How big is your matrix - how we bias or enjoyment.


Looking at cables all over again provided some fascinating insight into how people use information. The site that most surprised me, or made me aware of our weakness to bad data, was; http://www.roger-russell.com/wire/wire.htm

If you go read this post, it is most interesting in that the author is so sure of himself, and his data set, that he omits the ability to even question his authority. That being, zip cord is all you need in the proper AWG size for distance and don’t let it get any corrosion on the wire. He goes on to point out “other” peoples experiences with the famous black-box wire comparing devices (we had one, too, but we found out it was so colored that it masked the wire being compared!) to support his view. This gentleman is NOT a development engineer where open minded and constant re expression of data is necessary.

A perfect wire shares the same send and return path (impossible in reality) so it has no inductance (fields cancel), is in a vacuum so it has no capacitance, and it has no tilt (nonlinear attenuation or attenuation). I see no wire close to this ideal and the longer they get the less ideal they get. Physics says wire does make a difference, and is a series of compromises to arrive at a good “sound”. As time moves on, we adapt to physics to come closer to an ideal design.

How do we get wrapped up in this blind-sided “knowledge” acquisition? First, we, at some point, set a bias in our knowledge base. The data is true, but simply not the universe of all the information out there. Once we set our feeling of authority (we often use knowledge as just that, to form out tribe if you will) we further bias our data set by finding support for our side and that sets our directional bias. This cycle repeats until data can simply become a belief system and not even real data or knowledge at all and thus, restricting the ability to enjoy the remarkable things in the world around us. Stereo equipment being on that list.

I call this a serial approach to data acquisition. The real data on any subject is huge. We set a bias to cover but a small part of it (zip cord is all there is in the right size!) and serially hit only those data sets that support our bias. If this was a ski slope, gravity is your bias, and the moguls (those snowy bumps) are your data points to get over to arrive at your destination, to the bottom of the hill.

Data is not serial in nature at all. Data is a matrix, and even 3D matrixes of data. The slope matrix is flat (no bias) and to have true knowledge, every mogul has top be experienced. How do you do that? With Audio, you probably can’t. But to know you should allows a more open experience and enjoyment in this hobby.

Let’s have some fun with this. There was a time when only American cars were supreme in the 1950’s to 1970’s and two or three generation felt this way. Then, the Japanese cam along and drove (pun intended) another several generation to buy and support ONLY Japanese cars. Who was right? Actually, neither was right. Right is a constantly CHANGING argument. Let’s buy a car in 2000. ANY car is a vastly superior car than in the past, so either the old American iron followers, or the Japanese car followers, wouldn’t know to look at ALL cars marketed. Their “knowledge” base, which might have been roughly right at the time, is now worthless. They serially biased their data set, and some still are!

The Japanese, and others that followed them, and eventually the Americans, too, learned statistical process control or SPC for short. We taught the Japanese how to do manufacturing after the war but were too stubborn to listen to our own teachers. What SPC does, is allow the large data sets that seem unrelated to be all considered to arrive at a very probable answer, or how to do things with the highest odds of success. In short, it is a way to cheat the system as it were by NOT having to ski over every mogul on the ski slope, but enough to form an answer with a high degree of probability. Their data set was a MATRIX of information that was looked at in ALL directions. Bias was eliminated and better products produced with lower material cost and better designs. Some would say materials were even of less quality (a real cost on every widget) but with better overall design (cost is amortize over the life of all the widgets) overcoming the cheaper materials. But they only had to be a good car with fewer on the road problems than the American iron over a comparable time span at the time, and they were. They met success with a procedural removal of bias.

What does this have to do with Hi-End Audio? EVERYTHING! We are all subject to serial bias in this hobby like I have never experienced anywhere else. We like solid state amps, so we only buy and listen to them. Why? Because a tube amp didn’t sound good…sometime ago, anyway. Well, if we think we like solid state amps, it seems you would continue to extensively listen to, and evaluate, tube units. Why? Because if you don’t, you are serially biasing your data set to omit what might be BETTER than your limited experience of tube amplifiers. The MATRIX is living and expanding all the time. It is us that sit there and do nothing to experience it. We tuck our heads in and ski down the bias that creates our slope.

Getting back to CABLE, I have the best audience in the world. Two people who do not care about stereo at all. They don’t understand it, or know what it is even doing. My wife and stepson could not be any more different. My wife is a decades long choir singer in a large city choir that supports a city symphony. My stepson is an MP3 generation that isn’t even aware that there are two speakers on his head when he uses headphones. Music just comes out. Neither “knows” you can’t tell the difference in wire if it’s the proper AWG size! My own “knowledge” base was thirty years old (and was mostly zip cord based!). I was literally OFF the MATRIX, or my matrix was so small and unpopulated with new data to be essentially worthless.

I had them both listen to several sets of speaker cable (three that I made from scratch) to the tune of $30,000 bucks worth of stuff. I asked them what they would like to listen to given the choice that I turn the stereo on, walk away, and let them be. Both participants selected wire, and did so easily, based on the sound. I did this twice and they kept picking their “sound”. My stepson thought it was all the same cable and I was “adjusting” it somehow! He thinks it’s weird to have all that money in a stereo and no tone controls. Later, I did this with interconnects with just my wife, and she finished that with a favorite. By now, they both know I’m changing just the cables. When she turned to head upstairs she said, “I hope I saved you money”. Nope, she picked the most expensive XLR interconnect I had to audition.

How can two people clearly hear differences in cable that can’t be heard? I did not say rate them, so much as pick what you like. There is no right in audio. I know, you hate me saying that, but there isn’t. I have two different solid state amps. One sounds better on some material than the other does. I’m forced to decide which is better based on my “knowledge” of listening tests. Can I ever be right? No, I can’t. The reason is that the SOURCE MATERIAL is highly subjective in quality. If I only listen to rock (heaven help me), the compressed and synthesized material is hard to accurately judge and would bias the answer to the softer sounding amp. Or even simpler, my data is biased by my musical tastes (what I buy) and the quality of those recording. I may have mostly good recording, or mostly bad. That would influence what I think the amps abilities are. If all my source materials were real poor, I’d be better off with the cheaper amps. I already have the music, so a better amp won’t change that.

Go to; http://www.thecableco.com/Catalog/Speaker-Cables
Look at all the vendors of speaker cable out there, and this isn’t all of them. I bought cable on all the moguls I could ski over, but not nearly all that would make a statistical likelihood I picked the “best” cable to my ears. To compare cables at all, I had to use a given dealer who would let me walk away with $30,000 of cable and compare them at home. Is price a factor? No, it isn’t. And not for reasons we all like to think. Some manufacturers live on small margins, some live on large margins. They could make the exact same cable, but it would sell at a large price difference. Are materials? No, they aren’t. A good example is Teflon. It is not as good a dielectric as polyethylene. Teflon has a higher dissipation factor at the same solid state or equivalent foam factor (some call this the loss tangent) than polyethylene. The dielectric constant of Teflon is slightly lower than Polyethylene (2.1 verses 2.25) so a smaller cable can be made. But as far as “sound” goes, Teflon is mostly expensive. Isolated design can use Teflon to good effect (do things PE can’t), but Teflon is NOT universally the better material, and most of Teflon’s outstanding properties goes to no practical use in an audio cable. Consider for a moment too that the material cost of the cable is a minor point to the sell price. Manufacturers amortize development cost (the real cost) over their customer base but that doesn’t mean they still won’t levy an outsized sales margin, too! Big companies can absorb smaller margins, but they don’t have to. Design? Yes or no. Good design / good materials is a yes, good design with bad materials maybe (design may overcome materials) a no.

So what are we to do? I’d have to listen to subgroups of cable, and move the winner of that subgroup to text, and next and next till I heard every cable in the matrix. If I listened to 40%, or 60% or 80% of the cables, I could calculate the odds of picking the best one at each percentage of the universe of cables. But it is just the "odds” and not the answer. That would take awhile so I am forced to make a decision based on knowledge that really isn’t the answer. But I know this, and that keeps the matrix open for me.

I like Solid state amps, so I make a point of listening to tube amps and pre amps. I expand my matrix, I don’t create a downhill bias and follow the SS guys. I am aware that when I feel the urge to continue to push in one direction, that I’m probably heading in the wrong direction. I did this with speaker and interconnect cables. I stopped one day and asked myself, why am I afraid to re listen to cable? Knowledge is used as power. And if I found out I was wrong, my power was taken away from me. My “tribe” is gone. I’m alone again. Good. I can now go and experience the new MATRIX and remember to find “knowledge” that may never well be the “answer”, as I’ll never hear all the DATA. But isn’t that what makes this hobby so much fun?

Unlike the author of that recent paper @ http://www.roger-russell.com/wire/wire.htm, I’m fond of feedback. It expands the matrix that people need to get involved in. The real data matrix has the good, the bad, and the ugly in it, with no bias. The Japanese auto industry studied the Americans methods in the states incessantly to learn every thing we did in manufacture, but to be sure they recognized the biggest mistakes so they wouldn’t make them. They plugged those methods into the matrix, and use statistics to let the best methods be found using an unbiased SPC methodology. This is harder to do with audio as the audience is one, you.

We forget how to learn. We form tribes, collapse the matrix and get stuck in time. Sometime we crash and burn on a mogul that knocks the senses back in and sometimes we ski down the bunny slope all our lives. I’d rather be knocked around and expand my matrix.

OK, just to let you know, I selected the NORDOST tyr2 3 meter XLR interconnect to go between my XP-10 and MOON W-8.
rower30
Frogman,

??? Not so sure where you got all that. But if you like to form conclusions (...you have a problem with the notion that I feel confident about this specific aspect of audiophilia...) to make a point to someone...go ahead. I wasn't even talking about you. Masking knowledge by touting credentials isn't confidence. Confidence stands on it's own. Your opinion is fine (for or against), and I said so (I trust your answers just as much because it simply you, frogman.). But, if you really need to be that confident that you jump to conclusions, assume I'm referring to you (nope), fine.

Now, how many filaments did it take to invent the light bulb? Plenty, and if you took the "failures" one after the other to doubt the inventors intuition, we'd all have been in the dark a whole lot longer. Most would think the guy didn't know what he was doing, but an insight completed is worth 100 ideas never put to task. Failure is a "wrong" answer, but I'd rather fail to success than try to preserve my record and do little to nothing.

Medicine is a weird science. My wife had breast cancer. Funny, the doctors were wrong on eight of ten issues. Oh, but they were doctors. My wife sat behind a PC and figured it out on her own. Credentials aren't an "answer" but simply a sort of odds of getting a right answer. But, like I said, the answer still has to stand on it's own to be "right" or "wrong". More right answers in front of the wrong answer don’t change it.

So you have different examples of how "knowledge" is gained, used, and improved. Sure, you'd like to have "old" knowledge be accurate (medicine) over "new" (inventivness) as. But, better compare notes in either case if you can. Fool me once shame on you, fool me twice shame on me.

People want to simply give in to the "odds" of getting a right answer. Credential tell the probability of a right answer, and just says I'll more than likely find it correct. But find it I will. I examine each answer one at a time, and different perspectives as new information arises. The matrix isn't static. In God we trust, all else bring data.

So my system is never "done" per say, even if I keep it forever (that doesn't mean it's perfect by any means). I'll just know that things are better out that should I do so.
hi rower30:

i think you have been ambiguous in your analysis.

it seems your concern is listener bias.

for example, dislikintg solid state, before auditioning solid state components.

you have not stated such a concern or other analogous examples.

essentially, bias denotes a sentiment about that which has not been sufficiently experienced,. another example would be to generalize , based upon limited experience.

you have not commented on the common usage definition of bias, as applied to matters audio.

if you develop preferences based upon a sufficiently large experience set, or because of one's nervous system, one prefers odd order harmonic harmonics and certain coloration(s).

you have not distinguished between bias (prejudice) and genuine, justified preferences.

it is not unreasonable to have preferences, after extensive listening. those preferences do not constitute bias.

so, i would appreciate a coherent statement as to what you mean by "bias" in audio.
Mrtennis,

Good question. Bias is the conclusion that information that you have gathered is still "true", and "fair" or the most accurate representation of a physical outcome.

When we stop collecting additional data to support a "bias" we now have a "belief system", or a choice made using what is incomplete and not fair. Once you stop challenging your data, you can no longer be as confident that it is right. It still could be, mind you, you just can't say that it is statistically as accurate without new data involved.

Look at anti-oxidants and aging, new data says it's bunk.
The Gulf-stream mechanisms warming Europe is being challenged by global ocean sensors (more accurate than a hypothesis derived 100 years ago). The "old" model is being taken apart as wrong (we believed it for 100 years!).

I could go on, but that's my general premise...people (like me!) take what data we want to "believe" and then STOP the process once it agrees with our "belief" system. But, it's not totally true, is it? A belief system is static, and unfair.

Human nature and date is very odd.
We'll spend 20 bucks a month on the lottery. Our odds are millions to one to win, and have nothing. We could put the twenty in a mutual fund and in 30 years have six figures. What do most people do? Here we have a preference, though. We KNOW the odds are terrible, but we do it anyway. There is nothing wrong with that.

I don't call bias preference. Preference does not involve a degree of bias, it is a choice based on what is right. I HATE a truly accurate stereo as it has too little bass energy to make me happy. The test equipment is right, but I'm not happy. Still, I KNOW what is right from wrong. My PREFERENCE is more bass. Still, I know what is right...that's the key difference. It is a "fair" choice, as I know what right is.

From my original example of one small matrix I neglected to address were cables. I "knew" (stopped collecting data when I had enough data to support what I wanted to believe) cables made little difference. That "was" true with 1980's technology, but not so today. I didn't "prefer" to think cables made no difference, I though I knew that they didn't make a difference. This wasn't a "bias" in my mind as it was "true" (NOT!). I didn't think I was unfair, but to exclude new data says that I actually was. You can only have a preference when you know the truth (no bias), and decide to move to, or away, from it. So a preference is supported with an accurate depiction of what is possible to chose from.

BIAS - Prejudice in favor of or against one thing, person, or group compared with another, usually in a way considered to be unfair.

That said, I am fully aware of this bias. There are many aspects of things that I think I have the most "right" answer to, which are undoubtedly incorrect. This may not be a conscious bias to be unfair, but it is, even if I don't know it. You can never stop expanding the matrix with new data to reset your data sets "bias" to as small a value as you can. Any data set that does not include all the data on a given subject is "unfair" and hence, bias.

A preference does not have to be right, you just like it. And, there is absolutely nothing wrong with that when you make the choice against the most accurate representation of "right" at the time.

That's why this hobby is so much fun. We get it all "right" in our heads, or we all try to as best we can, and then place our "preferences" against the data sets. This is different than BIAS, where we exclude data so our preferences are less accurate as to what we "could" have chosen from. Time moves on...so does the data.

I didn't know I could change cables to improve my system 30 years ago. Today? I have a "preference" to cables I've listened to for sure, but my "bias" that they don't make a difference has been reduced, not eliminated, by listening to many, many cables. I can now select from a much less bias data set and get a much more accurate preferred sound.