I would personally agree that "there's no such thing as agreeable distortion or noise"....
However, there are absolutely types of both that are
less disagreeable to most people than others.
For example, assuming you have a noise floor that is low but still slightly audible, for the same S/N, many people find "analog hiss" like tape hiss much less annoying than digital quantization noise.
In some situations this may even extend to unequal amounts... for example, many tape fans cheerfully tolerate high levels of "harmless tape hiss", to avoid far lower, and barely audible, levels of digital artifacts that they find far more annoying.
An analogous situation exists with the latest h.265 video compression CODEC.
For a given
measured "loss of quality" the new CODEC substitutes a rather innocuous softening for the unpleasant blocky artifacts produced by the previous h.264.
(The new CODEC does actually deliver better measured quality for a given level of compression... but, even when delivering the
same measured level of
quality, the flaws it contains are less
visually unpleasant to most viewers.)
On the notion of manufacturing tolerances...
I understand what you're saying, and you are generally correct, but the details can sometimes be more complex, and the comparison between R2R and D-S still applies.
The thing you need to remember is that precision carefully matched parts are critical in some situations - yet deliver no benefit whatsoever in others.
For example, in a typical analog amplifier, certain pairs of components must be carefully matched, especially if you want good performance, while others perform perfectly even with considerable variation.
(It might be a nice idea to carefully sterilize your garden tools before planting flowers - but it offers no actual benefit at all.)
From a design and manufacturing perspective an R2R DAC
requires precision matched components in order to deliver reasonably good performance.
Specifically, precision carefully matched ladder resistors are
necessary in an R2R DAC.
Unless you use expensive, high precision, carefully matched resistors in the ladder network, it's performance will be audibly and measurably poor.
So, rather than reflecting "better manufacturing quality" they actually do represent "the cheapest part that will actually do the job even reasonably well".
(In engineering terms this is considered to be one of the drawbacks of the R2R topology... that it only performs at all well if you use expensive and carefully chosen parts.)
In contrast, a D-S DAC does NOT rely on a set of carefully matched resistors to deliver its linearity... so it can deliver similar or better performance
without the requirement for those precision components.
This was in fact one of the major factors motivating the development of D-S DACs....
(And it's why D-S DACs have almost entirely replaced the older R2R topology for almost all audio applications.)
I point this out because the manufacturers of many R2R DACs are quite proud of the fact that they use expensive, precise, carefully matched resistors.
And they point this out in their marketing literature to suggest that their product is "a quality product" because it uses "expensive carefully chosen precision parts".
However, in engineering terms, what it suggests is that, due to inherent design issues, the R2R topology forces the manufacturer to use unnecessarily expensive components.
I've seen many ads that infer, or claim outright, that a D-S DAC that uses less precise components will necessarily not perform as well - which is simply untrue.
However, in the audiophile world, and a few other niche markets, people are too readily willing to believe that "if it's too cheap they must have cut an important corner somewhere"....
A reasonable analogy would be if my economy car required expensive aviation fuel - in order to deliver the same performance that its competitors delivered with plain old regular gas.
(And if the manufacturer were to advertise that "it must be faster and more powerful - because it burns more expensive higher quality fuel".)
Very good point , I had discounted shall we say agreeable distortion . Harmonic distortion in tube amplifiers would be another example . Not convinced on agreeable noise since I am a black background type but I would be open to consider noise as agreeable coloration perhaps .
Tolerance I was referring to was in fact the sourcing and testing of suitably matched resistors as opposed to manufacturing of the unit as a whole . I have some personal experience of this and know it is difficult to find manufacturers that use tight enough tolerance in the first place and mind numbingly tedious to test and match them (resistors not manufacturers).
I am well aware of the roll off character in nos dacs but would hesitate to say 20khz is audible certainly not in my case .