Donald Trump and US politics

"

Wraeclastian wrote:

"

pneuma wrote:

I'll certainly say that if you think everyone in science is in it for "ultimate truths", you'll be sorely mistaken. It's very hard to find people like that in science... or anywhere.

Actually, those people aren't very hard to find though. They are most often the ones labeled as "crazy" or "conspiracy theorists" if they haven't already died due to suspicious circumstances.

"

ChanBalam wrote:

Another stupid generalization based on personal biases. Do you have any actual evidence for that claim?

"

Laurium wrote:

That bold part DL is talking about extends far beyond any particular discipline. Academic/Published research suffer from this problem of being reproducible. It's especially prevalent in computer science.

As an example, suppose I write a subroutine for a machine learning algorithm, and test it against some large database. In terms of being able to reproduce the results, I would need: the raw data, the source code for the subroutine, the source code for the statistical software tests, etc.

You can imagine all of this is neither volunteered nor readily available. Not sure if it's absolutely true in the instance of climate science as DL says, but in the abstract it's a general problem among high volume data, high complexity research. Stated differently, many times when you read through a paper you're forced to take the author(s) word for it.

"

Wraeclastian wrote:

"

ChanBalam wrote:

Another stupid generalization based on personal biases. Do you have any actual evidence for that claim?

How about you provide evidence that proves me wrong? You are the one challenging me, not the other way around.

"

ChanBalam wrote:

"

Laurium wrote:

That bold part DL is talking about extends far beyond any particular discipline. Academic/Published research suffer from this problem of being reproducible. It's especially prevalent in computer science.

As an example, suppose I write a subroutine for a machine learning algorithm, and test it against some large database. In terms of being able to reproduce the results, I would need: the raw data, the source code for the subroutine, the source code for the statistical software tests, etc.

You can imagine all of this is neither volunteered nor readily available. Not sure if it's absolutely true in the instance of climate science as DL says, but in the abstract it's a general problem among high volume data, high complexity research. Stated differently, many times when you read through a paper you're forced to take the author(s) word for it.

Thanks. The question remains as to whether the collection and interpretation of climate data over the past 20+ years suffers from the same problems faced by computer scientists doing research. Unless stifled politically, good science will figure out how to resolve its methodology problems.

"

ChanBalam wrote:

"

Wraeclastian wrote:

"

ChanBalam wrote:

Another stupid generalization based on personal biases. Do you have any actual evidence for that claim?

How about you provide evidence that proves me wrong? You are the one challenging me, not the other way around.

Ah, it was you who made the claim. As any thoughtful person knows, making claims about things requires substantiating those claims with some evidence. It is the process we use in law, medicine, insurance, common discourse etc.

You said: "Actually, those people aren't very hard to find though. They are most often the ones labeled as "crazy" or "conspiracy theorists" if they haven't already died due to suspicious circumstances." I would think that to make such a claim you would have some actual evidence to back it up.

If I said that mixing yellow and red produces green and then you disputed that, would you accept "prove me wrong" as an answer? Showing me that you can make red and yellow produce orange, doesn't prove me wrong, it just shows you don't know how to do it.

It is actually the way science works. Make a claim (hypothesis) and then show the evidence. It is also the way a rational discussion works.

Claim 1: Trump is a liar. If someone can show an instance of him lying, then the claim is true (and likely true for most of us).

Claim 2: Trump is a chronic liar. Producing multiple, ongoing cases of him lying would be evidence that this is true. It does not mean that everything he says is a lie, just that he has lied repeatedly on multiple occasions.

You made the claim. Show me that it is true.

"

Laurium wrote:

"

ChanBalam wrote:

"

Laurium wrote:

That bold part DL is talking about extends far beyond any particular discipline. Academic/Published research suffer from this problem of being reproducible. It's especially prevalent in computer science.

As an example, suppose I write a subroutine for a machine learning algorithm, and test it against some large database. In terms of being able to reproduce the results, I would need: the raw data, the source code for the subroutine, the source code for the statistical software tests, etc.

You can imagine all of this is neither volunteered nor readily available. Not sure if it's absolutely true in the instance of climate science as DL says, but in the abstract it's a general problem among high volume data, high complexity research. Stated differently, many times when you read through a paper you're forced to take the author(s) word for it.

Thanks. The question remains as to whether the collection and interpretation of climate data over the past 20+ years suffers from the same problems faced by computer scientists doing research. Unless stifled politically, good science will figure out how to resolve its methodology problems.

Right, so again I'll preface by saying I can't say for sure about climate science--not my field.

However, with a quick thought experiment I can see how climate science would be more prohibitive in this context of research being reproducible, not less, than say psychology.

Assume an endeavor to predict future climate patterns and its consequences from which political policy may or may not based off of. Further accept that all of your assumptions are true, that I can in fact have access to the raw data, results, and methodology (i.e., model construction, verified assumptions, solid kernel theory, etc.). Obviously, the only way to predict future weather patterns is with a highly sophisticated model that only a computer can process, even if it can only barely approximate the complexity of a planet creating/regulating its atmosphere.

Here's the issue: The only way said model can be run is on a supercomputer. A research capable machine is insufficient; you need the power of a SC. Thus, even if we cede all your assumptions, next to no one can reproduce the results as supercomputers are just not accessible, even for most universities.

Therefore, do we blindly accept the authors research? Maybe, maybe not. But I don't think a default stance of skepticism is all that wild.

"

ChanBalam wrote:

"

Laurium wrote:

That bold part DL is talking about extends far beyond any particular discipline. Academic/Published research suffer from this problem of being reproducible. It's especially prevalent in computer science.

As an example, suppose I write a subroutine for a machine learning algorithm, and test it against some large database. In terms of being able to reproduce the results, I would need: the raw data, the source code for the subroutine, the source code for the statistical software tests, etc.

You can imagine all of this is neither volunteered nor readily available. Not sure if it's absolutely true in the instance of climate science as DL says, but in the abstract it's a general problem among high volume data, high complexity research. Stated differently, many times when you read through a paper you're forced to take the author(s) word for it.

Thanks. The question remains as to whether the collection and interpretation of climate data over the past 20+ years suffers from the same problems faced by computer scientists doing research. Unless stifled politically, good science will figure out how to resolve its methodology problems.

"

ChanBalam wrote:

But how much of the climate change research is done on SCs? I would think only a small portion.

"

deathflower wrote:

Scientists can don't disclose their research. Scientists are also competitors and research cost money. They are competing for funding. Scientists are looking after their own interest.