As above. This is a curious view that comes from a common misunderstanding of sampling theory. Mathematically, the results of any of these systems is identical. The information content (from a pure information theory point of view) is identical. What oversampling does is free the designer so they can optimise the implementation in more areas in different ways. This freedom mostly lets them push more work into the digital domain, which with the current balance of technology is the better place.
There is a rather unfortunate idea in the NOS crowd that the results are somehow truer to the source data and that oversampling changes the data.
If we had a perfect implementation of both a NOS and an oversampling DAC, each bandlimited to the Nyquist frequency, the results would be identical. Where oversampling gets you an improvement in sound is a combination of the ability to create a physical reconstruction filter with better characteristics and an ability to craft the analog parts of the DAC with much less sensitivity to all the vagaries of implementation. A one bit element is vastly easier to craft than a 16 bit ladder. But you need the digital logic to make it work. Oversampling is the key to unlocking this.
