S/PDIF COAX, implemented per the more-than-20-year-old-specs is galvanically isolated and less noise-prone than USB 2.0 will ever be short of heroic, expensive, needless, measures.
Jitter, measurable, never mind audible, with TOSLINK (never mind ST) for R2R solutions is irrelevant. It might affect DS solutions, but it's out of band for any half-way competent multi-bit PCM implementation. No one relevant slaves the DAC's clock to the S/PDIF clock anyway ... it's invariably, minimally, if imperceptibly, buffered.
In my, now extensive, evaluations (with largely TOTL gear), DSD is different but not better, than PCM, so a bit of a red-herring. It's also a vanishingly small aspect of the audio landscape. Converting PCM to DSD is just pandering to specific DAC implementations, and unnecessary if you do things "properly" in that realm (e.g. PS Audio & Chord).
Do some listening, or some math, post some actual listening impressions and stop spouting ill-informed/miss-understood third-party dogma. The sniping-without-backup is getting really old. Disagreement is fine, but if you can't back it up with specifics it's just noise.
I say this as an engineer. As well as a music lover.