[higbvuyb]

 
I think your view and understanding of models is too narrow. Note that models come in different forms, and some are more specific and parameterized, while some contain unknown variables.
 
For example, you can do the math in a linear systems textbook, and you will find that you can make a lot of predictions about how linear systems behave that are generalized to all linear systems. You don't need to know the FR specifically. That's one kind of model.
 
Measuring FR of a specific device involves taking the math and using it to fit observations, and compute the FR. What you end up with is a more specific model.

That's not relevant to the point being made. Assuming that something acts linearly would be applying a specific model to the data. However, that does not make the data into a model, much like petrol isn't a car just because the car is using it.
 

  First of all, you aren't arguing that measurements aren't models, you are just arguing that they are very good models. Sometimes they are. As far as models of audio equipment, they are only useful insofar as they have been "rigorously" correlated with hearing.

You didn't understand. This is a side point; some models are rigorously tested and well understood, minimising the assumptions required to use them.
 
This has nothing to do with measurements being models or whatever. "a voltmeter measures voltage" is a model but it certainly isn't a measurement.
 

  Of course it's a model of hearing. An FR with a big peak in the bass is roughly correlated with the listening experience we report as "loud bass." It's a rough model, but this is a very important fact, because as I said, models in audio are only useful insofar as they have been correlated with the listener's experience.

No, the model being used is your assumption that a bass peak correlates with hearing loud bass. The measurement just gives you a number. It doesn't stand behind you with a baseball bat, telling you "you must interpret the data this way with these assumptions"
 
This isn't a difficult concept! Read this again and think about it until you understand the distinction being made here.
 
The measurement is the data.
 
Models are used to interpret data. Data is not a model. As with your above post, you keep incorrectly conflating the data itself with how you are interpreting it.

 

[higbvuyb]

  Assuming it's constant might be a perfectly useful model. It would be for computing BMI, for instance. Don't assume it's a bad model.
 
It depends on what use you want to put the measurement to.
 
Very simple idea, often overlooked.
 
The number 5'10" is not separate from the context within which it was computed, and knowing that context helps to make predictions.
 
That number was computed with a device that had error and noise, for instance. If I know how much error and noise, it helps future predictions.
 
If I know whether you measured that in the morning or evening, it helps future predictions.
 
It's really simple, folks. All measurements are models.
 
You can argue that it doesn't matter much, but in audio it does matter a whole heck of a lot.

Your argument boils down to:
 
"Data plus a specific interpretation is X, therefore data is X"
 
This is fundamentally logically flawed.
 
A measurement cannot be separated with how it was measured, but not because of why you think it is.

 

[manbear]

@raddle 

You ask a lot of rhetorical questions. Here's one for you -- what is the purpose of this idea that subjective impressions are more accurate than measurements? For example, frequency response graphs. Have you found that your subjective impressions disagree with frequency response graphs? Have you encountered headphones that measure with a peak in the bass but don't have a subjectively strong bass response?

I don't really see the point of an attack on measurements unless your subjective impressions actually disagree with them. You have theoretical issues with measurements, but if these models are really inadequate, you should be able to observe real life deviations from what is implied by measurements. Have you? If not, what is the purpose of this inquiry of yours?

I agree with some of the other posters about your conflation of information and interpretation. As well as your failure to see that giving priority to subjective impressions is a model in its own right (and the purpose of that model is often to confirm biases and justify spending money). But these are mostly academic games in the absence of a real life conflict between the measurements as models idea you object to and your subjective impressions. You need to show that there is some practical motivation for making this distinction in order to make the details worth hammering out. 

Another way to say this is: suppose you are right and the people who disagree with you are wrong. What benefits do you enjoy as a result of viewing the situation correctly? What mistakes are the incorrect folks doomed to make as a result of not understanding your point? Are these benefits and mistakes things that most Head-Fiers would or should care about?

Personally, I consider both measurements and subjective impressions. I don't really see the need to make some iron-fisted rule that one must always be more valuable or accurate than the other, insofar as it would increase my enjoyment of music or inform my purchases. 

 

[proton007]

@raddle
 


You ask a lot of rhetorical questions. Here's one for you -- what is the purpose of this idea that subjective impressions are more accurate than measurements? For example, frequency response graphs. Have you found that your subjective impressions disagree with frequency response graphs? Have you encountered headphones that measure with a peak in the bass but don't have a subjectively strong bass response?


I don't really see the point of an attack on measurements unless your subjective impressions actually disagree with them. You have theoretical issues with measurements, but if these models are really inadequate, you should be able to observe real life deviations from what is implied by measurements. Have you? If not, what is the purpose of this inquiry of yours?


I agree with some of the other posters about your conflation of information and interpretation. As well as your failure to see that giving priority to subjective impressions is a model in its own right (and the purpose of that model is often to confirm biases and justify spending money). But these are mostly academic games in the absence of a real life conflict between the measurements as models idea you object to and your subjective impressions. You need to show that there is some practical motivation for making this distinction in order to make the details worth hammering out. 


Another way to say this is: suppose you are right and the people who disagree with you are wrong. What benefits do you enjoy as a result of viewing the situation correctly? What mistakes are the incorrect folks doomed to make as a result of not understanding your point? Are these benefits and mistakes things that most Head-Fiers would or should care about?


Personally, I consider both measurements and subjective impressions. I don't really see the need to make some iron-fisted rule that one must always be more valuable or accurate than the other, insofar as it would increase my enjoyment of music or inform my purchases. 

I think the OP is getting confused, because he didn't set the definitions beforehand. Now he's caught in a web of changing definitions and interpretations.
The most basic rule of an enquiry of this nature is to be clear about the terms and their scope, because it also helps others in providing valid arguments.

The way I see it, the OP has already set the premise and conclusion, without taking into account the soundness of his argument:
-- measurements are models
-- models cannot be guaranteed to give accurate results
-- hence measurements cannot be relied on.

So what do we need to disprove? That measurements are not models?
Here goes:
Measurements are assignments of numbers to a physical property. The only property of a unit of measure is that it's agreed upon and consistent (my metre should be same as everyone elses).

So, a measurement is not a model, unless the OP can prove that assignment of numbers is a model.

 

[manbear]

I think the OP is getting confused, because he didn't set the definitions beforehand. Now he's caught in a web of changing definitions and interpretations.
The most basic rule of an enquiry of this nature is to be clear about the terms and their scope, because it also helps others in providing valid arguments.

The way I see it, the OP has already set the premise and conclusion, without taking into account the soundness of his argument:
-- measurements are models
-- models cannot be guaranteed to give accurate results
-- hence measurements cannot be relied on.
 

I agree that OP is a bit confused about measurements vs. models. My earlier point was that even if he wants to says that measurements (or models?) cannot necessarily be relied upon, since they are not guaranteed to be accurate, they can still be close enough to be useful. A real-world example of how this supposed inaccuracy is problematic is necessary. Does he actually have examples of how they are not close enough to be useful?

In a way, I'm just accepting what the OP is saying about measurements and models for the sake of argument. Even if the OP is right, why does it matter? What problems are the OP's ideas going to fix, if he were correct?

 

[KraftD1]

If we follow the argument to its end, external reality is just a model we each produce (or maybe just you... Solipsism).

 

[xnor]

No no no no. Measurements have to be 100.0% accurate, else they're useless. Science doesn't have all answers, so it's useless.
Perception on the other hand doesn't have to be accurate or complete. That is totally fine.
 
Put that way it's a clear example of double standards.

 

[SanjiWatsuki]

  No no no no. Measurements have to be 100.0% accurate, else they're useless. Science doesn't have all answers, so it's useless.
Perception on the other hand doesn't have to be accurate or complete. That is totally fine.
 
Put that way it's a clear example of double standards.

I think we allow leeway with measurements, though.
 
Rin Choi did a comparison of the different measurements, and everyone's raw measurements differ in their own way. Tyll's measurement set up is a bit different than everyone else's. Rin Choi's measurement set up typically has a bit more bass. I think Samsung's measurment set up has a bit less bass. PersonalAudio.ru is sometimes just flat out wrong in the bass, so they do two different measurements and say that one type is more accurate than the other for lower frequencies and the other is more accurate for higher frequencies. Plus, everyone struggles with seal.
 
And don't get me started on compensation curves.

 

[castleofargh]

Plus, everyone struggles with seal.

 
they have a right to struggle

 
we shouldn't cross reference those measurements.
but even inside the measurements done by the same person, I guess there could be some bias sometimes. like maybe trying a little more to get a good response out of an expensive gear and not giving as many chances to a 50$ phone. but we have to trust the methodology of those doing the experiment.
at the end of the day I'd say it's still more informative than "great soundstage, strong bass and muddy medium".

 

[SanjiWatsuki]

   
they have a right to struggle

 
we shouldn't cross reference those measurements.
but even inside the measurements done by the same person, I guess there could be some bias sometimes. like maybe trying a little more to get a good response out of an expensive gear and not giving as many chances to a 50$ phone. but we have to trust the methodology of those doing the experiment.
at the end of the day I'd say it's still more informative than "great soundstage, strong bass and muddy medium".

 
 
You're right. I'm just contesting that measurements need to be 100% accurate to be accepted, generally speaking. I probably shouldn't have responded to a facetious point, though.

 

[dvw]

Measurement and models are two different things. Measurement is 100% correct data while models are best guess based on prior knowledge. For instant, popular technology standard like a WiFi were simulated as a model before they were established as a standard. It would be too expensive and time consuming to do trial and error type development. Similarly silicon chip are simulated as well. And yes, noise and non-linearity are part of the equation. Are they perfect? No they are not. Frequently, they will have to have revision. But models will continue to improved as we get more experienced with it.
 
Biological models are statistically modeled  If I'm 5'10", the odd of I'm being 160lb is high but it is not certain. But for sure, it will not be able to predict with 100% certainty. But how are you going to predict how tall and heavy I will be with my measurement today? I will be taller in the morning, true, but by how much. What is the prediction mechanism?
 
An example that was cited often is FR. FR does not predict what you can hear. If the FR is 20 to 20KHz. But if you can only hear 60 to 16KHz, the only relevant portion is 60 to 16KHz. FR is a measurement although it can be use as a transfer function. And of course I can use model to predict a FR. The hearing model on the other hand is a biological function. Actually, there is such a model. It is primarily used by hearing aid researcher to simulate their design.
 
Here's another frequent argument on hearing is better than measurement because of xxxxxxx.  But if you take the measurement with two different instrument you will have the same result within tolerance. However, if you try it with humans, it will be all over the place. Even a single person can't have consistent result. In a blind test, 7 out of 10 is considered a good score. If an instrument does that, it will be going back to the factory for recalibration.

 

[proton007]

See? There's the confusion.
Are we talking about measurement itself? Or the act of measuring and the setup usdd to measure?
The former is absolutely consistent, the latter may have variations. Still, where's the model in that?

 

[raddle]

Everyone here is using a different meaning of "model" so it's a bit confused. I'm using the sense described in the Wikipedia entry on "Scientific Modeling" -- the essence of it is some kind of construct (mathematical or not) that represents some part of reality (or some part of a more complex system).
 
A lot of you are talking about measurements being data and models being how you interpret the data.
 
I'm looking at the big picture... numbers without a context are just numbers. There is always a context.
 
Whether you call one part of the data "data" and another part an "assumption" and another part a "model" or whatever you want -- my point remains the same, that measurements are representations of reality which are used in the ways models are used, such as to make predictions.
 
We are getting hung up on the word "assume." When I say that stating your weight is 160 pounds "assumes it's a constant" I don't mean that you put blinders off and go into denial. The "assumptions" of a model are in the mathematics. Maybe a better thing to say is that "treat your weight as a constant." Treating your weight as a constant is a way to predict your weight in the future. It's so obvious that people don't realize they are doing it.
 
You could have a better way to predict your weight in the future. You could plot your weight over time and fit a polynomial curve to it, then report to me the coefficients of the polynomial. Might be better in some situations.
 
FR is most definitely a model of hearing, by which I mean it takes a real phenomena -- a part of reality -- which is the experience of hearing a device with that FR--and represents that reality with a curve. It doesn't need to be perfect. It doesn't need to match everyone. It's still a model of hearing. Why do you think we put so much emphasis on getting a flat FR? Not so it can look pretty, but so it can sound pretty.
 
Models don't need to be perfect or useful in all contexts to be models.

 

[higbvuyb]

  Everyone here is using a different meaning of "model" so it's a bit confused. I'm using the sense described in the Wikipedia entry on "Scientific Modeling" -- the essence of it is some kind of construct (mathematical or not) that represents some part of reality (or some part of a more complex system).
 
A lot of you are talking about measurements being data and models being how you interpret the data.
 
I'm looking at the big picture... numbers without a context are just numbers. There is always a context.
 
Whether you call one part of the data "data" and another part an "assumption" and another part a "model" or whatever you want -- my point remains the same, that measurements are representations of reality which are used in the ways models are used, such as to make predictions.
 
We are getting hung up on the word "assume." When I say that stating your weight is 160 pounds "assumes it's a constant" I don't mean that you put blinders off and go into denial. The "assumptions" of a model are in the mathematics. Maybe a better thing to say is that "treat your weight as a constant." Treating your weight as a constant is a way to predict your weight in the future. It's so obvious that people don't realize they are doing it.
 
You could have a better way to predict your weight in the future. You could plot your weight over time and fit a polynomial curve to it, then report to me the coefficients of the polynomial. Might be better in some situations.
 
FR is most definitely a model of hearing, by which I mean it takes a real phenomena -- a part of reality -- which is the experience of hearing a device with that FR--and represents that reality with a curve. It doesn't need to be perfect. It doesn't need to match everyone. It's still a model of hearing. Why do you think we put so much emphasis on getting a flat FR? Not so it can look pretty, but so it can sound pretty.
 
Models don't need to be perfect or useful in all contexts to be models.

Assuming your definitions are valid, what is the point you are trying to make?

 

[proton007]

Everyone here is using a different meaning of "model" so it's a bit confused. I'm using the sense described in the Wikipedia entry on "Scientific Modeling" -- the essence of it is some kind of construct (mathematical or not) that represents some part of reality (or some part of a more complex system).

A lot of you are talking about measurements being data and models being how you interpret the data.

I'm looking at the big picture... numbers without a context are just numbers. There is always a context.

Whether you call one part of the data "data" and another part an "assumption" and another part a "model" or whatever you want -- my point remains the same, that measurements are representations of reality which are used in the ways models are used, such as to make predictions.

We are getting hung up on the word "assume." When I say that stating your weight is 160 pounds "assumes it's a constant" I don't mean that you put blinders off and go into denial. The "assumptions" of a model are in the mathematics. Maybe a better thing to say is that "treat your weight as a constant." Treating your weight as a constant is a way to predict your weight in the future. It's so obvious that people don't realize they are doing it.

You could have a better way to predict your weight in the future. You could plot your weight over time and fit a polynomial curve to it, then report to me the coefficients of the polynomial. Might be better in some situations.

FR is most definitely a model of hearing, by which I mean it takes a real phenomena -- a part of reality -- which is the experience of hearing a device with that FR--and represents that reality with a curve. It doesn't need to be perfect. It doesn't need to match everyone. It's still a model of hearing. Why do you think we put so much emphasis on getting a flat FR? Not so it can look pretty, but so it can sound pretty.

Models don't need to be perfect or useful in all contexts to be models.

The frequency of a signal is not a model. Its a naturally occurring phenomenon which is assigned a number.
The definition of a second is also tied to natural phenomenon (cesium atoms).
Hence measuring an FR is not a model, unless you can prove that a fourier analysis is a model.