I am late to the party finding this article, also late to the party beeing much more an objectivist and rejecting voodoo cables ;-) I guess so I am right here haha
I am not sure about the AI posture here, but the latest paid models are quite usable. I attached a file Claude created for me over more than 75 iterations, hourse of prompting, to create cables for MY system. And yes, I still want to use Neotech wire despite nobody hears the difference, just for my own "jewelry" satisfaction ;-)
Currently, my $100k system runs on $30/100ft 16awg OFC stranded wire and it sounds just as it should haha...but back to this topic, Claude Fable's take on Kunchur's "study" is this...
What the study actually shows
The paper is better than its fans and worse than its headline. Kunchur tested three 2 m single-ended RCA cables — a ~$500 StraightWire-class interconnect with porous PTFE dielectric (S), a ~$50 polyethylene one (M), and a sub-$5 PVC generic (G) — and the solid parts of his data
agree with the skeptics: R and inductance differences are "utterly negligible" (the generic would need to be 11 km long for a 3 dB loss), and the measured frequency response is flat within ±0.03 dB with phase within ±0.06°, even for the generic — responses he himself calls "sonically perfect". That's his own instrumentation confirming the doc's Chapter 12 position on R and FR.
The genuinely interesting findings are exactly where the doc's framework says they should be — dielectric and shielding, not conductor: the PVC generic's capacitance varies ~74% with frequency (32% within the audio band) where the PTFE and PE cables hold nearly constant; propagation speed ranks by dielectric constant exactly as Vp physics predicts; and noise pickup differs measurably, with ITU-R 468-weighted audible-band values of 25.3, 40.6, and 44.5 µV for the three cables. His non-ideal-capacitance decay tails and 4 ns pulse-reflection sequences are real oscilloscope phenomena.
Where it fails
The paper contains
no listening test. Audibility is asserted by juxtaposition with his earlier JAES work and a psychoacoustic bridge that doesn't hold: he argues the ear has "temporal resolution in the microseconds" — but that figure comes from
interaural time difference acuity (comparing arrival between two ears), which says nothing about detecting a microsecond-scale, ~100 dB-down linear smear on a transient in one channel. That conflation is load-bearing and it's wrong. The magnitudes never get compared to any masking threshold: his weighted noise levels were 59–64 dB below a 40.2 mV rms reference — and 40 mV is itself ~30 dB below real line level, putting these artifacts ~90 dB down in practice, measured into a 1 MΩ scope input at 500 MHz bandwidth rather than a real 10–27 kΩ amplifier input that loads and filters RF. And his own data quietly breaks the price-ranks-quality narrative: the $50 cable had the
highest full-bandwidth RF noise (2.6 mV vs the generic's 0.49 mV), and cable S showed the smoothest reflections despite the worst impedance match — "contrary to expectations," attributed ad hoc to its connectors. One sample per "grade," n=1, no statistics. Also worth knowing: IOSR is a pay-to-publish outfit that has appeared on predatory-journal watchlists — this is not the peer review the JAES paper got.
The ASR critique and Kunchur's rebuttal
Amir's central point is correct and is the whole ballgame — measurable differences don't translate to audible improvements, audio signals contain no 4 ns pulses, and no cause-and-effect between the noise and perception was ever established. But he was sloppy on specifics, and Kunchur's rebuttal wins those skirmishes cleanly: the brands
aredocumented (footnote g), the multi-range triggered-averaging scope technique is legitimate (his HDR-photography analogy is apt), and his correction of Amir's T/2ⁿ⁻¹ temporal-resolution claim is right. What the rebuttal never does is engage the magnitude question — instead it's credentialism and invective ("cult leader," "juvenile understanding," "zero journal publications"), and it defends the IOSR paper with the JAES paper's four reviewers. Winning the instrumentation skirmishes while dodging the audibility artillery is not a defense; it's a change of subject. The ************ article, meanwhile, launders "measurable electrical differences exist" into "expensive cables do matter in sound quality" — a claim the paper never demonstrates and its own FR data undercuts.
Verdict
Ironically, the paper is a strong argument for
your build philosophy at a fraction of the price: everything that separated cable S — stable C(f), high v/c, low noise — is a property of
PTFE dielectric plus competent shielding/geometry, not of $500. Your ToT-2 26/22 hits those exact mechanisms (PTFE throughout, quadrupole rejection, channel separation — the paper's own advice, since the generic's noise was correlated between the molded-together left and right channels) for $37 of wire. Kunchur measured real physics, proved the cheap PVC molded cable is electrically sloppy, proved R/L/FR are non-issues, and then let a pop-audio site and his own microsecond-hearing conjecture carry the conclusion further than the data goes.
