6 ms·

Why would emails be more important than the data? The data should speak for itself, only a politician is looking for "gotcha" phrases they can print in huge let

by lemevi 11y ago

Why would emails be more important than the data? The data should speak for itself, only a politician is looking for "gotcha" phrases they can print in huge letters for a poster board that they put up while making speeches denying global warming. It's not that these politicians don't understand how science works, they understand how to manipulate people to favor legislation that is actually against their interests. Edit: For those that are suggesting that you need the emails to verify that scientists aren't making up the data, the data itself includes information on its providence. All the information you need to trust the data come from independent verification of the data and from information in the data itself. You don't need emails to verify research, but if you did then there's a problem with how research is published.

oberstein 11y ago
The data should speak for themselves. Unfortunately people who don't know statistics think they can't, because as we found out from emails, the data were "manipulated" via statistical methods: https://en.wikipedia.org/wiki/Climatic_Research_Unit_email_controversy https://en.wikipedia.org/wiki/Climatic_Research_Unit_email_c... Honestly I'd rather we press scientists of all sorts for all their source code and modeling tool configurations including the raw data that hasn't been post-processed to hell in an "acceptable" way (recalling that the statistics involved in generating p-values were "acceptable" for psych journals until recently).
- gaius 11y ago
  Any science based on simulation, any member of the public should be able to download, compile and run on their PC and get exactly the same results. It's the only way to be completely transparent. C and FORTRAN compilers are free.
  - mrow84 11y ago
    It's important to recognise that a lot of the model runs are done on super-computers, and the runs themselves aren't always easy to replicate from a practical point of view, because of things like resource limitations and configuration issues. I certainly agree that things should be open sourced, probably allowing for some kind of embargo system for publication, but we shouldn't allow a rhetoric to develop suggesting that people should be making things easily replicable, because that essentially means making very portable and resource-flexible code, which is not trivial, even in less complex domains, as I'm sure many people on this forum recognise. As an example, many models make use hundreds of cores via MPI (along with preposterous amounts of RAM), and whilst I am no expert, I get the impression that porting such things to more commodity hardware is essentially impossible. As an aside, I have seen more bad-faith practice (i.e. manipulation of results) in industry than in science, and have spent approximately equal periods of time in both.
    - gaius 11y ago
      Today's supercomputer is a GPU in 5 years time.
      - mrow84 11y ago
        Maybe in terms of various metrics (though I'm not so certain things are progressing that quickly), but not in terms of source compatibility, which was the point I was trying to make.
  - dragop 11y ago
    You can find NOAA's current climate model here, with source code, compile scripts, and sample data: http://www.gfdl.noaa.gov/cm2-5-and-flor http://www.gfdl.noaa.gov/cm2-5-and-flor However, as mrow84 points out, you will need a supercomputer, unless you have a lot of time on your hands.
- dragop 11y ago
  No evidence was found of wrong-doing at the Climate Research Unit. The only result was it provided a source of comments to be taken out of context by climate sceptics. Which is exactly what Lamar Smith is hoping to achieve here.
SilasX 11y ago
In theory, yes. Among ideal rational agents, yes. But in practice, science depends on the trust that scientists aren't actively subverting the process: failing to consider alternate hypothesis, fabricating data, ignoring data, deliberately filtering for confirmatory data, groupthink, etc. So an import safeguard is the ability to detect when someone is doing the above, even if the detector doesn't know the literature well enough to refute it. With that said, that doesn't mean there's enough evidence of misconduct to justify a subpoena in this case; I'm just saying, it's not simply a matter of "just look at the data rather than the people".
- gmarx 11y ago
  A subpoena doesn't require evidence of misconduct. As I understand it congress asked one of their agencies to provide information and they refused. I think subpoena is the legal way of saying, your refusal is invalid. Give us the information
- snowwrestler 11y ago
  While it would certainly make science more efficient, science as a whole does not depend on colleagues within one program working perfectly. Science is global endeavour based on data. If one scientists is lax or fraudulent, there are numerous incentives for competing scientists to prove so.
- cracker_jacks 11y ago
  Then have a separate group of scientists reproduce the results and review the methodology used. Their publication should provide all the information necessary to reproduce and evaluate their findings. All the possible 'evidence' of misconduct will be there in the data. Using a subpoena to threaten scientists for finding results you don't agree with is a form of censorship.
  - sanderjd 11y ago
    Unless the data collection itself is biased. The solution to that is to have another team collect its own data and compare its findings. But now you once again have scientists doing science, and if those scientists confirm the other findings, then they might also have used biased methods, so you once again have to either trust them, or not. So to build trust, yet more scientists have to collect and analyze yet more data. But what if those scientists are also in on the bias? So maybe at this point, you hire your own scientists that you trust implicitly because you're the one paying them. Maybe their findings contradict all the earlier findings, and so you trust that. But now nobody else trusts their findings, because you paid them for the study, so people think they are probably biased toward your perspective. This is literally what has been going on for years. So now what? How do you collect and analyze data when trust in those doing the collection and analysis is not a given? One approach is to implicitly trust scientists, because they are experts in their field, while you are a novice. This is what a lot of us do, and I think it's a perfectly fine solution, but it is a trade-off; you might misplace your trust. Another approach is to train up on the science yourself, so that you can independently verify things. This is a great solution, but it isn't scalable. Most people have other jobs and other interests, and in any case, you can never be an expert in everything that matters. Another approach might be to try to build trust using out-of-band signals, like transparent communication. Transparency often sounds nice, but it also has a major chilling effect, because people communicate more efficiently in the sorts of candid conversations that are hard to have publicly. Personally, I think the witch hunt for biased climate scientists is disgusting and beside the point, but I also find it distasteful to suggest that it isn't a really hard problem to disseminate trusted information about complex subjects in a highly specialized society.
moonchrome 11y ago
>The data should speak for itself Are you serious ? Not only is there multiple way to interpret same data (after all there are so many competing models to explain it) and at best you can debate confidence intervals and models - it's far beyond non-experts expecting them to review the data directly is ridiculous - the only way they can work with experts is based on trust and credibility.
- acqq 11y ago
  Wrong. You don't have to be a rocket scientist to repeat most of the basic calculations which give similar results.
  - moonchrome 11y ago
    Who cares if I can run a formula/program that tells me Y from X if I can't understand/trust the program/formula ? If you think that stuff like economics/climatology/etc. has the same level of rigor and validation applied to it as established and verified theories from hard science then boy do I have some bad news for you (struggling for SD 2 as ideal vs SD 5 for eg.).
    - acqq 11y ago
      I see you don't want to trust even yourself in actually learning the formulas and the basic principles of physics behind them, but that's how it is. Physics works. Humans produce immense amounts of CO2, CO2 in the atmosphere increases the temperature of the planet. It's as simple as that. Reading other people's mails won't refute the physics. And no, other factors aren't the answer: http://www.bloomberg.com/graphics/2015-whats-warming-the-world/ http://www.bloomberg.com/graphics/2015-whats-warming-the-wor...
      - moonchrome 11y ago
        >Humans produce immense amounts of CO2, CO2 in the atmosphere increases the temperature of the planet. It's as simple as that. CO2 is a green house gas - good job mr. high-school physicist - now show me the temperature projections along with human CO2 production graphs and estimated impacts and error margins and past model prediction performances vs measured data. You know the actual things that matter for policy implications. See that's where things go beyond high school.
tptacek 11y ago
The emails aren't more important than the data, but that doesn't matter: these are government employees communicating on government servers about government work, and whether we like it or not (in this instance, I don't!), Congress is entitled to the messages.

by lemevi 11y ago

oberstein 11y ago

The data should speak for themselves. Unfortunately people who don't know statistics think they can't, because as we found out from emails, the data were "manipulated" via statistical methods: https://en.wikipedia.org/wiki/Climatic_Research_Unit_email_controversy https://en.wikipedia.org/wiki/Climatic_Research_Unit_email_c... Honestly I'd rather we press scientists of all sorts for all their source code and modeling tool configurations including the raw data that hasn't been post-processed to hell in an "acceptable" way (recalling that the statistics involved in generating p-values were "acceptable" for psych journals until recently).

gaius 11y ago
Any science based on simulation, any member of the public should be able to download, compile and run on their PC and get exactly the same results. It's the only way to be completely transparent. C and FORTRAN compilers are free.
- mrow84 11y ago
  It's important to recognise that a lot of the model runs are done on super-computers, and the runs themselves aren't always easy to replicate from a practical point of view, because of things like resource limitations and configuration issues. I certainly agree that things should be open sourced, probably allowing for some kind of embargo system for publication, but we shouldn't allow a rhetoric to develop suggesting that people should be making things easily replicable, because that essentially means making very portable and resource-flexible code, which is not trivial, even in less complex domains, as I'm sure many people on this forum recognise. As an example, many models make use hundreds of cores via MPI (along with preposterous amounts of RAM), and whilst I am no expert, I get the impression that porting such things to more commodity hardware is essentially impossible. As an aside, I have seen more bad-faith practice (i.e. manipulation of results) in industry than in science, and have spent approximately equal periods of time in both.
  - gaius 11y ago
    Today's supercomputer is a GPU in 5 years time.
    - mrow84 11y ago
      Maybe in terms of various metrics (though I'm not so certain things are progressing that quickly), but not in terms of source compatibility, which was the point I was trying to make.
- dragop 11y ago
  You can find NOAA's current climate model here, with source code, compile scripts, and sample data: http://www.gfdl.noaa.gov/cm2-5-and-flor http://www.gfdl.noaa.gov/cm2-5-and-flor However, as mrow84 points out, you will need a supercomputer, unless you have a lot of time on your hands.
dragop 11y ago
No evidence was found of wrong-doing at the Climate Research Unit. The only result was it provided a source of comments to be taken out of context by climate sceptics. Which is exactly what Lamar Smith is hoping to achieve here.

SilasX 11y ago

In theory, yes. Among ideal rational agents, yes. But in practice, science depends on the trust that scientists aren't actively subverting the process: failing to consider alternate hypothesis, fabricating data, ignoring data, deliberately filtering for confirmatory data, groupthink, etc. So an import safeguard is the ability to detect when someone is doing the above, even if the detector doesn't know the literature well enough to refute it. With that said, that doesn't mean there's enough evidence of misconduct to justify a subpoena in this case; I'm just saying, it's not simply a matter of "just look at the data rather than the people".

gmarx 11y ago
A subpoena doesn't require evidence of misconduct. As I understand it congress asked one of their agencies to provide information and they refused. I think subpoena is the legal way of saying, your refusal is invalid. Give us the information
snowwrestler 11y ago
While it would certainly make science more efficient, science as a whole does not depend on colleagues within one program working perfectly. Science is global endeavour based on data. If one scientists is lax or fraudulent, there are numerous incentives for competing scientists to prove so.
cracker_jacks 11y ago
Then have a separate group of scientists reproduce the results and review the methodology used. Their publication should provide all the information necessary to reproduce and evaluate their findings. All the possible 'evidence' of misconduct will be there in the data. Using a subpoena to threaten scientists for finding results you don't agree with is a form of censorship.
- sanderjd 11y ago
  Unless the data collection itself is biased. The solution to that is to have another team collect its own data and compare its findings. But now you once again have scientists doing science, and if those scientists confirm the other findings, then they might also have used biased methods, so you once again have to either trust them, or not. So to build trust, yet more scientists have to collect and analyze yet more data. But what if those scientists are also in on the bias? So maybe at this point, you hire your own scientists that you trust implicitly because you're the one paying them. Maybe their findings contradict all the earlier findings, and so you trust that. But now nobody else trusts their findings, because you paid them for the study, so people think they are probably biased toward your perspective. This is literally what has been going on for years. So now what? How do you collect and analyze data when trust in those doing the collection and analysis is not a given? One approach is to implicitly trust scientists, because they are experts in their field, while you are a novice. This is what a lot of us do, and I think it's a perfectly fine solution, but it is a trade-off; you might misplace your trust. Another approach is to train up on the science yourself, so that you can independently verify things. This is a great solution, but it isn't scalable. Most people have other jobs and other interests, and in any case, you can never be an expert in everything that matters. Another approach might be to try to build trust using out-of-band signals, like transparent communication. Transparency often sounds nice, but it also has a major chilling effect, because people communicate more efficiently in the sorts of candid conversations that are hard to have publicly. Personally, I think the witch hunt for biased climate scientists is disgusting and beside the point, but I also find it distasteful to suggest that it isn't a really hard problem to disseminate trusted information about complex subjects in a highly specialized society.

moonchrome 11y ago

>The data should speak for itself Are you serious ? Not only is there multiple way to interpret same data (after all there are so many competing models to explain it) and at best you can debate confidence intervals and models - it's far beyond non-experts expecting them to review the data directly is ridiculous - the only way they can work with experts is based on trust and credibility.

acqq 11y ago
Wrong. You don't have to be a rocket scientist to repeat most of the basic calculations which give similar results.
- moonchrome 11y ago
  Who cares if I can run a formula/program that tells me Y from X if I can't understand/trust the program/formula ? If you think that stuff like economics/climatology/etc. has the same level of rigor and validation applied to it as established and verified theories from hard science then boy do I have some bad news for you (struggling for SD 2 as ideal vs SD 5 for eg.).
  - acqq 11y ago
    I see you don't want to trust even yourself in actually learning the formulas and the basic principles of physics behind them, but that's how it is. Physics works. Humans produce immense amounts of CO2, CO2 in the atmosphere increases the temperature of the planet. It's as simple as that. Reading other people's mails won't refute the physics. And no, other factors aren't the answer: http://www.bloomberg.com/graphics/2015-whats-warming-the-world/ http://www.bloomberg.com/graphics/2015-whats-warming-the-wor...
    - moonchrome 11y ago
      >Humans produce immense amounts of CO2, CO2 in the atmosphere increases the temperature of the planet. It's as simple as that. CO2 is a green house gas - good job mr. high-school physicist - now show me the temperature projections along with human CO2 production graphs and estimated impacts and error margins and past model prediction performances vs measured data. You know the actual things that matter for policy implications. See that's where things go beyond high school.

tptacek 11y ago

The emails aren't more important than the data, but that doesn't matter: these are government employees communicating on government servers about government work, and whether we like it or not (in this instance, I don't!), Congress is entitled to the messages.