Win $10,000!
Nov 4, 2002 at 7:02 AM Post #2 of 16
24 correct responses in a row? Sorry, that's pretty damn hard to do, especially considering the relatively tight conditions of the test. It would be pretty difficult to win this, which is why the guy is making the offer.
 
Nov 4, 2002 at 7:35 AM Post #3 of 16
yeah, i agree. i know he's trying to prove a point... but i think it would be more fair if he changed the amount you have to guess. he should have the ratio of correct/wrong = the statistical amount that indicates a non-random situation. sorry, forgot the term... but there's a statistical formula for figuring out whether a certain outcome was due to chance or is a real value.

but anyway... still... one would think that the way some people say that there's a "huge" difference between this and that amp would seem to suggest that 100% correct isn't that hard. he does let you choose which amp you will use anyway.

hmm... you know, i don't have $10,000. but i'd be willing to do something similar for a cable test... let's say that we both put up $100... the experiment will be conducted in my studio, under very controlled conditions. except, rather than 100%, i would do, say 75% correct, since 50-50 would be statistical random chance. if anyone has the courage, you can email me. just for fun
very_evil_smiley.gif


heh he... i wonder how many "experiments" he has already finished... but yes, i think 100% correct for this experiment is unfair.... one does make mistakes, no matter how good one is.
 
Nov 4, 2002 at 2:45 PM Post #5 of 16
Quote:

Originally posted by andrzejpw
Question: is there that large of a difference in car audio amps?


Yep. The cheapo ones are just a bunch of IC's, whereas the decent ones are actual transistorised amplifiers. There's probably MORE variation than in Headphone amps, just because of the much higher loads involved.
 
Nov 4, 2002 at 2:58 PM Post #6 of 16
He was suggesting doing the test in a home system -- did that include home amps, or was it still car audio amps?

Someone should post this on AA -- I bet there are a lot of guys who would give it a try.

kerely
 
Nov 4, 2002 at 6:54 PM Post #9 of 16
A set of amps was tested in the finnish 'Hifi' magazine under blind conditions in identically labelled plywood boxes in which even had identical weights (the units has been matched with additional weights). Levels were matched.

Switching was done with an electrically "invisible" (as much as possible) switcher.

The listeners not only been were to discern the amps from each other repeatedly over, but the listeners ratings of the amps correlated statistically strongly with the measured quality (and to most degree price) of the amps.

These amps measure very similarly in terms of most commonly used amp measurement characteristics (output power peak/rms/reactive load/8/4/2ohm, frequency response, 15kHz attenuation with 5 kOhm input impedance, signal/noise ratio, channel balance, output impedance, harmonic distortion, sensitivity, power consumption, etc).

Maybe he should do some reading before repeating tests done by others? Oh, it's not in English? Not my problem, maybe he should understand that knowledge exists in other languages as well (there are very interesting German tests as well in Stereoplay and Audio magazine).

Getting 24/24 right borders on the very limits of successful and experimentally useful number of repeat tests in psychoacoustic testing. Doing such a thing with ABC testing with just noticeable differences is often VERY difficult, even if you can get 10/10 correct identifications 100 times in a row in consecutive days.

But 24 out of 24 is hard in one go - even it the difference is there and can be correctly proven with another kind of set up.

The test proposer either knows this and is deliberately using it against the test subjects (tester bias) or then he is a total neophyte in psychoacoustic testing and shows that his test setup isn't worth a lot in terms of scientific proof. In either case, I think his test setup is invalid and shows bias which totally destroys the credibility of the whole testing scenario. If you set out to test something does not exists, you have already shown your bias and made your whole test methodology suspect.

A scientific test is set up to test whether something exists. It can't be set up to prove that something does not exist or it becomes a null test.

Furthermore, I think he is delusional in trying to prove that there is no audible difference. You can't prove such a thing, except in some statistical certainty under some limited conditions for some particular test subjects for some particular test signal and test gear.

If he thinks otherwise he needs to go back to his philosophy of science books and read up on epistemology.

I'm personally sick and tired of argumenting with such folks as they should clearly stick to engineering on which they may know a thing or two.

But psychological testing is the domain of human sciences, not engineering and you can't use engineering logic, test setups and definitely not engineering attitude to do it in scientifically correct way. Humans are not (known) machines.

If I had a dime for every such "I bet I can prove it to you that X doesn't exist" argument that I've tried to rebut, I'd be one hell of a rich person...

Ah well, I got excited again... don't take offense about what I wrote... other than the clearly unscientific attitude (and somewhat questionably selected methodology) the person in question is probably just trying to do good. Ten points for that!

regards,
Halcyon
 
Nov 4, 2002 at 7:04 PM Post #10 of 16
Quote:

Originally posted by halcyon
All of the amps tested in the finnish 'Hifi' magazine are tested under blind conditions in identically labelled plywood boxes in which even the weight of the units has been matched with additional weights.


I was under the impression that this 'plywood boxes' test method was only used once, the point being that the listeners could take the amplifiers to their homes and listen to them at their leisure.

Is this really their regular practice nowadays?


Regards,


L.
 
Nov 4, 2002 at 7:46 PM Post #11 of 16
Leporello,

Ah, I wrote in misleading manner. At least the amps are listened to under blind conditions in Mr. Eräpuu's auditions. I think other people listen to at home without the boxes (i.e. non-blind). This is always indicated in the test.

So the several people listening under blind conditions was a special case, I'll modify my post to indicate this correctly.

In this regard you are right that not all the people listen to all the amps under blind conditions in 'Hifi'.

Good correction, thanks.
 
Nov 5, 2002 at 4:45 AM Post #12 of 16
well, he does state that there IS A DIFFERENCE between amps. he says clearly he does not question the fact that there are measurable differences. the problem he's asking is, "Are these differences audible." now... he pointed out that many people say there is a "night-and-day" difference between this and that amp... so, based on the fact that these people believe that there is such a huge difference, it should not be hard to have a 100% rate, comparing amps.

now, the problem is... most people criticizing this experiment do doubt that there is a "night-and-day" difference. and that's why i think you think it's hard to get 100% correct... and is also why I personally think this experiment is also misleading.

so, i guess i agree with you that this experiment is not ideal in proving his point, but we are assuming that differences between amps are going to be small.

but there are people out there that swear that one amp sounds completely different from another... and those are the people that should take his wager.
 
Nov 5, 2002 at 6:13 PM Post #13 of 16
Orpheus,

good points, but human senses don't work like that, they adapt.

100% hit rate with 6 repeats might turn into 50% hit rate with 30 repeats. Does this mean that the first 6 were merely right guesses. Not necessarily. If the person can perform the same 6/6 right on consecutive days again and again it can be shown, that they were not mere guesses.

Even REALLY BIG differences that any person can spot hundreds of times correctly in tests with small repeat rates of, will get get lost in longer tests if the test is done so that sensory adaptation kicks in.

This is why a person who does not understand the neurology of perception should not try to set up listening tests to prove anything (scientific).

The tests can be tailored to prove almost anything, if done incorrectly.

As a case in point, take a look at the below linked picture and tell me what you see:

http://www-bcs.mit.edu/people/adelso..._illusion.html

That image uses sensory adaptation of ganglions and neurologically still unknown effect of blurred contours to fool your vision. It will trick almost 100% of people, due to the way it uses our sensory path to trick us.

Similarly I can build a test which will mask even really audible differences statistically just by using some psychoacoustic tricks and little tidbits from neurology of senses. But the same test done properly will reveal the differences even to a total bronze ear in a statistically valid manner.

His test setup is just invalid and he has clear tester bias, which renders his whole setup suspect. Not good science, but attitude masked as scientific proof.

Best regards,
Halcyon

PS Even caps have audible sound signatures. This has also been proven with measurements in Electronics World (latest four issues has really nice harmonic distortion graphs for many different types of capacitors).
 
Nov 6, 2002 at 1:48 AM Post #15 of 16
I'm not sure I understand the test. Could someone please reiterate it in 30 words or less?
 
You must log in or register to reply here.

Users who are viewing this thread

Total: 2 (members: 0, guests: 2)
Back
Top