Tír Na HiFi

jkeny wrote:Tony, so that's what your expectation bias sounds like?
What's a TZ sun, btw?

Nige, I doubt it's to do with impedance

mjock3 wrote:Thanks for the reply Tony. Nice mini review!

Well Nige your cable faired pretty well, and looks great. Must feel good to have an attempt do so well. I tried making a USB cable once and it sounded just horrid. Not to mention was a royal PITA to try and solder. Guess I am getting a bit old.

If you do thorough testing to establish there's no physical effect, be it the claimed effect or another effect that might not have occurred to anyone, then fine.

Most people on a forum who question audiophile anecdotes do no such thing though. It's straight to synthesising theory and denial.

If people had the humility to use the words probable and improbable I would take less exception, but many don't.

The difficulty is what passes for 'thorough' is far from thorough enough, both from the alpha nerds and from the fairy-foo-folk.

The calls for ABX tests are little more than pointless dick-waving by those who want to appear 'sciencier than thou'. Those few who actually understand the mathematics involved in this are working stochastically rather than deterministically, because the audio engineering world is starting to wake up to the notion that all this listening stuff might have something to do with signal detection theory, rather than some 1950s mechanistic view of of how the brain processes sound.

http://lsbaudio.com/publications/AES127_ABX.pdf

If you work to this model, however, several big things fall out of the mix. First, if you run a test with less than about 50 test subjects, each one being tested the statistically significant number of times to achieve 95% confidence, your results are rendered invalid due to the potential for 'false alarms' (Type II errors in standard stats models). Then, if you have a group of test subjects anywhere between 50-200, the only way to eliminate a strong bias toward false alarm is to crunch the numbers using Beyesian stats methods, which has happened precisely never. Third, this has been well known in practically all branches of sensory science since the 1920s, and even the secret military knowledge bit with respect to ASDIC was declassified before stereo hit the streets.

So, we don't know what tweaks work from a scientific setting, because the science used to test is fundamentally wrong. As it stands, it is designed to flood the results with false negatives, and the people who hold to this are too rigid to see the science has moved out of the 1950s.

Ultimately, there's a move to replace bad test models with even worse test models. The most random sighted, non-level-matched, non-comparison listening test succeeds on the 'even a stopped clock tells the right time twice a day' standard, but that doesn't mean it should be replaced with a clock that always tells the wrong time.

jkeny wrote: Just thought I'd bring people's attention to it