Reality is analogue. Whenever you convert A to D, you lose information. What actually matters is whether your analogue method loses more data than the conversion from A to D.
CDs are 65,536 levels, 44.1 kHz. If you converted that signal to vinyl, you could imagine that the resulting groove would have a 'squareness' or a 'lumpiness' that vinyl made from an analogue recording would not. How big would these 'lumps' be? The answer is 1/65536 of the depth of the groove (0.05 mm) - this gives a 'lump size' of 700 pm, that's marginally larger than the chlorine atoms in the PVC molecules, let alone the complex molecules that make up the structure of the vinyl itself, or indeed the carbon granules that give the material its strength.
So, the best possible vinyl reproduction would be 'lumpier' than the size of the lumps coming from the inaccuracies of CD ADC. If you want a 'best possible' estimate of the digital equivalence of a vinyl recording it would be about 12-bit, 36 kHz.
It only gets worse from there, numerous other effects degrade the performance of vinyl: intermodulation between the left and right channels (vinyl has poor separation of channels, whereas CD faithfully reproduces both channels), and there are minor bleed-through effects from one groove to a neighbouring one, as well as numerous mechanical limitations, such as change in rotation speed of the disc due to frictional effects, etc. Never mind what actually wear does to the groove from contact with the stylus.
Certain niche audiophile vinyl recordings may use different specifications - e.g. deeper grooves, higher rotation rates, together with special needles, etc. but these are not necessarily compatible with more standard equipment.
It's also interesting that people disagree on what the purpose of a good sound system? Is it to faithfully reproduce the sound as it existed in the studio? If so, why are devices such as 'tube amps' so popular among audiophiles - their strong distortion and modulation dashes hopes of faithful reproduction. Perhaps it's because many people want the sound reproduced as they would like to hear it.
Digital can always be improved by adding more bits, or a faster sample rate. But, what are the useful limits?
Current high-end digital converters are 24 bit, 192 kHz. This gives a theoretical dynamic range of 144 dB, and 96 kHz of bandwidth. In fact, this dynamic range is so high, that even the finest quality analogue circuits can't come close to matching it - even the internal analogue circuits in the converter are a significant limiting factor, giving about 120 dB as the limit.
But what about sound sources? Even the best studio microphones can't offer better than 100 dB - in other words, the microphone alone produced 10x more noise than the ADC. Then the preamps add noise.
Then finally at reproduction time you've got the amps and speakers. 100 dB is about the limit for amps, and 90 dB the limit for speakers.
There's no point going any higher that 24/192 because all you'll be recording/reproducing is noise. There are, perhaps, some theoretical benefits in going higher, because it gives more headroom for sophisticated digital processing, but that could be obtained by resampling the data to a faster sample rate.