Measurements fanatism:
if good engineered:
- no difference between output stages
- no difference between dac chipsets
- no advantage of filters that have slower roll off of minimum phase
- all differences that are appreciated can be emulated (if difference can be measured)
- Delta sigma has no drawbacks
- tube has no advantage
- discret output has no advantage
All of this is wrong.
you change the reality so that it fits your pseudo science.
Science is modeling the reality.
It's not changing the reality so that it fits it.
Show some ABX results to show if it's wrong or not. I did test proving there is difference between filters in dx3pro and there are measurements supported. Very minor difference. If you can show 95+% accuracy of ABX for different filters along with measurements then we talking. Now you are just talking out of your ass without any support of evidence from controlled listening test nor scientific references.
Show me perfectly engineered output stages. There are literally less than handful strictly transparent amplifier in the world. Most are less than ideal. And you say you prefer somewhat less ideal measured ones makes no sense at all.
Delta sigma does have limitations but its benefits out weights the drawbacks which allowing the performance to be so good. Poorly designed deltasigma will measure like shit and will sound like shit. R2R especially discrete ones are never live up to remotely close. If you go back to pcm1704 and try implement to the full potential it will beat all discrete r2r as well as poorly implemented delta sigma. It doesn't matter how you achieve it, it's the resul that matters.
Measurements do as well have limitations, not that you can't measure it just that the design of an measurement and the interpretation of the measurements are not so perfect. Like the ess imd hump. It doesn't affect on traditional imd test at max level but shows clear modulation in different tests when with full level. Different construction of multitone signal also has affect.
Measurement is not everything, but without measurement it's likely to achieve nothing at all. None of the companies avoid at least using some sort of measurements in the designing phase and/or production. If you say otherwise then make one yourself.
Then tubes have minimal advantage not only because it needs high voltage to extend the linear region, it has very low gain. So in comparable circuits BJT will be better. Same with mosfets. It's not cut and dry that tube is always worse but the result is what matters. Measurements will clearly show the better one. This is not at the limitation of measurements, far from it. Whether you can use tubes to make really good performance is obviously possible. But just you need to work harder.
Discrete output has no advantage if you can make same or better performance using ICs. For the most part, ics don't have better power dissipation ability nor enough biasing current to allow low distortion at low impedance load. But it's not saying it's impossible. Especially for headphones, tpa6120 within a composite feedback loop will beat any discrete design within its designed condition. Some discrete designs are indeed very good like Schiit ragnarok. It's miles ahead the magni in terms of performance but you are essentially using a power amplifier for headphones. How much power does it dissipate?
The thing can go on and on. Just because you don't understand something doesn't mean it doesn't work. Just because you don't feel comfortable about some concepts doesn't mean it's wrong.