Despite Jason mentioning designers will probably disagree with what he is 'claiming' I don't think they will.
I don't in any case.

Firstly PCB design is very important (even more so in digital IMO) and certainly when higher currents are in play (speaker amps)
Also I agree with his feedback assesment (crap happening in the amp being 'nulled' with a counter input voltage at the input stage)
Furthermore I don't think topology/components used is that important by itself.
Accuracy (on a number of aspects) is important to get good results.

I use non overall and local (if any) feedback in my designs myself and even though it is having higher amount of distortion I don't think those levels of distortion are detrimental to the sound if they remain low enough under real world circumstances (isn't the same as on a test bench).

IMO opinion an amp should be tested under 'real loads' with real music and nulling input and output and listening to the result (the nulled) signal can give a good indication of how good an amp is.
test bench results are handy to show some basic and comparative numbers that should be viewed as a whole (and not as individual parameters) and should also be accompanied with some plots to show relations.

It is impossible to test out any speaker amp with all speakers and see which may not be driven well and standard measurements (which could be used for comparing aspects) WON'T show all aspects.
It is NOT that it cannot be measured or quantified or is impossible to measure but there are no 'standards' on how a speaker or headphone should/must behave.
They certainly aren't linear by a long shot.
So... yes .. measurements don't say it all and not because differences cannot be measured BUT because there is no standard to which they should comply and can be compared.
Having a 'standard' speaker load emulation (a selection of inductors, coils and resistors and perhaps even back EMF ?) will not be enough as amp designers could 'cheat' and get that to perform well but react poorly to other loads.

There is really no way to test/compare it all. Even a differential amp cannot show everything because it also reacts to phase differences which may not be detrimental but do show up as an error signal.
Listening to a differential signal under different difficult loads (NOT a digitally recorded dif signal) is a really good way to asses how an amp performs in real life situations with music.

Having more than 24 bits won't enhance resolution in practice simply because of S/N ratio and accuracy/linearity of components the DAC is made of are the limiting factor.
Ladder DACs with real 24 bit linearity simply cannot be created simply by lack of accuracy (tolerance) of parts.
This is easier to do with oversampling as higher frequencies aren't as difficult to make as higher accuracy.

Audio isn't voodoo or an art form it's just that not all designers know what matters the most and some of them (mostly bigger companies) only design to earn a living and make devices that must be able to do more or other things than other brands. Also in order to increase sales.
To me it appears as though Schiit doesn't give a schiit about it but wants to bring well made stuff for 'low end' prices.
They could easily move their schiit into a much higher price range (at slightly higher production costs) and sell it for 10x as much.
In that case they would likely sell 10x less schiit and thus not get richer so I REALLY appreciate their philosophy of good affordable schiit for the audio minded masses.