacetime is not stationary there (the KVF is null on the horizon and spacelike inside it), so there are no well-defined surfaces of simultaneity defined by the KVF.
Ibix said:
if I consider a shell of delta-function thickness I can have my clocks arbitrarily close together. So do I have any choice about which event just inside and just outside the shell are simultaneous? You seem to think so, if I understand you correctly
Sort of. The issue I am thinking of is that, heuristically, the Schwarzschild simultaneity surfaces outside the shell and the Minkowski simultaneity surfaces inside the shell are not parallel--except possibly at one particular instant of time. Heuristically, you can see this by imagining trying to match Rindler simultaneity surfaces with Minkowski (inertial frame) simultaneity surfaces, across a boundary at some finite value of Minkowski ##x## to the right (positive ##x## direction) of the origin. Except for the surface at Minkowski ##t = 0##, the surfaces are not parallel. Schwarszschild simultaneity surfaces, heuristically, behave like Rindler simultaneity surfaces, so the same issue will arise with them.
You could argue that the "kink" in each simultaneity surface is due to the idealization of a shell of zero thickness, and that for a shell of finite thickness the kink would be smoothed out into a curve. But then there is ambiguity about how to do the smoothing.