I feel like there's a whole world of information theory that either astronomers or (I suspect) people making arguments on behalf of astronomers are unaware of.

I feel like there's a whole world of information theory that either astronomers or (I suspect) people making arguments on behalf of astronomers are unaware of. Long-duration observations can't be disrupted badly by a quick satellite jaunt across the sky. That's just a noise blip. Less than 0.00001% of the data collected. Isolate it and scrub it. Similarly, wide-field observations have a lot of data to work with to figure out what the field is supposed to look like vs. what it looks like with a satellite-pass event. Isolate them and scrub them. Satellites and deep stellar phenomena are different scales of event at different ranges and speeds. A Fourier transform should make them show up as obviously separable. (... and again, all of this is assuming you have to scrub this data without a-priori knowledge, when the disrupting event is a well-documented passover with clear ephemera. If it is mandated that satellite companies publish the ephemera, which I think is a perfectly reasonable burden upon them, it should be even easier to separate signal and noise). I do see your point regarding (b) though; space telescope science may require a wholly new approach relative to terrestrial science.