Although the deleted post on DSP-based oversampling contained some glaring errors (most of which were nullified by removing it, and explaining the source of the misconception), it contained one error that might take on a life of its own unless I address it specifically. That error is the assumption that a digital filter affects only frequencies above the cutoff frequency, as if a formula and its implementation in digital hardware could create a physical low-pass analog filter. The fact is that a digital filter rejects the frequencies above the cutoff frequency by calculating the samples representing the DESIRED waveform. The additional precision of a DSP-chip-based filter creates more precise samples, and hence superior sound.
The deleted post was otherwise one of the better layman-oriented explanations of oversampling I've seen. So, I'll revise it and eventually repost it somewhere.