Schiit Happened: The Story of an Improbable Startup (2014)

You're misrepresenting both schools here.

First, the "objective" school consists largely of internet loudmouths who have never done an actual A/B test. They're saying what seems intuitive to them, and imagining it's logical.

Second, most serious subjectivists recognize the value of experiments and testing. Where they disagree with the would-be objectivists is that when their subjective experience disagrees with the test, they suspect the test might be wrong as well as the possibility that their subjective experience might be wrong. The so-called objectivists argue that when the test disagrees with the subjective experience, then the subjective experience must be purely imaginary. This perspective could be easily disproven with a test...

Ok, since I brought up THD as a bad measurement, let's take a look at what, exactly, is being measured. A standard THD test measures harmonics generated from a sine wave at 1khz, into a resistive output load.

So, how much does a 1khz sine wave represent an actual musical signal? And how much does a resistive load represent speaker impedance? The answers are "not at all", and "not at all". Do you see the problem yet? Of course you do.

And that's the problem with amplifier specs. They're measuring what is easy to measure, not what reflects sonic behavior. And it gets worse, because it leads to designing for specs rather than designing for sonics.

The easiest way to get "low distortion" spec-wise in an amp is large amounts of global negative feedback. But the negative feedback loop isn't just the outgoing signal. It also picks up back EMI from the speaker. A speaker is an electric motor, but it's also an electric generator hooked to a spring. Energy put into it is stored by and distorted by the suspension, then fed back into the output of the amplifier in a time and phase delayed manner. Yuck. And unless the output of the amplifier has infinite impedance (which it does not), then that back EMI is getting injected into the global negative feedback loop, doing things an 8 ohm resistor in the THD test doesn't do to the signal. The negative feedback loop is reducing input distortion, but increasing speaker EMI.

At which point the "distorted" impedance-matched output transformer of an old school tube amp starts looking like a much better proposition...

I'd disagree with you on the point of transducer theory. It's well understood and poorly published. In particular, proprietary models are significantly more reliable than what is publicly available (both for old school lumped-element and newer finite-element analysis), and the major manufacturers keep those close to their chest. Even the published models by those engineers are glorified approximations of internal research. The limiting factors are in manufacturing, where DIYers and small shops do not have access to the materials and testing equipment of the big players.

I think the real elephant in the room is that audiophiles spend a lot of money and don't like being told they're wrong. There's a weird psychology in high end audio (both professional and consumer) - sales are made based on the users' preconceived notions about the product/brand, and you make money by playing into those notions instead of trying to subvert them.

PS: based on your username, I think I apprenticed under you a few times.

I think that's a lazy argument. Context matters a great deal, and "blind AB tests" don't mean what you think they mean.

I play guitar, and have recorded/produced numerous albums. When I'm playing, I can easily tell the difference between subtleties like which kind of guitar pick I'm using. But listening back to recordings of my own playing that I made, I sometimes couldn't even tell you which guitar I used.

Now, I don't think you're suggesting that all guitars sound alike here. But as an example of the kind of context distortion of the AB test, that's what it suggests.

I mean, so do dozens of other audio companies:

https://www.head-fi.org/sponsors/