Harshness and digital glare have been part of digital audio for years. All you have to do is listen to a good record (aka vinyl) played through a tube system and you recognize the characteristics of the two different approaches. In many case, the ultimate goal of digital design is to get to the liquidity of an all analog system. For others, an all analog system is perceived as mushy, without the details some people want. You say "tomato", I say "tomato" - never did know how to write that out.
For CDs, digital glare is often associated with the brick wall filter typically implemented at 22.05 KHz (44.1Khz/2). This eliminates the high frequency noise at the Nyquist frequency which comes from the D to A conversion. This filter also causes artifacts in the normal hearing range. Higher sample rates moved the filter to a much higher frequency and allowed for it to be much less steep. This moved the artifacts out of the normal hearing range. That is one of the reasons that DAC designers went to oversampling in their DACs. Most modern DACs do oversampling, although there are a small number of NOS (non-oversampling) DACs out there. High res files were originally designed to help overcome the brick wall filter effects, but with most DACs doing upsampling, that aspect of high res files is less important.
R-2R DACs are a very simple D to A converters, but they do require quite precise resistance values. And, the real magic usually comes not in the ladder, but in the filters that are applied after the ladder conversion. The same is true for DSD conversions. The filters are a key to the final sound.
MSB has been a leader in high end audio for many years. The link below is a description of their R-2R ladder DAC implementation. Obviously, this is a manufacturer hyping its technology, but they clearly believe in their approach and MSB DACs are highly regarded in the industry.
http://www.msbtechnology.com/faq/why-ladder-dacs/DSD does not require the same detailed mathematically conversion that PCM does. It is therefore often viewed as being simpler and more accurate, without any of the artifacts associated with brick wall filters.
Modern DACs, both PCM and DSD, often implement multiple filters that the user can select based on their own hearing. This is an acknowledgement that no D to A process is entirely accurate and that, even if it were, people hear differently.
The interesting thing to me is the emphasis DAC manufacturers and software designers are spending on filter design. Many openly market their filter design and many provide multiple options. The hardware design is still important but their is a current emphasis on software, in the form of filter design, for both PCM and DSD.
With all that said, lets not forget that the main purpose for most people is simply to get music that sounds good to them. And we all hear differently.
