Just to recap the discussion on the mailing list:
The star code takes the daylight level to determine what magnitude of objects to show. At that point, a simple bias property can be inserted ('show me the default level +1 mag or default level -2 mag,...). That property can default to 0, but be changed by the airplane based on however the plane computes how dark the cockpit is. If the plane doesn't access the property, the user can always set it (via commandline or property browser, maybe even via a dialog, (?)).
This doesn't look an outrageously difficult addition to the patch (I believe I can even point out where to insert the bias if that is needed) and would basically solve all of my issues and allow for easy future extension (once it is property driven, we can use an arbitrary set of property rules to include all sorts of effects).
How about that?
So I would suggest that increasing the number of stars potentially visible in the sky in no way diminishes the original perceptual model, because there never was one.
It happens to be good in many cases by accident. Whether this was originally intended or not in no way influences the level of realism it displays.