James Randi vs. Anjou Pear - once and for all

hi tbg:

we are all subject to two types of errors, namely perceiving what isn't there and failing to perceive what is present. it is difficult to determine when an error is made.

the point is to assume that mistakes will be made and not worry whether a difference that is perceived is true or not.
one makes decisions on the basis of confidence, usually as result of induction. sometimes the confidence is not justified.

risk is the name of the game and unfortunately the occasional dissatisfaction from a component purchase.

hi brizonbovizier:

you are correct on the need for multiple tests. if we were not separated by the atlantic ocean, we could design an experiment to be performed , say 50 times. that way, i could clean your pocket of most of your waging money.

you also raise the question of reliability of perception. the notion that one can be confident of one's perceptions may be open to question. anytime we hear a difference based upon sense perception, we may not always perceive such a difference over multiple repititions. so you could apply your logic to other components as well.

as a practical manner, there is a risk when purchasing components. we may deicde we don't like its presentation after some period of time, evn though we may have auditioned the component for 30 days prior to purchase.

one cannot be certain of one's perceptions. they are a lot like opinions. they are probably true and probably false, and it is difficult to test them.

Components clearly sound different and that is what I buy on. This can be repeated blind. Cables cannot. The likelihood of a particular random outcome can be assessed from the sample size and test regime. For example if you have one listener (you) telling apart two cables in a test on one occasion you have a 50% chance of being right just by chance. You would need to repeat the identification a statistically significant number of times in repeated tests for it to have 99% significance. Noone has ever done this and i doubt you would either. Significance - means a significant deviation from chance distribution given the test conditions.

The connectors are hardly expensive - probably £10 a pair. They break earth last which is why I buy them so I can hot swap without turning components off. The importance of good connection can also be demonstrated repeatably if you put purposefully make a poor connection the sound quality suffers. Good quality connectors also minimize RF issues. Again - all well understood science. I have compared with expensive cables lent by numerous shops and they made no difference. Also I have observed ridiculous claims like skin effect and cable directionality. I used to work in rf engineering testing and i recognize this is all ridiculous and just marketing BS.

>>>Components clearly sound different and that is what I buy on. This can be repeated blind. Cables cannot.<<<

Perhaps you are listening to the wrong cables- I easily hear the difference.

brizonbovizier, all I can say is that we differ greatly on what is science and statistical significance, and how we personally choose components. I think that my listening pleasure would be sharply reduced were I to be driven in making decisions on audio based on limited engineering concepts that you view as fundamental. I think EE has very limited understanding of why things work and uses a fundamentally unscientific "good enough" perspective on circuits, parts, wire, and even what is safe.

What you say about statistical significance is in error. As I said, you are talking about probabilities, not statical significance. To use significance is to make believe that you are talking about meaningful significance when you are not.