The Impossible

[Davian]

What about immaterial things that arent conscious?

What about the elephants stampeding in Zimbabwe?

 

[Loudmouth]

Well, sort of. We make assumptions about what is real and base models on that.

We make measurements of reality to the best of our ability, and we use those measurements to test our hypotheses. Assumptions really aren't a part of it.

It didn't seem this question was aimed at me, but I'll answer it. No, that was not my intent. I've seen a significant amount of bluster about science showing this or that to be impossible, so I decided to put some focus on the issue. It seems to me science is an attempt to show what we think is possible, and never really shows much of anything to be impossible ... though there are cases of showing something to be highly unlikely.

Since science is tentative it really isn't set up to say this or that is impossible. All science can say is that the empirical evidence is consistent or inconsistent with a given hypothesis.

I'm paraphrasing (and taking great liberty with the paraphrase), but it's interesting to see people working through what they don't believe in ... coming to a conclusion that if some really powerful entity existed it probably would be possible ...

That's a big IF. That would seem to apply to any deity we choose, or any deity, trickster spirit, or demon that we make up on the spot. That's the problem. People think that an unevidenced belief holds the same weight as an evidenced claim simply because it could be possible. That makes zero sense.

 

[Resha Caner]

It lacks a positive ontology.

I don't think that means science has declared it impossible. It would only mean science has not yet defined it.

It is a term that you have used; perhaps you could explain it.

Did I? I suppose I might have. Regardless, I would assume the troublesome part is the adjective. Were we to agree on a definition of consciousness (and I'm not sure we have, though we might be able to accomplish it), the "immaterial" would be the sticky part, wouldn't it?

 

[Resha Caner]

Since science is tentative it really isn't set up to say this or that is impossible. All science can say is that the empirical evidence is consistent or inconsistent with a given hypothesis.

Sure. I won't be proclaiming victory here. If the thread helped add some clarity to the discussion, I'm happy.

And observing human behavior never ceases to be entertaining.

That's a big IF. That would seem to apply to any deity we choose, or any deity, trickster spirit, or demon that we make up on the spot. That's the problem. People think that an unevidenced belief holds the same weight as an evidenced claim simply because it could be possible. That makes zero sense.

Understood. Though I've been careful with my words. All I said was "powerful entity", not "deity". The U.S. military could be considered a powerful entity within the proper context.

We make measurements of reality to the best of our ability, and we use those measurements to test our hypotheses. Assumptions really aren't a part of it.

Yes to the first sentence. No to the second. There is a branch of science regarding measurement, and, as with everything, it starts with assumptions.

Or are you claiming something self-evident here?

 

[keith99]

It is impossible, and always will be, to cover the entire surface of this planet with liquid water.

Garbage.

That is an engineering problem that is not even all that difficult in theory.

Don't try to 'get more water', simply make the ground level. Dumping the Himalayas into the Marianas Trench would be a pretty good start.

 

[Loudmouth]

Yes to the first sentence. No to the second. There is a branch of science regarding measurement, and, as with everything, it starts with assumptions.

Perhaps you could give us an example?

 

[Resha Caner]

Perhaps you could give us an example?

Of the assumptions? Wow. There are so many it's difficult to know which thread to yank first.

Before I begin, I'll note that you needn't take just my word for it. There are many books on the topic. I've linked to one:
Theory and Design for Mechanical Measurements - Fourth Edition: Richard S. Figliola, Donald E. Beasley: Amazon.com: Books

As always, the assumptions range from the simple & obvious to the obscure. A simple assumption would be the arithmetic basis of the measurement (linear, logarithmic, etc).

Slightly more complex is the assumption that all significant forms of error can be mitigated (a ruler expands with temperature) or statistically quantified (the noise is Gaussian and not completely random).

At the obscure end are assumptions that the measurement is unaffected by location or time (i.e. that physical laws govern both the thing & the measure such that if the thing behaves consistently here and there, the measure will also behave consistently here and there).

 

[Loudmouth]

Of the assumptions? Wow. There are so many it's difficult to know which thread to yank first.

Before I begin, I'll note that you needn't take just my word for it. There are many books on the topic. I've linked to one:
Theory and Design for Mechanical Measurements - Fourth Edition: Richard S. Figliola, Donald E. Beasley: Amazon.com: Books

The very title of the book says that measurement is a theory, not an assumption. Theories and assumptions are not the same thing.

As always, the assumptions range from the simple & obvious to the obscure. A simple assumption would be the arithmetic basis of the measurement (linear, logarithmic, etc).

That is tested on a regular basis. Our math makes predictions that we can test.

Slightly more complex is the assumption that all significant forms of error can be mitigated (a ruler expands with temperature) or statistically quantified (the noise is Gaussian and not completely random).

Standard error is always included as part of any scientific study. It is not assumed that error in measurement can be mitigated. It is always controled for.

At the obscure end are assumptions that the measurement is unaffected by location or time (i.e. that physical laws govern both the thing & the measure such that if the thing behaves consistently here and there, the measure will also behave consistently here and there).

That is also checked in such fields as astronomy and physics. That is not assumed.

 

[Davian]

I don't think that means science has declared it impossible. It would only mean science has not yet defined it.

I think 'science' would declare 'get back to me when you have a positive ontology'.

Did I? I suppose I might have. Regardless, I would assume the troublesome part is the adjective. Were we to agree on a definition of consciousness (and I'm not sure we have, though we might be able to accomplish it), the "immaterial" would be the sticky part, wouldn't it?

You are the one using the term. When you find a scientific explanation of consciousness that leaves the door open for "immaterial", let me know.

 

[Resha Caner]

The very title of the book says that measurement is a theory, not an assumption. Theories and assumptions are not the same thing.

Your reply (in its entirety, not just what I've quoted) implies not only a lack of familiarity with the subject, but a superficial response. I'm used to more from you, so I'll give you a chance to reconsider.

It doesn't take much effort to understand what I've said. So, it's not like you'll have to go to the library and read the book I mentioned. I grabbed something at random from Google Books:

"The basic tenet of classical measurement theory evolved from the assumption that random error is an element that must be considered in all measurement."
- Measurement in Nursing and Health Research by Elizabeth R. Lenz

So, let's take one of the assumptions I mentioned - that of Gaussian noise. Gaussian noise is very convenient because it allows for noise terms to be mathematically manipulated within the model (per Bendat's book on Nonlinear Systems) in a way that allows the base model itself to be correlated.

But how does one "prove" the noise is Gaussian? Well, there are tests such as Kolmogorov–Smirnov, but those tests include assumptions of their own. Further, it doesn't give a yes/no answer. It becomes a matter of degree, and so people develop rules of thumb on what they will accept as Gaussian. So, then, suppose you decide the noise is not Gaussian. Is there always a mathematical model for all types of noise? No.

That doesn't mean it will never be found, but considering how developed statistics is, it becomes less and less likely. So, it raises the question: is there a phenomena that is completely random (i.e. which doesn't have a distribution?) I don't know. But in order to proceed, we have to assume that is not the case - at least that such noise is not significant.

 

[Resha Caner]

I think 'science' would declare 'get back to me when you have a positive ontology'.

So science has no responsibility to develop definitions? Hmm. Regardless, this is a red herring. Nothing in your statement implies impossibility.

You are the one using the term. When you find a scientific explanation of consciousness that leaves the door open for "immaterial", let me know.

I didn't use it in this thread. You brought up the subject.

Since your memory of my past threads is so good, you must recall that I did provide a definition of "immaterial".

 

[Loudmouth]

Your reply (in its entirety, not just what I've quoted) implies not only a lack of familiarity with the subject, but a superficial response. I'm used to more from you, so I'll give you a chance to reconsider.

Reconsider what?

Scientists are looking for evidence that the laws of physics either have changed temporally or by location. They have even set up mazers that measure the anisitropy of spacetime as the Earth moves about the Sun to see if laws change by locality. The recent kerfuffle over faster than light neutrinos highlights this topic once again.

Perhaps you want to reconsider your position?

It doesn't take much effort to understand what I've said.

It doesn't take much effort to understand what I've said either.

So, let's take one of the assumptions I mentioned - that of Gaussian noise.

Not assumed. You can acutally measure Gaussian noise in your data.

But how does one "prove" the noise is Gaussian? Well, there are tests such as Kolmogorov–Smirnov, but those tests include assumptions of their own.

Those tests include tests of their own.

That doesn't mean it will never be found, but considering how developed statistics is, it becomes less and less likely. So, it raises the question: is there a phenomena that is completely random (i.e. which doesn't have a distribution?) I don't know. But in order to proceed, we have to assume that is not the case - at least that such noise is not significant.

Why would we have to assume that phenomena are not random? In a pile of radioactive uranium the atom that decays next is random, as one example.

 

[Resha Caner]

Scientists are looking for evidence that the laws of physics either have changed temporally or by location. They have even set up mazers that measure the anisitropy of spacetime as the Earth moves about the Sun to see if laws change by locality. The recent kerfuffle over faster than light neutrinos highlights this topic once again.

Of course the possibility is under consideration. It's mentioned in the measurement theory books I've been talking about. But has it been found? Not that I'm aware of. As such, even though it is an acknowledged possibility, the assumption we currently use is that measures are constant.

If you're saying changing physical laws have been verified, that is new information to me, but I don't see how it invalidates my point. It was certainly an assumption of the past. Would it mean, going forward, that we will know for every circumstance whether the measure is constant or changing? I don't think so. It would remain an assumption.

Not assumed. You can acutally measure Gaussian noise in your data.

Those tests include tests of their own.

You're saying you know of a test that will give a yes/no answer that noise is a particular distribution? Again, that would be new information for me.

You don't measure Gaussian noise. You measure. That which doesn't fit the model and can't be explained by the model is noise. If the noise is essentially bell-shaped (to oversimplify), you assume it's Gaussian. I've never seen it be exactly Gaussian.

... atom that decays next is random ...

is random? Were you part of earlier discussions on this? The camps divided into two basic definitions of random:
1) Something that is not governed by a law
2) Something that can't be predicted

The first definition is a problem because it supposes new laws can never be found to explain what we currently can't explain. The second is the safe definition, but it is a merely pragmatic one. It makes no metaphysical claim that something actually is random.

 

[Loudmouth]

Of course the possibility is under consideration. It's mentioned in the measurement theory books I've been talking about. But has it been found? Not that I'm aware of.

Just goes to show that it isn't assumed. It is actually tested, and continually tested.

As such, even though it is an acknowledged possibility, the assumption we currently use is that measures are constant.

That's not true at all. Have you ever worked in a lab with sophisticated equipment? Have you evern been involved in designing complex experiments? Lab equipment has to be constantly monitored, calibrated, and zeroed. The very last thing you would ever do is assume that your measurements are constant.

If you're saying changing physical laws have been verified, that is new information to me, but I don't see how it invalidates my point. It was certainly an assumption of the past. Would it mean, going forward, that we will know for every circumstance whether the measure is constant or changing? I don't think so. It would remain an assumption.

Are Newton's laws invariable at every velocity? Nope. Are Newton's laws invariable in all gravitational wells? Nope.

If anything, science is the history of finding laws that aren't laws, finding out where laws breakdown.

You're saying you know of a test that will give a yes/no answer that noise is a particular distribution? Again, that would be new information for me.

All you do is plot your data and see if it fits a Gaussian distribution. It isn't that hard to do.

You don't measure Gaussian noise. You measure. That which doesn't fit the model and can't be explained by the model is noise. If the noise is essentially bell-shaped (to oversimplify), you assume it's Gaussian. I've never seen it be exactly Gaussian.

Height of humans in a large population will probably fit a Gaussian curve. I don't think that equates to noise. What you claim is noise is actually a proposal, not an assumption. You produce statistical models to bolster your claim that it is noise, but in the end it is still a testable claim that is falsifiable.

is

random? Were you part of earlier discussions on this? The camps divided into two basic definitions of random:
1) Something that is not governed by a law
2) Something that can't be predicted

The first definition is acausality. The second definition is random.

The first definition is a problem because it supposes new laws can never be found to explain what we currently can't explain. The second is the safe definition, but it is a merely pragmatic one. It makes no metaphysical claim that something actually is random.

What we can do is show that the measurements are consistent with a random model.

 

[Resha Caner]

That's not true at all. Have you ever worked in a lab with sophisticated equipment? Have you evern been involved in designing complex experiments?

Yes. Have you?

The very last thing you would ever do is assume that your measurements are constant.

You are confusing what I called the "measure" (the model for measuring) and the measurement itself. If I have a control system to maintain the temperature of a room, I expect it to vary. It will be x +/- y deg Celsius. What I expect to be constant is the relationship between heat and temperature - the model for measuring.

If I measure that a wall needs to be 8 feet tall here, and cut my board to a length of 8 feet there, I expect 8 feet to be the same thing in both places even if the wall is actually 8 +/- 0.2 and the board is 8 +/- 0.1. If they aren't the same, the measure is pointless.

Are Newton's laws invariable at every velocity? Nope. Are Newton's laws invariable in all gravitational wells? Nope.

That is a case of Newton being wrong. It doesn't mean reality has changed since the time of Newton.

All you do is plot your data and see if it fits a Gaussian distribution. It isn't that hard to do.

By what test and how close must the "fit" be? Should I give you some data and you tell me whether it is Gaussian or not? It just so happens I was struggling with exactly this problem about a year ago. I'll strip the labels off and give you the data.

What we can do is show that the measurements are consistent with a random model.

How consistent must they be? At what exact yes/no point is the model confirmed to be right or wrong?

 

[Loudmouth]

Yes. Have you?

Yes, and I never assume that my measurements are consistent. I do a lot to make sure that the machine is calibrated, zeroed, and working as it should. For a scintilation counter I run samples with known amounts of radioactive decay to determine if the equipment is correctly the number of disintegrations per minute. For an HPLC, I run standards of known concentration to test both peak height (quantity) and retention times to verify that the instrumentation is working properly. For a spectrophotometer, we run caffeine standards of known concentration from time to time to make sure there is no drift in the equipment.

At a more fundamental level are the theories that describe why the instrument should work. We can just as easily test those theories. None of it is assumed.

You are confusing what I called the "measure" (the model for measuring) and the measurement itself. If I have a control system to maintain the temperature of a room, I expect it to vary. It will be x +/- y deg Celsius. What I expect to be constant is the relationship between heat and temperature - the model for measuring.

Heat and temperature are one in the same.

If I measure that a wall needs to be 8 feet tall here, and cut my board to a length of 8 feet there, I expect 8 feet to be the same thing in both places even if the wall is actually 8 +/- 0.2 and the board is 8 +/- 0.1. If they aren't the same, the measure is pointless.

So we can determine if they are not the same? If so, it shoots a hole in your claim about assumptions.

That is a case of Newton being wrong. It doesn't mean reality has changed since the time of Newton.

But it does show that scientists are continually testing scientific "laws" to see if they breakdown under specific circumstances. They are not assumed to be constant in all conditions. Entire careers have been made by testing and disproving the very things you claim are just assumed.

By what test and how close must the "fit" be? Should I give you some data and you tell me whether it is Gaussian or not? It just so happens I was struggling with exactly this problem about a year ago. I'll strip the labels off and give you the data.

There should be plenty of statistical programs that will do that for you. The most common, but arbitrary, value is a 95% confidence value. That is, there is only a 5% chance that a non-Gaussian distribution would mimic a Gaussian distribution just by chance.

It's not really my field, but I have heard that most physicists look for a 5 sigma level of confidence, or 5 decimal places which is 99.999% that the statistical model is correct. You can never reach 100% certainty, btw. What statistics allows you to say is that there is something worth studying further, that there is a strong indication of cause and effect.

How consistent must they be? At what exact yes/no point is the model confirmed to be right or wrong?

It never is confirmed 100% because science is always tentative. Newton's laws made it through 400 years of constant prodding and probing without any problem until the accuracy of our instruments improved. Only then did we see the deviations from the laws in extreme environments.

 

[Davian]

So science has no responsibility to develop definitions? Hmm.

Where did I say that?

Regardless, this is a red herring. Nothing in your statement implies impossibility.

And here I am thinking that the contrary to possible is impossible.

I didn't use it in this thread. You brought up the subject.

Did I misunderstand the point of your OP?

Since your memory of my past threads is so good, you must recall that I did provide a definition of "immaterial".

That I do not recall. Enlighten me.

 

[durangodawood]

impossible means it cant be done.
not possible means it cannot be done.

Subtle difference.

 

[Resha Caner]

So we can determine if they are not the same? If so, it shoots a hole in your claim about assumptions.

I didn't say we could determine they are the same. I said we must assume they are the same.

Heat and temperature are one in the same.

It is unfortunate it has come to this. Our conversations are usually enjoyable.

Often we think that heat and temperature are the same thing. However, this is not the case. Heat and temperature are related to each other, but are different concepts.
How does Temperature Differ from Heat?

I did notice that you took the effort to edit out my quote in post #90, which explicitly noted assumptions made in measurement. I let it go because I hoped you would change course and accept a clarification rather than digging in to make more and more blatant errors. So, what authority would convince you assumptions are required for measurement? I could add a whole list of such quotes. For example, Mary J Allen, in her Introduction to Measurement Theory devotes an entire section (3.1) to assumptions.

On page 103 of Scientific Method, A.D. Ritchie states:
The vital point of the whole matter is the nature of our first assumption H0 and its probability because it is ex hypothesi not derived from any previous term ...

Since you say you are not an expert in assessing Gaussian noise, what is H0? It is the "null hypothesis" of a statistical test. Therefore, when you refer to the 95% rule, one must ask (as we engineers always do) why 95%? The constants of nature are rarely so nicely framed, and when one sees a round number, it is a red flag that it is an assumption. So why 95%? Does that mean at 95% we are guaranteed correlation and at 94.9999999999% we are guaranteed failure? Why do some use 95% and, as you noted, others use 99.999%? Because it's an assumption of what constitutes acceptability.

It never is confirmed 100% because science is always tentative.

Why? Is science's only challenge an uncertainty principle, i.e. a wall that limits the precision of measurements?

 

[Resha Caner]

Did I misunderstand the point of your OP?

I don't know, but don't bring up an example and then say its my job to define the terms of your example.