Steve,
Not sure I agree with time domain being the biggest factor (although I agree it has its importance). I think oversampling actually helps reduce noise from DAC non-linearities and this may be the biggest benefit from oversampling (not the fact that a less aggressive filter can be used to better preserve time domain info). Oversampling is actually a form of dither. You can also think of DAC non-linearities as a form of error similar to jitter (it is bad as it can correlate to audio signals). Oversampling decorrelates the DAC non-linearity errors.
Sorry but I can't explain further as it really requires a deep understanding of time series analysis which can be rather mathematical.
As for 16 bit, it is indeed enough for a recording simply for playback but greater bit depth in the DAC itself can help with better digital filtering/processing.
Not sure I agree with time domain being the biggest factor (although I agree it has its importance). I think oversampling actually helps reduce noise from DAC non-linearities and this may be the biggest benefit from oversampling (not the fact that a less aggressive filter can be used to better preserve time domain info). Oversampling is actually a form of dither. You can also think of DAC non-linearities as a form of error similar to jitter (it is bad as it can correlate to audio signals). Oversampling decorrelates the DAC non-linearity errors.
Sorry but I can't explain further as it really requires a deep understanding of time series analysis which can be rather mathematical.
As for 16 bit, it is indeed enough for a recording simply for playback but greater bit depth in the DAC itself can help with better digital filtering/processing.