Preferred word in lieu of "GW Denier"

jwoodward48 wrote:
For you, sure, skeptic. But people like IB are not skeptics, and take offense from the term.

Tim the plumber wrote: There is only one here who denies that the earth's temperature has risen since 1970. Ibedadenier.

Tim the plumber wrote: There are those, such as me, who do not deny science but I will point out the difficulties and limitations of the data.

jwoodward48 wrote:Trends. Almost all organizations that track temperature, whether independent or government-funded, show a gradual rise in temperature over the last several decades.

spot wrote:

jwoodward48 wrote:
For you, sure, skeptic. But people like IB are not skeptics, and take offense from the term.

IB takes offence at everything unless you acknowledge that he is source of all knowledge on everything.

jwoodward48 wrote:
Trends. Almost all organizations that track temperature, whether independent or government-funded, show a gradual rise in temperature over the last several decades. The chances of a particular measurement being outside the margin of error are 2.5%, but the chances of the average being outside the margin of error is 2.5% divided by the square root of the number of measurements. The margin of error is 0.05 degrees for recent years, 0.1 degrees for 1900. The measured increase in temperature is agreed by most such organizations to be about 1 degree since 1900. This exceeds the standard deviation of the difference (about 0.11 degrees) by more than ninefold. The Earth is heating up.

Into the Night wrote:
Bad math. The problem is the requirements of data used for the input stream in the first place. Statistics is very specific about the requirements of this stream.

First, statistics REQUIRES sampling to be done through selection of data by a random means independent of the data. This means the thermometers MUST be equally spaced. Weighting of data imposes bias on the data itself, even if you weight it for location in space.

Second, data MUST be compared against data not sampled (through the use of the random numbers) to generate a valid margin of error.

Third, the points of measurement along the surface is essentially infinite. The best you can do it space thermometers to some acceptable level of temperature gradient.

Fourth, that gradient has been observed to vary by as much as 20 deg F in a single mile, such as stretching that gradient across a fast moving cold front. This gradient of an infinite number of temperature points between them MUST be factored into the margin of error.

Fifth, air moves. It can move pretty fast. So does the surface of the Earth as the Earth rotates against the source of energy, the sun. This means you MUST measure all readings simultaneously, or the readings don't make any sense at all as an average.

Sixth, all instruments have a known accuracy of calibration. No thermometer is perfect. That too must be factored into the margin of error.

Increasing the number of samples is only useful if the rest of the requirements are met along with it. Whether you use one thermometer in Seattle or a thousand thermometers in Seattle, they tell you nothing about Yakima and very little about Everett.

jwoodward48 wrote:

Into the Night wrote:
Bad math. The problem is the requirements of data used for the input stream in the first place. Statistics is very specific about the requirements of this stream.

First, statistics REQUIRES sampling to be done through selection of data by a random means independent of the data. This means the thermometers MUST be equally spaced. Weighting of data imposes bias on the data itself, even if you weight it for location in space.

No, statistics does not require that. First, your second sentence does not follow the first. Random =/= perfectly even. Second, do you have a source for that? I've Googled it, and found nothing on it. I don't think you're right.

Second, data MUST be compared against data not sampled (through the use of the random numbers) to generate a valid margin of error.

I never said how they calculated their margin of error. I don't know. I don't think that the scientists intentionally messed up their science.

Third, the points of measurement along the surface is essentially infinite. The best you can do it space thermometers to some acceptable level of temperature gradient.

Of course we can't measure the temperature exactly. That's why we have margins of error.

Fourth, that gradient has been observed to vary by as much as 20 deg F in a single mile, such as stretching that gradient across a fast moving cold front. This gradient of an infinite number of temperature points between them MUST be factored into the margin of error.

That's the thing: the cold front is fast moving. So perhaps quite a few measurements are not good representatives of their location - but this won't necessarily be the case tomorrow, or even in the next hour. (Also, I was using degrees Celsius for my measurements before. This is science. Let's use metric.)

Fifth, air moves. It can move pretty fast. So does the surface of the Earth as the Earth rotates against the source of energy, the sun. This means you MUST measure all readings simultaneously, or the readings don't make any sense at all as an average.

But I thought that randomly selecting the space and time coordinates of observations would be fine!

Anomaly-measurement (not sure what the technical term is) involves taking the average daily temperature for a specific, small region over thirty years, and then measuring all data by how it varies from that baseline, for that day. (The same temperature would be warmer than average in winter and colder than average in summer.) The overall average change from the overall baseline is the average of all of the individual changes from the individual baselines.

Why do I mention this? Because the change in solar energy is already taken into account.

Sixth, all instruments have a known accuracy of calibration. No thermometer is perfect. That too must be factored into the margin of error.

Naturally, it is. That's probably the first thing they calculated when figuring out the margin of error.

Increasing the number of samples is only useful if the rest of the requirements are met along with it. Whether you use one thermometer in Seattle or a thousand thermometers in Seattle, they tell you nothing about Yakima and very little about Everett.

Not necessarily.

Let's suppose that besides the real, actual thermometers, we have magic thermometers placed wherever our real ones are. These magic thermometers have the magic property that if you take the average of every one of them, the true global temperature is the result.

If you only have one thermometer for the entire world, there's going to be a huge chance of huge error between the real and the magic thermometers. But as you increase the number of thermometers, the average difference between real and magic thermometers will go down. That's the power of big numbers: the chances of every thermometer being warmer than its magic counterpart goes down dramatically as the number of thermometers increases.

If you replace the magic instantaneous thermometers with magic average-over-time thermometers, a similar effect takes place. As the number of thermometers increases, the margin of error decreases.

TL;DR: A million thermometers would measure the Earth better than one, so increasing the number of observations improves the margin of error.

jwoodward48 wrote:
IB, just read this: http://earthobservatory.nasa.gov/blogs/earthmatters/2015/01/21/why-so-many-global-temperature-records

See? Actual data. Not made up.

Into has much better arguments. Yours are just disguised (or not disguised) ad hominems, combined with baseless claims.

jwoodward48 wrote:
IB, just read this: http://earthobservatory.nasa.gov/blogs/earthmatters/2015/01/21/why-so-many-global-temperature-records

See? Actual data. Not made up.

Into has much better arguments. Yours are just disguised (or not disguised) ad hominems, combined with baseless claims.

jwoodward48 wrote:Dogmatic... Conclusions... No data...

jwoodward48 wrote:
But it's not. The government cannot influence scientific data. On the one hand, you have scientific integrity. On the other hand, you have the same factors which prevent just anybody from making up results. If the government can get scientists to construct a hoax, why do you even trust science?

jwoodward48 wrote:
I just figured out why I hate you. You're an armchair psychologist. (Can't claim to be innocent there, but hypocrisy doesn't make me wrong.)

So, share another gem. What makes the non-US government-funded sources unusable? A worldwide conspiracy, perhaps?

jwoodward48 wrote:
Into, there is a DEGREE of inaccuracy introduced by uneven spacing, whether spatial or temporal. But you can't just ignore all data unless it comes from an infinite number of thermometers placed at every point on Earth. Error happens. Some errors can be alliviated, at least partially, by statistics. Keep in mind that statistics isn't a pure math; it deals with real-life data to produce usable results.

jwoodward48 wrote:
...how long did it take for IB's manner to get old? A few hours? A few weeks?

jwoodward48 wrote:
Wait, no, it didn't. Double standards anyone? I'm only doing tit-for-tat with him.

Into the Night wrote:

jwoodward48 wrote:
Into, there is a DEGREE of inaccuracy introduced by uneven spacing, whether spatial or temporal. But you can't just ignore all data unless it comes from an infinite number of thermometers placed at every point on Earth. Error happens. Some errors can be alliviated, at least partially, by statistics. Keep in mind that statistics isn't a pure math; it deals with real-life data to produce usable results.

Statistics is pure math. It has rules just like any other part of math. It uses values from real-life situations just like any other part of math.

I guess you fell asleep in 2nd and 3rd grade, when they started introducing story problems.

You talk about ignoring data. You are ignoring the requirements of statistics. You are also ignoring the all important margin of error and where it comes from.

All I can say is...stay out of Vegas.

jwoodward48 wrote:

Into the Night wrote:

jwoodward48 wrote:
Into, there is a DEGREE of inaccuracy introduced by uneven spacing, whether spatial or temporal. But you can't just ignore all data unless it comes from an infinite number of thermometers placed at every point on Earth. Error happens. Some errors can be alliviated, at least partially, by statistics. Keep in mind that statistics isn't a pure math; it deals with real-life data to produce usable results.

Statistics is pure math. It has rules just like any other part of math. It uses values from real-life situations just like any other part of math.

I guess you fell asleep in 2nd and 3rd grade, when they started introducing story problems.

You talk about ignoring data. You are ignoring the requirements of statistics. You are also ignoring the all important margin of error and where it comes from.

All I can say is...stay out of Vegas.

Statistics is not a pure math.

jwoodward48 wrote:
Physics is not mathematics.

jwoodward48 wrote:
It uses mathematics as a languge and framework.

"Applied mathematicians focus on the real-world uses of mathematics. Engineering, economics, physics, finance, biology, astronomy—all these fields need quantitative techniques to answer questions and solve problems. Pure mathematics, by contrast, is mathematics for its own sake."

That is, applied math answers questions with the data that we have. These questions come from engineering, economics, physics, finance, biology, astronomy, etc., but those fields are not mathematics. Pure math is more rigorous and a bit less relevant to the real world - it still is, but not in an obvious way.

In 2nd grade, they gave me applied mathematical problems (how many apples) along with pure mathematical problems (2*4=?).

However, it appears that you do not understand the pure/applied division of mathematics. (Admittedly, it is a fuzzy line, but it exists.)

Applied mathematics is motivated by the real world. "Pure mathematics is not about applications. It's not about the "real world." It's not about creating faster web browsers, or stronger bridges, or investment banks that are less likely to shatter the world economy."

In fact, read all of this. It describes how research is a "dialogue" (that sounds like consensus!)

Statistics is very much applied mathematics. It doesn't exist in a vacuum. It gets real, flawed data and needs to produce results and answers to our questions. "THERE IS AS YET INSUFFICIENT DATA FOR A MEANINGFUL ANSWER" depends on your definition of meaningful. Our margins of error are quite good - the Earth is warming.

Into the Night wrote:

jwoodward48 wrote:

Into the Night wrote:

jwoodward48 wrote:
Into, there is a DEGREE of inaccuracy introduced by uneven spacing, whether spatial or temporal. But you can't just ignore all data unless it comes from an infinite number of thermometers placed at every point on Earth. Error happens. Some errors can be alliviated, at least partially, by statistics. Keep in mind that statistics isn't a pure math; it deals with real-life data to produce usable results.

Statistics is pure math. It has rules just like any other part of math. It uses values from real-life situations just like any other part of math.

I guess you fell asleep in 2nd and 3rd grade, when they started introducing story problems.

You talk about ignoring data. You are ignoring the requirements of statistics. You are also ignoring the all important margin of error and where it comes from.

All I can say is...stay out of Vegas.

Statistics is not a pure math.

Statistics is pure math. There are no components of statistics or its rules that exist outside of mathematics.

jwoodward48 wrote:
Physics is not mathematics.

True. None of science is mathematics. It is why science by itself does not have the power of prediction. It must turn to mathematics, which does have that power, for that.

jwoodward48 wrote:
It uses mathematics as a languge and framework.

"Applied mathematicians focus on the real-world uses of mathematics. Engineering, economics, physics, finance, biology, astronomy—all these fields need quantitative techniques to answer questions and solve problems. Pure mathematics, by contrast, is mathematics for its own sake."

That is, applied math answers questions with the data that we have. These questions come from engineering, economics, physics, finance, biology, astronomy, etc., but those fields are not mathematics. Pure math is more rigorous and a bit less relevant to the real world - it still is, but not in an obvious way.

In 2nd grade, they gave me applied mathematical problems (how many apples) along with pure mathematical problems (2*4=?).

However, it appears that you do not understand the pure/applied division of mathematics. (Admittedly, it is a fuzzy line, but it exists.)

Applied mathematics is motivated by the real world. "Pure mathematics is not about applications. It's not about the "real world." It's not about creating faster web browsers, or stronger bridges, or investment banks that are less likely to shatter the world economy."

In fact, read all of this. It describes how research is a "dialogue" (that sounds like consensus!)

Statistics is very much applied mathematics. It doesn't exist in a vacuum. It gets real, flawed data and needs to produce results and answers to our questions. "THERE IS AS YET INSUFFICIENT DATA FOR A MEANINGFUL ANSWER" depends on your definition of meaningful. Our margins of error are quite good - the Earth is warming.

Statistics is pure mathematics. Its rules do not depend on data. The reason it works does not depend on data.

The same is true for probabilities and the definition and generation of random numbers.

The same is true for algebra, calculus, trigonometry, or even what a binomial operator means.

Since you have demonstrated you do not know statistics, and have rejected any discussion of what it is and what it does, making a blind claim that the margin of error is 'good' is bullshit.

jwoodward48 wrote:
If you'd like to create new terms, Into, that's fine. But pure and applied mathematics already are terms with existing meanings.

It turns out that you're sort of right, in a sense:

"Theoretical statistics is all about justifying mathematically the methods that applied statisticians use. In other words, proving rigorously that the methods of applied statistics are good."

But I am referring to applied statistics:

"Applied statistics is like science without any particular subject matter: how to collect data, make sense of it, quantify error in measurement, and build models to predict future observations."

In applied statistics, you can't just throw out data because it's not continuously measured. It's about "making sense of data." You can't be mathematically snooty about the data - we need a conclusion and a margin of error. Applied statistics, like science, needs useable results, even if the result is a 2C margin of error - just saying "the observations were not randomly placed so statistics will just throw up its hands and give up" will not cut it.

jwoodward48 wrote:
I'm not separating them. The WORLD is. The scientific world. Don't blame me.

Besides, what rules?

jwoodward48 wrote:
What rules?! What. Rules. I don't recall any rule in statistics that says that all data must come from evenly-spaced points.

jwoodward48 wrote:
" Almost every sample in practice is biased because it is practically impossible to ensure a perfectly random sample."

You are describing selection bias, which exists almost everywhere in greater or smaller amounts. The mere existence of selection bias does not invalidate the data. It merely increases the chances of the data misrepresenting reality. As more and more independent measurements of GW are done, the chances of GW being the result of error decrease. Selection bias is just another source of error, and we have error already. It just means we have to work a bit harder to get a good confidence interval.

jwoodward48 wrote:
Also,

"Statistics is pure mathematics. Its rules do not depend on data. The reason it works does not depend on data."

You just described theoretical statistics!

You understand a kind of statistics, good for you.

But don't call me an idiot for understanding another type. Statistics has ways of dealing with error. Usually, it's known as "get more data." We have more data, ergo we can deal with the error from sampling bias. All data is flawed. You can't throw out any data that doesn't meet your ridiculously impractical and unfounded requirements.