Data Validation and error analysis
Simulator flights would not generate handshake data, consequently would not yield any information about the errors associated with the observed data from the satellite.
Any observed data will have an inherent measurement errors that have to be quantified so that subsequent calculations (particularly innovative and untested calculations) will be useful.
Hopefully the Inmarsat guys have done exactly that, established the total error possible for each of their original conclusions. More specifically, what is the error and confidence that surrounds the "40 degree arc".
I'm sure that engineering folks have been frustrated by the legend given in the BFO plots provided by the AAIB. Assume that the green (triangle) plot points are controlled observations derived from actually replicating the MH370 flight through 17:07 UTC (and not as the legend states "predicted"), and the blue plot points ( diamonds) are the actual MH-370 observations.
Making those assumptions, one is left to explain the difference in BFO values at 16:49 and also at 16:55. Is that the potential for error, or is there some other explanation?
I'd agree that the investigation team (at this point) has no obligation to release error analysis, however I'd ask: Is there any reason withhold that information?