Fear and loathing not required

[InfernoSTi]

A corollary of this is that all empirical "facts" are really hypothesis-laden observations. So rather than say we can't call facts "objective", what I would suggest is that we can't call facts indisputable. (If they aren't objective, then they can't be proven to be wrong.) They are hypotheses, and like all others, have to be tested. All of our knowledge, including empirical observables (facts) is in this sense tentative.

 

To a degree, this is what I was trying to say. The test may be "objective" but the hypotheses ("a supposition or proposed explanation made on the basis of limited evidence as a starting point for further investigation") is inherently "subjective" and we were willing to accept returns that are less than 100% as valid. And our selection of both the hypotheses and how much less than 100% to accept are both subjective choices.

 

Why is an R Square of 0.9 OK and not accept 0.95 as the minimum? Why not 0.5? Generally accepted practice? That is a fancy way of saying belief/faith/enough people thought the same way.

 

Further, as our knowledge grows, we find that we explain things more simply and directly with a new explanation than the old intricate answer or that we now can explain to a higher degree of explanation. This seems to be the "diminishing degree of wrong" that was mentioned earlier except when the new knowledge is a radical departure from the fundamental base previously held to be true, as in my non-euclidean example.

 

One common outcome at that point is some holding onto the old knowledge as if it were a religion rather than accepting the new understanding. Another common outcome you see that some of what we thought before was actually false, even if we can still measure it on the other hand. It wasn't the measurement that was wrong, it was the test that was wrong.

 

Because it is all iterative (understanding of reality -> hypotheses -> measurement equipment -> understanding of reality->repeat), and because we know our understanding of reality will improve (sometimes incrementally sometimes radically), then we must give some amount of credence to the "unknown unknowns" and that amount seems to vary by person subjectively. It isn't the facts that are not objective, but the human selection of what to consider and what to reject that is subjective.

 

I guess some people believe that the selection of what to test is "objective" but it isn't if you have ever tried to create new knowledge. If you are trying to create new knowledge, you are constantly considering new hypotheses, new blends of old tests, old tests with new measuring methods, adding one thing to the "knows" in the hope that you get some new result to follow.

 

But I am going to reread the PMs I got last night and consider that I might need to grow my thinking. Thank God I still love learning and that I would rather grow than hold onto a limited idea.

 

I am very open to being shown in a respectful and civil way, that there is another perspective that I, if I were to consider it seriously, might actually agree with. Learning is a joy…

 

Best,

John

[Jud]

And our selection of both the hypotheses and how much less than 100% to accept are both subjective choices.

 

Why is an R Square of 0.9 OK and not accept 0.95 as the minimum? Why not 0.5? Generally accepted practice? That is a fancy way of saying belief/faith/enough people thought the same way.

 

Best,

John

 

I agree re the subjectivity, or exercise of judgment, in choosing the commonly accepted tests of statistical significance. I would disagree that it is a matter of belief or faith. Rather, it is a matter of just how unlikely the alternative must be in order to allow us to feel subjectively comfortable with our (provisional as always) conclusions.

 

If it were really a matter of faith, then there would be no way to calculate probabilities and thus no way to test for significance.

[Jud]

And just with regards to hi-res, I personally feel that much of the advantage of such material is that it lessens the burden (and unavoidable sonic fingerprint) on the compromised digital filters in most DACs. Well-tuned SRC--to 352.8/384KHz--done in a computer goes a long way towards making Redbook sound like native hi-def material. Yet as you point out, everyone argues about if the extended bandwidth is/is not the important thing and if we can or can not hear it.

 

So many dogs, barking up so many wrong trees. The squirrels are in the tress behind the house.

 

Just an observation. If if SRC to 352 goes a significant distance in making Redbook sound like native hi-def, then extended bandwidth isn't needed or at least it doesn't support that idea. Maybe the filtering or other factors, but even rate converted it has nothing above 20 khz. And I did notice you did not say it made Redbook equivalent to native hi-def.

 

I think you two are agreeing here. Oversampling makes sense in terms of filtering, not in terms of somehow capturing something not present in the original samples.

[InfernoSTi]

I agree re the subjectivity, or exercise of judgment, in choosing the commonly accepted tests of statistical significance. I would disagree that it is a matter of belief or faith. Rather, it is a matter of just how unlikely the alternative must be in order to allow us to feel subjectively comfortable with our (provisional as always) conclusions.

 

If it were really a matter of faith, then there would be no way to calculate probabilities and thus no way to test for significance.

 

Not religious faith, but faith that the calculated probability will fill one with confidence. For some, that is a singular matter, for others it means that many must also feel the same way. It is a feeling, a belief, not an objective point that one can measure. I call that belief "faith" that it is right even if you can calculate it as a number. Perhaps I give too great a power to the word faith…all I mean is that it isn't a fixed amount that all will feel equally confident from. It varies person to person, some needing more statistical evidence, others less.

[ringenesherre]

With most null tests in other Audio areas, the results are more often than not, not a complete null, but the person performing the tests invariably claims that the results are beyond the limits of audibility. How arrogant is that ?

 

I will answer your question "How arrogant is that?". Obviously, arrogance in this situation is primarily a function of how much the result is attenuated (other factors would include the type of the residual signal). For example, if the result of nulling two 24-bit recordings peaks at -140 db, the level of arrogance in assuming that this is inaudible is very close to 0. On the other end of the spectrum, if the result has components at 0db, the level of arrogance is close to infinity.

 

Hope my answer helped

 

Cheers,

Peter

[wgscott]

Much less than this arrogance:

 

Mr. Professor of a non related audio area.

I will NEVER accept I am wrong ...

[sandyk]

QUOTE] Originally Posted by sandyk View Post

 

Mr. Professor of a non related audio area.

I will NEVER accept I am wrong ...

 

Much less than this arrogance:

 

So it's arrogant when you can demonstrate that what you are reporting about .wav files with identical check sums sounding different is 100 % correct ?

Why should I admit that I may be wrong in this area when I am not ?

YOU are the arrogant one who thinks that just because you are a Professor in a non related discipline gives you any credibility in Audio related matters. On the other hand, I now have confirmation from 3 separate qualified EEs, one of whom is well respected as a technical writer and technical journalist ,as well as confirmation from a high profile Recording and Mastering Engineer and also quite a few C.A. members. You , on the other hand have no credibility whatsoever in the area of Audio.

Although in itself, the other confirming reports that I haven't made public are not sufficient in themselves to change the mind of a closed minded sceptic, I am prepared to share those reports with Chris C. in confidence if requested.

 

BTW, typical selective editing of my reply, of the kind I have come to expect from people like yourself.

[wgscott]

I think it is somewhat arrogant that you cannot even entertain the possibility that you could be wrong, let alone state under what conditions you would be willing to accept that your hypothesis has been refuted.

 

I gave you two examples of conditions under which I would be willing to accept that the null hypothesis is wrong.

 

Try it. Just as an exercise in intellectual humility. It doesn't mean you actually have to believe you are wrong. It simply means you have to be able to conceive of the possibility.

[Snowmonkey]

Alex, I'm at a loss to understand why electrical engineers are any more or less qualified to hear differences between audio files than professors of Greek, Biochemistry, Ancient History or Mathematics (or janitors for that matter).

[sandyk]

Alex, I'm at a loss to understand why electrical engineers are any more or less qualified to hear differences between audio files than professors of Greek, Biochemistry, Ancient History or Mathematics (or janitors for that matter).

Snowmonkey.

I didn't say they weren't, but wgscott is setting himself up as someone with superior knowledge in an area where he has no expertise. I doubt that he has even had any formal training in electronics other than what he was taught at college, yet he is dismissing outright reports by highly qualified EEs such as Martin Colloms. At least I have had a previous subjective report of mine confirmed in the Nov. 2008 "Current Mirror" thread in DIYAudio, where even Nelson Self was a participant.

Alex

[crisnee]

Alex,

You really should stop jumping in so quickly to defend yourself and your position; you sound paranoid and really, almost no one, if that many, care about your position. Besides, we all know it. And if you feel you must inform the uninformed, those new to the forum, how many of those do you think will care, never mind be able to hear the differences that you can hear? And how important would those differences be to those few?

 

I thought Bill's original post was excellent and it had nothing to do with you, no matter what you may think. And I would much rather read comments and thoughts about it then to once again have to wade through numerous posts of your same old arguments and the inevitable responses to them.

 

Alex, perhaps you could write a blog explaining your exact position and then just post, See Blog "Alex' Position" every time you have the urge to defend yourself or your position.

 

Chris

[Snowmonkey]

Well I'm not in the habit of defending Prof. Scott, but I think he can reasonably claim some professional expertise in the application of scientific method, which is more what this is about than electronics. Personally, I'm sympathetic to your subjectivist viewpoint in that I don't believe I'm in a position to tell you what you can of can't hear. I live not too far from Sydney and I would love to come and test your assertions some time. However, I have to agree with the Prof. that unless you can conceive of a situation where you would accept you were wrong, the whole exercise becomes pointless.

[sandyk]

Chris

You have already made your position abundantly clear to me on several previous occasions.

You can always use the IGNORE feature.

 

Alex

[crisnee]

Chris

You have already made your position abundantly clear to me on several previous occasions.

You can always use the IGNORE feature.

 

Alex

 

I have? I think you must be confusing me with someone else. I have no position regarding you other than the developing one which is as stated above. It's just enough already re your position on this topic. But I don't want to IGNORE you in general.

 

However if you guarantee to me that you will only post on this particular topic in the future, then I will gladly use the IGNORE function on you.

 

Chris

[sandyk]

Snowmonkey

 

Please check your PMs

 

Alex

[goldsdad]

Martin Colloms

 

That name will soon be in every damn thread!

[christopher3393]

From A.N. Whitehead, Science and the Modern World, p186, published 1925:

 

“In formal logic, a contradiction is the signal of a defeat: but in the evolution of real knowledge it marks the first step in progress towards a victory. This is one great reason for the utmost toleration of variety of opinion. Once and forever, this duty of toleration has been summed up in the words, 'Let both grow together until the harvest.' The failure of Christians to act up to this precept, of the highest authority, is one of the curiosities of religious history. But we have not yet exhausted the discussion of the moral temper required for the pursuit of truth. There are short cuts leading merely to an illusory success. It is easy enough to find a theory, logically harmonious and with important applications in the region of fact, provided that you are content to disregard half your evidence. Every age produces people with clear logical intellects, and with the most praiseworthy grasp of the importance of some sphere of human experience, who have elaborated, or inherited, a scheme of thought which exactly fits those experiences which claim their interest. Such people are apt resolutely to ignore, or to explain away, all evidence which confuses their scheme with contradictory instances. What they cannot fit in is for them nonsense. An unflinching determination to take- the whole evidence into account is the only method of preservation against the fluctuating extremes of fashionable opinion. This advice seems so easy, and is in fact so difficult to follow.

One reason for this difficulty is that we cannot think first and act afterwards. From the moment of birth we are immersed in action, and can only fitfully guide it by taking thought. We have, therefore, in various spheres of experience to adopt those ideas which seem to work within those spheres. It is absolutely necessary to trust to ideas which are generally adequate, even though we know that there are subtleties and distinctions beyond our ken. Also apart from the necessities of action, we cannot even keep before our minds the whole evidence except under the guise of doctrines which are incom*pletely harmonised. We cannot think in terms of an indefinite multiplicity of detail; our evidence can acquire its proper importance only if it comes before us marshalled by general ideas. These ideas we inherit – they form the tradition of our civilisation. Such tradi*tional ideas are never static. They are either fading into meaningless formulae, or are gaining power by the new lights thrown by a more delicate apprehension.

They are transformed by the urge of critical reason, by the vivid evidence of emotional experience, and by the cold certainties of scientific perception. One fact is certain, you cannot keep them still. No generation can merely reproduce its ancestors. You may preserve the life in a flux of form, or preserve the form amid an ebb of life. But you cannot permanently enclose the same life in the same mould.”

[kumakuma]

That name will soon be in every damn thread!

 

Too late. Do a search and see how many hits you get.

[esldude]

Too late. Do a search and see how many hits you get.

 

Well it includes every thread I started in the last two years. So I seem to bring the Martin Colloms out. Suppose I need to quit starting threads.

[InfernoSTi]

Based on an off-line conversation with Dennis, my thoughts, or perhaps my language, have changed a little on this subject (whatever that actually is).

 

First, I believe that it is possible to objectively test for subjective preferences. Where before I was saying the selection of the test was a subjective part of a whole, now I am saying that while the preference is subjective, the test is objective and the results predictable.

 

Second, I am concerned with matters of scale of importance. I don't see enough discussion about how significant a particular test is in the mix of all the accepted tests. This comes from thinking about speaker specifications being +/-3 dB 20-20K Hz yet most untreated listening rooms being +/-20 dB 20-20K Hz. What is more important at this point: getting a "flatter response" from one's speakers or getting a "flatter response" from one's room? I think we need more of an understanding of what are the big things and what are the little things.

 

Third, I think a lot of tests fail to address the timbre of the musical reproduction. I don't see a lot of objective tests showing what sounds "more like a real instrument" A or B. I think this is how we get a lot of the "that sounds better to me" discussion. It isn't that the objective data should be questioned, it is that there are differences in how amplified recordings sound from source to amplification to transduction. In the end, this seems more a preference than a question of accuracy of reproduction. The discussion of how different styles of tweeters can all measure the same but sound different from each other is related to this issue. So I think we need more acknowledgement of timbre in the reproduction of music.

 

In summary: (1) while the preference is subjective, the test is objective and the results predictable; (2) we need more of an understanding of the scale of importance of the test relative to other measures; and (3) we need better testing of timbre in the reproduction of music.

 

Does any of this make sense to anyone else or am I just wandering in the woods? I'd like to hear comments…

 

John

[kumakuma]

Well it includes every thread I started in the last two years. So I seem to bring the Martin Colloms out. Suppose I need to quit starting threads.

 

I doubt that would help.

[esldude]

Based on an off-line conversation with Dennis, my thoughts, or perhaps my language, have changed a little on this subject (whatever that actually is).

 

First, I believe that it is possible to objectively test for subjective preferences. Where before I was saying the selection of the test was a subjective part of a whole, now I am saying that while the preference is subjective, the test is objective and the results predictable.

 

Second, I am concerned with matters of scale of importance. I don't see enough discussion about how significant a particular test is in the mix of all the accepted tests. This comes from thinking about speaker specifications being +/-3 dB 20-20K Hz yet most untreated listening rooms being +/-20 dB 20-20K Hz. What is more important at this point: getting a "flatter response" from one's speakers or getting a "flatter response" from one's room? I think we need more of an understanding of what are the big things and what are the little things.

 

Third, I think a lot of tests fail to address the timbre of the musical reproduction. I don't see a lot of objective tests showing what sounds "more like a real instrument" A or B. I think this is how we get a lot of the "that sounds better to me" discussion. It isn't that the objective data should be questioned, it is that there are differences in how amplified recordings sound from source to amplification to transduction. In the end, this seems more a preference than a question of accuracy of reproduction. The discussion of how different styles of tweeters can all measure the same but sound different from each other is related to this issue. So I think we need more acknowledgement of timbre in the reproduction of music.

 

In summary: (1) while the preference is subjective, the test is objective and the results predictable; (2) we need more of an understanding of the scale of importance of the test relative to other measures; and (3) we need better testing of timbre in the reproduction of music.

 

Does any of this make sense to anyone else or am I just wandering in the woods? I'd like to hear comments…

 

John

 

Excellent post and excellent thinking John.

 

I think those areas you bring up are very pertinent and could bring real benefits if they were investigated more. I feel some of the other minutia even when real is often of rather minor difference in the sound we hear.

[Jim Sylva]

Deleted

[Jim Sylva]

So I think we need more acknowledgement of timbre in the reproduction of music.

 

John

 

I agree. For me, timbre in the reproduction of music is very important. So is just wandering in the woods.

[Paul R]

Hi John - you have to account for intent in the testing. Just as in the case of drug tests being skewed when sponsored and directed by a pharma company with a financial interest in the results, so to are many audio test skewed to favor one or another set of results.

 

This is, unfortunately, just a true on the side of the so called 'objectivists' as it is on all the other sides. Perhaps more so because the loudest adherents believe they already know the answers. This is also a common circumstance, at least in the history of western civilization.