Interesting Thesis on IC's

[Steve Eddy]

By the way, when I was responding to the post about the dog and pony show "listening test," I was going to address the part about how 61% of the 39 people chose the Monster cable as their preference. But I had to stop short and go pick up my niece.

Anyway, I grabbed a quarter and flipped it 39 times. It came up heads 23 times, and came up tails only 16 times. In other words, it came up heads 59%.

This illustrates that just because something happens over 50% of the time, it's not necessarily an indication of anything.

But more germane to the other test that Ajani mentions, during those 39 flips, flips 5 through 11 came up heads and flips 19 through 24 came up heads. So that's heads seven times in a row and six times in a row respectively.

Just because you have a 50/50 chance of the coin coming up heads or tails each time you flip it doesn't mean it's going to alternately come up heads, tails, heads, tails, etc.

Some years ago during a similar discussion, I grabbed a quarter and flipped it 10 times, each time willing it to come up heads with my mind.

To my surprise, it came up heads 10 out of 10 times.

Did that prove that my mental powers were causing it to come up heads?

No.

I did nine more trials of 10 flips and the best I was able to get after that was 9 out of 10. And only once. And sometimes it only came up heads twice.

So just because JA got 4 out of 5 and MF 5 out of 5, it doesn't mean there was necessarily anything going on. It could very well just be nothing more than statistical "noise" and the confidence level isn't very high.

That's why you need to do more trials.

If there's nothing going on, the statistics will trend toward 50/50. If there is something going on, it won't.

se

[Ajani]

What exactly do you mean by "no trouble"?

Apparently none of the 39 who participated had any "trouble" stating their preference for A or B.

And what exactly makes them "experts"? Just because they write for an audio magazine?

Consider this and then tell me if you truly think MF at least qualifies as an "expert."

The common stated purpose behind high end audio is to preserve the signal and not alter or otherwise damage it in any way. There is endless marketing literature out there about all the pains taken to do this, using the purest conductors, the finest dielectrics, etc.

Some years ago Harmonic Technology introduced their "CyberLight" cables. While intended to be used between analogue components, they were in actually an optical cable. Electro-optical converters built into each end converted the source's electrical signal into an optical signal and at the other end, from optical back to electrical.

However these converters ultimately perform worse than the cheapest opamps you're likely to find in mass market gear. They had a huge hump in the low frequency response and massive amounts of harmonic and intermodulation distortion.

They mangled the signal so badly that when JA ran measurements on them, he wrote "If this were a review of a conventional product, I would dismiss it as being broken."

He further wrote "I am puzzled that Harmonic Technology, which makes good-sounding, reasonably priced conventional cables, would risk their reputation with something as technically flawed as the CyberLight."

And the real nail in the coffin, "I really don't see how the CyberLight P2A and Wave cables can be recommended."

However this is how the "expert," MF, summed them up in his review:

Harmonic Technology's Light Analog Module Photon Transducer is the most significant single technological breakthrough I have experienced in my career as an audio reviewer. It is immediately superior in every way.

So again, what exactly qualifies him as an "expert" if he finds huge frequency response aberrations and gross amounts of distortion to be "superior in every way"?



Yes, I'm familiar with that test as well.

And I stated at the time that just because the average of everyone who took the test wasn't statistically significant, then JA's and MF's results shouldn't have been dismissed out of hand.



No, you can't combine them like that. That's just as flawed as dismissing them because the average of all participants was no better than chance. You can only rightly consider them individually.



While 5 out of 5 may be statistically significant, the confidence level isn't very high.

So while I don't think that result should have been dismissed, neither do I think it provides any sort of conclusive evidence that there were actual audible differences between the cables.

More trials should have been done in order to improve the confidence level in the event there actually were audible differences.



Again, I don't think their results should have been dismissed the way they were. Demonstrating actual audible differences doesn't require some group of individuals all score high. All it takes is one person.



But it's only through adequate controls and statistics that we can establish actual audible differences with any confidence.

Don't be critical of those who ran the test for dismissing JA's and MF's results out of hand and then turn around and be just as dismissive yourself.

se

So let me get this straight, you don't think Fremer and JA's results of 5/5 and 4/5 respectively should have been dismissed, yet you say things like this:

Sure, there have been many people over the past 30 some odd years who have made such claims. But so far, no one has ever demonstrated this ability under controlled conditions.



But as yet, no one has demonstrated actual audible differences under controlled conditions except when the differences were trivially measurable and within known thresholds of audibility.

If someone has, please point me to it so I can check it out.

It seems to me that you are dismissing the possibility that audible differences exist, yet you acknowledge that with (smaller than your ideal sample size of 10 to 20 trials) Fremer and Atkinson have shown that they can determine differences...

Fremer and Atkinson both claim to hear differences in cables, and the limited tests they've taken so far have done nothing to contradict that claim... Actually the results so far have only strengthened their claims...

It's theoretically possible that Fremer or Atkinson would pass tests of 1 and 5 and then turn around and fail a test of 10 or 20. However, I see the limited tests as providing at least enough evidence to make a reasonable person question whether the "accepted science" that there is no audible differences between cables is correct...

IMO. there are 2 unreasonable stances in audio: 1) Every tweak, mod or dollar thrown at a system makes an audible difference & 2) Because we don't know how to measure something or it hasn't been statistically proven, means it doesn't exist.... I find both positions to be equally ridiculous...

Finally let me ask you this: If Atkinson and Fremer agreed to do tests of 20, under scientific conditions you approved of, what would the results have to be for you to believe that they can hear differences in cables? 15/20? 20/20? What if Atkinson got 20/20 and Fremer got 10/20?

[Ajani]

By the way, when I was responding to the post about the dog and pony show "listening test," I was going to address the part about how 61% of the 39 people chose the Monster cable as their preference. But I had to stop short and go pick up my niece.

Anyway, I grabbed a quarter and flipped it 39 times. It came up heads 23 times, and came up tails only 16 times. In other words, it came up heads 59%.

This illustrates that just because something happens over 50% of the time, it's not necessarily an indication of anything.

But more germane to the other test that Ajani mentions, during those 39 flips, flips 5 through 11 came up heads and flips 19 through 24 came up heads. So that's heads seven times in a row and six times in a row respectively.

Just because you have a 50/50 chance of the coin coming up heads or tails each time you flip it doesn't mean it's going to alternately come up heads, tails, heads, tails, etc.

Some years ago during a similar discussion, I grabbed a quarter and flipped it 10 times, each time willing it to come up heads with my mind.

To my surprise, it came up heads 10 out of 10 times.

Did that prove that my mental powers were causing it to come up heads?

No.

I did nine more trials of 10 flips and the best I was able to get after that was 9 out of 10. And only once. And sometimes it only came up heads twice.

So just because JA got 4 out of 5 and MF 5 out of 5, it doesn't mean there was necessarily anything going on. It could very well just be nothing more than statistical "noise" and the confidence level isn't very high.

That's why you need to do more trials.

If there's nothing going on, the statistics will trend toward 50/50. If there is something going on, it won't.

se

So therefore if the entire group in the test of 5 had gone for 20 trials and had similar results, then Fremer getting 20/20 and Atkinson getting 16/20 could still be dismissed as statistical noise... If it was 100 trials and they got 100/100 and 80/100 respectively, then it could still be dismissed as statistical noise....

Do you start to see the problem many of us have with DBT statistics yet?

[Steve Eddy]

So let me get this straight, you don't think Fremer and JA's results of 5/5 and 4/5 respectively should have been dismissed, yet you say things like this:

Sure, there have been many people over the past 30 some odd years who have made such claims. But so far, no one has ever demonstrated this ability under controlled conditions.

But as yet, no one has demonstrated actual audible differences under controlled conditions except when the differences were trivially measurable and within known thresholds of audibility.

If someone has, please point me to it so I can check it out.

It seems to me that you are dismissing the possibility that audible differences exist...

Not at all.

I'm simply saying that to date, no one has demonstrated that there are. At least not when the measured differences are below known audible thresholds.

...yet you acknowledge that with (smaller than your ideal sample size of 10 to 20 trials) Fremer and Atkinson have shown that they can determine differences...

I acknowledged no such thing.

Neither MF nor JA showed that they could determine differences.

What I said was, their results shouldn't have been dismissed out of hand. By that I mean that they could very well have been due to pure guessing. But five trials is too few to rule that out and that more trials were warranted to find out.

Fremer and Atkinson both claim to hear differences in cables, and the limited tests they've taken so far have done nothing to contradict that claim... Actually the results so far have only strengthened their claims...

The results so far are too ambiguous to strengthen anything.

It's theoretically possible that Fremer or Atkinson would pass tests of 1 and 5 and then turn around and fail a test of 10 or 20. However, I see the limited tests as providing at least enough evidence to make a reasonable person question whether the "accepted science" that there is no audible differences between cables is correct...

The "accepted science" is simply that no one to date has demonstrated to any reasonable level of confidence that there are. That's not the same as saying there are none.

IMO. there are 2 unreasonable stances in audio: 1) Every tweak, mod or dollar thrown at a system makes an audible difference & 2) Because we don't know how to measure something or it hasn't been statistically proven, means it doesn't exist.... I find both positions to be equally ridiculous...

As do I.

Finally let me ask you this: If Atkinson and Fremer agreed to do tests of 20, under scientific conditions you approved of, what would the results have to be for you to believe that they can hear differences in cables? 15/20? 20/20?

I would like to see something on the order of 90%. And preferably I'd like to have a bit more than 20 trials.

What if Atkinson got 20/20 and Fremer got 10/20?

As I said earlier, it only takes one person to demonstrate audible differences. So it wouldn't matter that MF got 20/20.

se

[Steve Eddy]

So therefore if the entire group in the test of 5 had gone for 20 trials and had similar results, then Fremer getting 20/20 and Atkinson getting 16/20 could still be dismissed as statistical noise...

Who said anything about dismissed?

Not me.

Do you start to see the problem many of us have with DBT statistics yet?

What I'm starting to see is that you're reading things into what I've written that aren't there.

Where did you get that I said anything about dismissing such results?

se

[Ajani]

Who said anything about dismissed?

Not me.



What I'm starting to see is that you're reading things into what I've written that aren't there.

Where did you get that I said anything about dismissing such results?

se

Fine, what you said:

So just because JA got 4 out of 5 and MF 5 out of 5, it doesn't mean there was necessarily anything going on. It could very well just be nothing more than statistical "noise" and the confidence level isn't very high.

If it is regarded as statistical noise, then what is it other than dismissed? Perhaps you just dislike the word I use but the point remains that results can easily be disregarded...

[Ajani]

I would like to see something on the order of 90%. And preferably I'd like to have a bit more than 20 trials.



As I said earlier, it only takes one person to demonstrate audible differences.

So back to the question: how many trials to satisfy you then? 30? 100? What if the person only got 85% in 100 trials?

[Steve Eddy]

Fine, what you said:

So just because JA got 4 out of 5 and MF 5 out of 5, it doesn't mean there was necessarily anything going on. It could very well just be nothing more than statistical "noise" and the confidence level isn't very high.

If it is regarded as statistical noise, then what is it other than dismissed? Perhaps you just dislike the word I use but the point remains that results can easily be disregarded...

Nothing is being disregarded. My saying "It could very well just be nothing more than statistical 'noise'" is not a dismissal, but simply the recognition that there can be more than one possibility for the result.

You're not saying that audible differences would be the only possible reason for a 5 out of 5 result are you? If not, then I don't see how you could have read my acknowledging another possibility as disregarding anything.


se

[Steve Eddy]

So back to the question: how many trials to satisfy you then? 30? 100?

I'd like to see about 50 trials at about 90%.

What if the person only got 85% in 100 trials?

I could accept that as being a pretty good indication that there is an audible difference.

se

[Ajani]

Nothing is being disregarded. My saying "It could very well just be nothing more than statistical 'noise'" is not a dismissal, but simply the recognition that there can be more than one possibility for the result.

You're not saying that audible differences would be the only possible reason for a 5 out of 5 result are you? If not, then I don't see how you could have read my acknowledging another possibility as disregarding anything.


se

Ahh I see.... perhaps a rephrase will make it clearer: I regard what you're saying as being that 5/5 is inconclusive... is that a fair assessment?

Inconclusive means you can draw nothing from the results... If you drew nothing from the results, then the results were essentially disregarded or 'dismissed'...

[blackraven]

That was an interesting read. I suppose if all he had to test back in the day was R.S. -vs- Monster...then I may have also reached the same conclusion...lol

I would not knock the Monster M1000i's. I picked up a pair of those and the M900i's on ebay for $30. They are well made and sound great. Certainly no worse than the Audio Quest Corals that came with my Music Hall CD player. I'm using them in my second system now and I am using BJC's in my main system because I perceive the BJC's to have more prominent treble probably because of their low capacitance, but otherwise the bass, and overall sound is the same. I'm not a big monster supporter and would never buy any new gear from them but they do make some good over priced stuff.

[Ajani]

I'd like to see about 50 trials at about 90%.



I could accept that as being a pretty good indication that there is an audible difference.

se

So 50 trials at 90% would be a yes differences exist, and 85% in 100 trials would be less conclusive?

OK then.... So is it possible that even if John Atkinson agreed to do 100 trials and scored 85%, that many DBT fans would still not be satisfied that he can hear differences in cables?

[Steve Eddy]

Ahh I see.... perhaps a rephrase will make it clearer: I regard what you're saying as being that 5/5 is inconclusive... is that a fair assessment?

Yes, that would be a fair assessment.

Inconclusive means you can draw nothing from the results...

No, not nothing. Just nothing conclusive.

If you drew nothing from the results, then the results were essentially disregarded or 'dismissed'...

But I'm not drawing "nothing" from the results.

If the results were to be disregarded and therefore "dismissed," then the whole thing would end with the 5 out of 5, as it did in the original test.

As I said previously, I think the 5 out of 5 is deserving of further investigation. So while the results may have been inconclusive, it doesn't mean I am drawing nothing from them and therefore am not disregarding or dismissing them. They are a starting point.

se

[Steve Eddy]

So 50 trials at 90% would be a yes differences exist, and 85% in 100 trials would be less conclusive?

My statistics is a bit rusty, so I don't know exactly how the two would compare in terms of confidence level.

OK then.... So is it possible that even if John Atkinson agreed to do 100 trials and scored 85%, that many DBT fans would still not be satisfied that he can hear differences in cables?

I can't speak for anyone else.

se

[Ajani]

I can't speak for anyone else.

se

I didn't ask you to... but I think you see my point with DBT and statistics... There are no clear rules about the number of trials required, number of participants required, the makeup of participants (whether experts or the average man or even what exactly makes someone an expert) and the percentage of answers that need to be correct...

So while 90% in 50 trials, with just one participant might satisfy you, it is possible that other persons would not be satisfied with such a test...

[atomicAdam]

Ok -

Let try this.

If we can't measure the difference, if all we can measure is the voltage and amplitude over time - comparing what is in the cable to what is coming out of the speaker - might be the wrong test.

What if we take what comes out of the cable/speaker and compare that to what is on the original source?

It could be a better cable aligns better if you were to overlay the two graphs. It would seem, comparing what going on in the cable and out of the speaker is apples to apples, where what we really want to compare is apples to oranges, and the differences are the changes that we hear.

[Steve Eddy]

If we can't measure the difference...

Can't measure what difference?

...if all we can measure is the voltage and amplitude over time...

In the electrical domain, what else is there but voltage and current versus time?

What if we take what comes out of the cable/speaker and compare that to what is on the original source?

Yeah, I suppose you could do that.

It could be a better cable aligns better if you were to overlay the two graphs.

Could be. Though wouldn't it be much easier to simply look at how the cable itself modifies the signal fed through it?

It would seem, comparing what going on in the cable and out of the speaker is apples to apples, where what we really want to compare is apples to oranges, and the differences are the changes that we hear.

But there's no point in doing that until you first establish that the changes you hear are actually changes which are heard.

In other words, first you need to establish that there are actual audible differences. Then you can go looking for the cause.

Otherwise, if the difference is nothing more than "placebo effect" (I'm using the term very broadly, hence the quotes), then you just end up chasing a phantom.

se

[atomicAdam]

Can't measure what difference?



In the electrical domain, what else is there but voltage and current versus time?



Yeah, I suppose you could do that.



Could be. Though wouldn't it be much easier to simply look at how the cable itself modifies the signal fed through it?



But there's no point in doing that until you first establish that the changes you hear are actually changes which are heard.

In other words, first you need to establish that there are actual audible differences. Then you can go looking for the cause.

Otherwise, if the difference is nothing more than "placebo effect" (I'm using the term very broadly, hence the quotes), then you just end up chasing a phantom.

se

Sound like you've got a grant proposal right there!

[Steve Eddy]

Sound like you've got a grant proposal right there!

Cool!

Got any connections at the NSF?

se

[Mr Peabody]

se, have you personally listened to various brands of cables? If so, have you detected any differences? We want to be sure what playing field you are on.

MF & JA have listened to more gear than we dream of hearing, this gives them some expertise. Another playing field that should be determined is if you are talking about "Average Joe" hearing a difference or some one who is a "audiophile" or at least has experience hearing high end gear. I have no doubt an "Average Joe" could have difficulty. Surprising to me many people can't detect the difference between an average system opposed to a high end system at first listen. When you educate them as to the detail to listen for then it becomes a bit more apparent. The same as when you give some one a guitar who has never played and told them to tune a string, they wouldn't know the first thing about it but when they hear the tuned string or it become in tune then they know the difference or can recognize when it's in tune.

[Poultrygeist]

If it sounds better because you believe it sounds better, then it is better.

I love my Placebo IC's.

[Mr Peabody]

If it sounds better because you believe it sounds better, then it is better.

I love my Placebo IC's.

I haven't tried that brand, where did you get them

I wish the placebo effect did work, I'd convince myself a $299.00 HTIB out performed my gear and then Audiogon here I come to recoop some money.

[atomicAdam]

I haven't tried that brand, where did you get them

I wish the placebo effect did work, I'd convince myself a $299.00 HTIB out performed my gear and then Audiogon here I come to recoop some money.

AudioReview has a classifieds as well.

[Steve Eddy]

se, have you personally listened to various brands of cables?

Yes, including my own (I'm a cable manufacturer and just now updated my signature after getting some clarification from Adam).

If so, have you detected any differences?

Yes, I perceive differences. Though I can't say whether they're do to actual audible differences. I'm human same as everyone else and just as susceptible to human weaknesses such as those which can cause us to perceive differences even when there are none.

But at the end of the day, I don't care whether they're due to actual audible differences. I'm only interested in the pleasure and enjoyment I derive from listening to reproduced music.

MF & JA have listened to more gear than we dream of hearing, this gives them some expertise.

So? It doesn't make them any less human.

I've a friend with nearly 30 years of recording, mixing and mastering experience under his belt. However that didn't prevent him one day from spending over an hour EQing a final mix only to later discover the EQ wasn't patched into the board.

Another playing field that should be determined is if you are talking about "Average Joe" hearing a difference or some one who is a "audiophile" or at least has experience hearing high end gear.

Talking about when I say what exactly?

se

[Steve Eddy]

AudioReview has a classifieds as well.

Shhhhhhhhhhhh! That information's classified.

se