A program that cannot be wrong is not a program. Here are the four places the Fluid Spacetime picture is most exposed.
F-1 · binary pulsar
Quadrupole residual outside bounds
The current source record narrows the radiation problem to an outgoing quadrupole normalization residual, with the 4PN tail diagnostic split as delta_Q + delta_tail at linear order. Once a frozen branch returns those quantities, binary-pulsar or gravitational-wave damping data can test them directly. A confirmed mismatch outside observational bounds would falsify the fluid route to quadrupole radiation.
→ see topic 10
F-2 · ring-down
Realized throat-response spectrum
Topic 11 does not yet supply a finished ring-down spectrum. What it does supply is the framework and branch-data contract that a realized throat solution would have to turn into one. Once that branch is fixed, gravitational-wave spectra that exclude its predicted response would rule out that throat-identification route.
→ see topic 11
F-3 · anomalous moment
g−2 branch when completed
The current anomaly package organizes the residue into a common quotient / prefactor-slope branch packet; it does not yet justify claiming that the published stack has already derived the full electron or muon anomaly series. If the completed branch realization yields the wrong sign or wrong series shape against experiment, that would falsify the geometric-anomaly route.
→ see topic 09
F-4 · EoS exponent
n ≠ 5 from an independent constraint
The weak-field / PN carry-forward sits on the stiff polytrope exponent n = 5. In the current source record this is fixed by weak-field optical consistency, not by downstream fitting. An independent calculation that forced the same branch to use a different n would force a rebuild or demotion of the picture.
→ see topic 01