Reply to IXP
IXP objects strongly to one statement of mine; he quotes me claiming that "Wellfed's protocol" cannot yield results because I claim that no human test subject can refine and reify "differences" per Wellfed' protocol, and then adds his objections as follows:
>
This is blatantly false.
Assuming the GSIC does something to the disc, and assuming that you do not know what it does, you cannot possibly know that the proposed protocol could not be passed.
Let us say that the GSIC superposes a 1000 Hz steady tone at the maximum amplitude representable by the binary coding on the disc. You have asserted that WellFed, indeed that no human being, could not possibly distinguish a treated from an unteated disc 10 times out of 10. This is simply absurd.
<
I appreciate IXP's objections (and I'll overlook his example, which I *suspect* might be so far out of context here as to be a strawman argument.) The sticking point is, as I see it, "what is Wellfed's protocol"?
Not each and every one of us here is on exactly the same page, and I see that my concept of it differs from IXP's and others' concept.
Wellfed postulates (my summary) "real, subtle differences that improve a CD sound"; then he provides a test process.
IXP et al. postulate (my summary) that "differences are irrespective of quantity or quality. We are concerned with the most utterly reduced ultimate meaning of non-identity. In effect, if ONE BIT is different -- or ALL bits, a difference exists".
The difficulty I have in leaping from "Wellfed" to "IXP" is that I don't conceive of "audible differences that a critical listener can accurately detect" as being the black box under test, when they are defined the way IXP might seem to me to define them.
I don't see an agreement, anywhere, of what constitutes "differences". We all seem to have our own idea of what they may be; some haven't analyzed them but propose only a word: "differences". This symbol is intended by them to represent a Platonic archetype, and they do not feel that they need to know what it is; only that it exists.
In the reduced view, a difference is a very simple "archetypal" construct; i. e., a lack of identity between two objects. It is ultimately Platonic and theoretical: a semantic construct represented by the symbolic string of characters "differences".
We audio engineers and sound enthusiasts have a great deal of difficulty accepting that, because of the fact that "musical listening" is such a complicated process, as I've explained.
As I see it, IXP wants to test an archetypal non-identity Platonic difference which actually could be -- fully reduced -- "1 bit difference" between the content of a treated disk and an untreated one.
Remember: I'm doing ULTIMATE REDUCTIONISM, as I believe others are also doing. Whether we are doing it the same way is open to argument.
But unfortunately for Platonists, the CD medium has nothing to do with archetypal absolutes. Compact disk error correction uses a complicated methodology that places four redundant copies of the data onto the CD, in linear quadrature. Using Reed-Solomon error correction algorithm, you produce a data stream that is converted into "an analogue of the quantized steps", based on that algorithm's subroutines that check for data-redundancy, loss of digital information, or large non-coherent digital errors between successive words, making some kind of best-guess interpolation if not all four "bunches" of digital data in the buffer, per one rotation of the disk, are equivalent.
Since the data errors that accumulate in each playing are also non-identical, the Reed-Solomon error correction algorithm process produces a unique solution -- actually, digital-word by digital-word, it varies within ranges defined by the algorithm to correct uniquely each quantized step of each sample non-repeatably -- thus, output subtly differs EACH time a given disk is played. By now, the Platonic non-identity is phenomenologically in existence, milliseconds before anybody hears anything, or even if the CD player audio output is not connected to anything.
We have moved far away from the possible theoretical existence of "ultimate archetypal Platonic identities", in comparing any two plays of ONE disk, let alone when comparing two disks.
In effect, you get a slightly different result when you play any one disk, compared to any other time you play it.
Reed-Solomon error correction introduces unremovable uncertainties that nullify postulates of theoretical identitical repeatable results, thus *guaranteeing*, at some level, non-identity in any playback, compared to any other playback.
Don't forget that "compact disks" are, when examined microscopically, full of flaws. Each "set" of flaws is unique, disk to disk.
I own many, many "marginal" disks which have errors that cause unrepeatable variations, play after play. Sometimes you get "a tick here, a tick there" and never in the same order.
How far do we go to reduce to "irreducible differences"?
Furthermore, no two copies of a given title are actually identical, bit-by-bit.
Irrespective of any human observer, on the level of molecules and within the dynamic range of the sound on the disk, there are measurable differences that cause a defined existence of "non-identity" in any comparison of (a) two successive playbacks of a given CD track, even on one machine; and (b) a controlled playback comparison of two individual copies of a CD title.
There is no actual, *observable in matter*, Platonic ideal of identity, comparing any one playing of any one CD to another, or comparing any experiment playing of similar musical passages of different copies of the same title.
Repeating more explicitly: it is actually true that if you analyze the data stream applied to the CD player lowpass filter, due to variations and errors in photodiode response, jitter and nonrepeatable vibration sequences unique to each individual playback experiment, etc. , that no two playings of one disk are actually alike, and thereby non-reducibly identical! Furthermore, some "marginal" disks have very distinct audible, unpredictable error patterns.
That is the sticking point: a conflict between elegant theory, and what happens in reality.
I can hear the objections that will immediate follow, so I offer them below, leaving out cries of "BS, PT! You are a detestible obscurantist!"
"We don't agree with your 'archetypal non-identity' definition";
"We don't CARE about variables that occur each time a disk is played";
"We don't care if the earth's spin rate varies slightly..."
On and on.
Where do we all agree "to draw the line" and define what any human test subject tries to verify in his claim? What is a "difference"?
Just because Wellfed has done a terrible job in trying to describe a difference, doesn't mean that in a good intellectual examination of a proper protocol, we HAVE to use his vaguely worded, imprecise, and incomplete description.
There are actually TWO possible ways to EVADE here: either use Wellfed's vague generality, or use a Platonic ideal symbolic term not represented by any potentially real event that could occur during an actual test. By following either arguably false premise with exactitude, you end up doing TWO tests, which both fail.
To use a quantum analogy, I've tried to show the existence of an actual classical physical uncertainty principle that prevents two successive playbacks of even one disk from being Platonically identical.
Human test subjects have variations in ability to apply their own unique skills of pattern-recognition, which has an uncertainty principle, to wit: no one test subject can "hear", with robotic non-human reliabilty, an identically repeatable sequence. ("Hearing" being defined as the total cognitive act of perceiving sound waves.)
Unfortunately, in the disastrous collision of an audio engineer with philosophical speculators reducing what is at least arguably a Platonic archetype (the black-box symbolic concept of "differences"), we have a profound cultural - logical - practical misalignment.
Furthermore, academic neurologists, neuropsychologists, and digital audio sound equipment designers will ALSO have profound difficulties, I'd argue, in accepting IXP's "elegant" premise.
I am having difficulties abandoning my life's training and work, and understanding of digital sound reproduction, to move over to your ultimate reductionist syllogistic simplification.
You also see this sort of "diving into complexities" offered AS A RUSE by True Believers, who try to EVADE. They do it all the time by constantly trying to change the rules... claiming pseudoscientific variables of dubious reality... baffling with BS, and so forth.
When you perceive me stopping to examine details, your understandable and permissable skepticism raises terrible red flags.
"PianoTeacher pretends that he is a Skeptical Rationalist, but he is actually an undisciplined pseudoscientist, surely a crank, and possibly a fraud offering evasions."
Have I put that succinctly? I hope so.
In fact, it is actually true that I may be deluded and that though I *think* I'm right, and can propose an argument to try to support my position, I am incapable of determining my own validity; that is up to external parties.
It is arguable that, however, I'm right. I may not have given, in this post, the best possible argument, however.
I do appreciate your "red flags", and since my life has involved hooking up real machines and examining what they really do, and trying to figure out what I can really perceive, versus subtleties that I can't see but which I CAN measure with devices -- correcting for errors and noise -- I am not trained to reduce this kind of examination as far as you are trained, purely theoretically and semantically, to reduce it.
I work in "a different department". I am down the hall, over in the engineering lab. You are upstairs, in a philosophy department seminar.
I apologize for my logical shortcomings here, if you are correct and that I *indeed* have shortcomings. Whatever; your "logical shortcomings" are my "toolbox".
As you can see, we have created another subdivision amongst skeptics:
Skeptical Reductionists examining irreducible archetypes, versus Skeptical Practical Technicians.
I guess that goes about as far as my mind can extend, at present, to analyze IXP's objections and answer them. I feel I am in one group of thinkers with respect to this task, versus the other group of critical thinkers.
As such, if you are right in asserting that I'm wrong, and I cannot falsify you, then you prove your assertions. I very much doubt that my explanations above have falsified anything to *your* satisfaction.
And it may follow that, assuming it is a correct and not false premise proposed by IXP and other critics of mine, that I am therefore unqualified to create a protocol for testing "Wellfed's claims". A true scientist must allow for correction of his own errors and neutralization of his own biases.
It seems to me then, trying to be critically detached from my own ego, absolutely possible to argue that (a) PianoTeacher is fundamentally wrong; and (b) his input should be ignored by wiser logicians.
I could *try* to falsify that... but I don't know that I can without exerting more effort, intellectually, than I have ever done before in all my years in college philosophy and logic classes. And it has been a LONG time since I've tried to argue with such sophisticated and intelligent critics!
So, you may be right.
I may stipulate that, and also prefer to decide that theoretical and fundamental logical objections I might not have answered to your satisfaction up to now, clash in my mind with practicality.
PianoTeacher
