CRU's "very artificial correction for decline" is dead code
jgc.org
jgc.org
And the story just keeps getting better and better, too. Lots of twists and turns.
So now we have code comments which are incendiary put ahead of, as it turn out, dead code. It's as if we found IRS code that had pieces in it like "This part makes sure all environmentalists get audited" but then the next part never runs. So it looks awful, but it does nothing.
So I guess that's the key for any kind of code analysis, right? What does it do? You run the code. Has anybody taken the raw input and ran the code to make sure this piece of code is currently representative of the code used at the time?
No wait -- we can't do that, can we? Because the data was lost/destroyed, right? So we have some version of some abstract pieces of code with incriminating comments in them that aren't being ran. On one hand, if you're looking for some kind of solid proof of scientists gone wild, you're never going to find it -- no matter how many times you add up nothing and nothing you'll get nothing -- because there are data and reproducibility problems that we can never resolve So it's always going to be he-said, she-said. On the other hand, that's been the problem with this issue all along -- data, models, and reproducibility. This case just brings it all out into the glorious sunshine for programmers like us to view.
Outstanding.
"As other have repeatedly pointed out, that code was written to be used for some kind of presentation that was false. The fact that the deceptive parts are commented out now does not change that at all.
It might get them off the hook if we knew — for certain — that it had never been shown to anyone who didn’t know beforehand how the data was cooked and why. But since these peiple have conveniently lost or destroyed primary datasets and evaded FOIA requests, they don’t deserve the benefit of that doubt. We already know there’s a pattern of evasion and probable cause for criminal conspiracy charges from their own words."
In other words, he doesn't know a damn thing. I can't help but feel that the anti-warming crowd (which includes ESR if I remember correctly, he wrote some long rant about peak oil?) is trying to blow a tiny mite up to be an elephant. It's typical revisionist behavior, just like the creationists who find a small bone that is not where biologists have predicted it would be, and they think they have refuted Darwin.
In any case, this all seems to be very ad hominem (revisionists trying to discredit scientists instead of providing data). Even if somewhere out there in the vast internet a false chart is floating around, it does not refute global warming. (Note: I don't claim that global warming is real - I don't know. I just don't like the style of this "mite-gate").
The way to resolve this seems to create some proper charts off proper data.
I've come to believe that AGW probably exists but am not married to the idea either - but nor am I convinced we need massive corrective policy versus adaptation (heck - I live in Canada after all, and after last winter we could use a little warming). That said, and I'm not saying you're doing this but many proponents of AGW are attempting to minimize the implications in the emails but their impact cannot and should not be denied: http://www.guardian.co.uk/environment/georgemonbiot/2009/nov...
Also - given the lack of transparency and the emails that have now come to light, the researchers most certainly open themselves up to questions that include whether or not the underlying data has been manipulated/adjusted such that the coding adjustments are no longer necessary http://wattsupwiththat.com/2009/11/26/mcintyre-data-from-the...
The whattsupwiththat thing was discusses in another thread, again I don't think it is all it is cracked up to be. It is just anti-warming people jumping at a chance, addressing the dumb public who doesn't understand scientific jargon or processes.
If this leads to a better grasp of the public about this, it might at least be some benefit. I am not claiming that academia is perfect these days.
It might as well be. It wouldn't be an exaggeration to point out that the consequences of basing political and economic decisions on the content of those papers will run into the trillions of dollars.
If there were ever a field in which it's important to avoid even the appearance of impropriety, climatology is it.
Also, this is pure speculation. From ESR's article I gather that it is not even known how and when the data was used. They could not yet point to an actual published article.
And, who are these people, are they like the most famous climate researchers ever, the Einsteins of climate research? Are they the only climate researchers on the planet? I can't imagine that they have that much influence.
If you claim P=NP, and I show a flaw in your proof, am I also required to provide a proof of my own? Of course not.
This is how science should work; both scientists, their competitors, and anyone else with a passing interest will carefully try to discredit existing work. In this case, it looks like a specific attempt to discredit it failed.
You are correct that creating proper charts off proper data would be the best solution. Unfortunately, much of the data has been "lost", making this impossible.
I simply used an example most people here would be familiar with; my broader point remains unchanged for data analysis. One can legitimately critique an analysis of data without providing new data or even performing a correct analysis.
So far I also haven't seen any "flaws in proofs". I have only seen words like "trick, milk data" that can or can not mean bad things - take your pick depending on what you want to believe.
This is what happens in the evolution debate on a regular basis: someone comes up with some minor technical complaint about peppered moth observation, spends a long time debating it, and then declares that evolution is "in controversy" despite the fact that even if their complaints were valid, there are thousands of other pieces of evidence in favor of the larger theory.
I don't know enough about global warming to know if its critics are merely hacking away at a weak sapling to draw our attention away from the dense forest that they can't cut their way through. But I know that tactic has been applied in the past, so I don't think it's unreasonable to wonder.
Scientists tend to get caught up arguing over whatever particular claim is being criticized, but I think in this situation, it would serve the climate scientists well to point the public eye towards less vulnerable data and claims, if they're out there...
It is like physics: millions of buildings have been built employing Newtons laws. Sometimes an architect gets it wrong and the building falls down. But that doesn't refute the laws of physics - it is just some guy or a team of builders making an error.
However, fraud and errors are different. Science is (to a great extend) based on trust. When I referee a paper, I trust that Fig 1 is really generated by Algorithm 2 if the caption says so. If it turns out that Fig 1 is photoshopped, my report is meaningless. Peer review is meant to find mistakes and evaluate importance, not eliminate fraud.
If a scientist cheats, we must consider all their work to be flawed until they can prove otherwise (i.e., release all data, code, etc). We need to stop and scrutinize all the fruit of their poisoned tree.
AGW may be true, and the science may be solid. But until we have more openness, we will never really know.
No, it's easier to assume trust, but science works best when there is almost zero trust in any single authority. This is why it's expected for you to present your methods, data, math, and then your conclusions. Ignoring malice a tiny mistake can drastically change an experiments outcome. So an experiment yet to be independently verified is next to worthless, but it's not sexy to work like that and it does not help your reputation.
Edit: IMO, what's missing from this discussion is the idea that independently verifying their results is a reasonable thing to do. This does not mean plugging their numbers into their code and looking at the results but starting from scratch using their methods and testing it yourself.
PS: One of the reasons that social science has such a poor reputation is the willingness to accept minuscule sample sizes as representing real research. A little less trust would go a long way to revitalizing those fields. Unfortunately neuroscience seems to be falling for this trap.
In reality, referees do trust that authors are making a good faith effort to present their results accurately (except for a little bit of cherrypicking and glossing over messy parts). As a referee, I'm simply not paid enough to search for fraud.
To search for mistakes, I skim a paper, think "yeah, that graph looks about right", [1] and give an opinion.
To search for fraud, I would need the entire source tree for the paper: data, programs, etc. I'll need to verify that the source code does what the paper says, and that the output of the program is really the source of the graphs. (This ignores the issue of fraudulent data, but that's really tricky to find.)
Since the data was destroyed (oops, I mean lost), this simply can't be done.
[1] I also check the proofs for mistakes, but that's irrelevant for this discussion.
Also I don't think this was the only data set in the entire world that relates to global warming. So it hardly invalidates all climate research if some random researcher formats his hard drive.
As for means, isn't the whole fossile fuel industry supposed to be behind the warming skeptics? They should have shitloads of money. In fact I wonder if by now they have their own research stations in the arctic, drilling for historic ice. I mean, they should? They really can not afford to buy their own thermometers?
I think they would probably have enough money to buy dozens of universities dedicated to climate research.
Climate science appears to have a lower set of standards.
Some of the CRU emails criticize climate journals which were considering adopting a "release all data" policy, and argued for not submitting papers to those journals. Other CRU emails discuss deleting data to prevent Steve McIntyre from getting it, and conspire not to release the data to him or other skeptics. This is why climategate is such a big deal.
As for other data sets, there are a few others. However, the CRU is one of the major sources of paleoclimate data. They aren't the only game in the world, but they are a big game. Both their data, and various analysis which are based on their data, are now inherently suspicious.
To make an analogy, imagine that a major open source regex library (maybe PCRE) is completely tainted (e.g., a rootkit). How large an impact would that have on open source software?
http://www.timesonline.co.uk/tol/news/environment/article693...
I see some emails that give a strange impression. How about investigating more details before calling it fraud? ESR can not even point to an actual application of the alleged corrections. I think for fraud we would need a little more.
Note I don't say it wasn't - I don't know anything about the people involved, their standing or their research. But I believe in the good old innocent until proven guilty. I think it is one of the cornerstones of civilization.
OK, but kinda the point of this is that these scientists themselves attempted to discredit other scientists who disagreed with them, rather than refuting them through data. That lays them open to ad hominem attack.
Phil Jones, the director of the East Anglia climate center, suggested to climate scientist Michael Mann of Penn State University that skeptics' research was unwelcome: We "will keep them out somehow -- even if we have to redefine what the peer-review literature is!"
Yep. That's a great summary of the contents of the emails.
"; Apply a VERY ARTIFICAL correction for decline!!"
Folks, please!!! So it is known (from the other emails about tree-ring data not corresponding with thermometer readings after the 1960s or something) that some correction is needed. So it seems plausible that a researcher starts with some crude correction to test things out, before they take the time to find the proper adjustments. This is probably just some throwaway code (like approximating pi = 3,14 before you take the time to make an algorithm for calculating Pi to higher precision). I don't think it ended up in a chart that was presented to Obama in person.
The whole case looks very, very weak, and to be honest, I just lost a whole lot of respect for ESR. Climate change be true or not, I don't like brainless mob behavior.
I think the problem again is with the wording, "artificial" can mean different things - again take your pick depending on what you want to believe.
In the end, I don't see how "artifical charts" like that could end up in scientific papers. Surely if you present a chart, you also have to present the way you derived it, at which point any fakery should become evident. If no such process is included, it simply is a weak paper.
Though now that you mention it, that seems like it would save a heck of a lot of work.
Surely if you present a chart, you also have to present the way you derived it, at which point any fakery should become evident.
If nothing else, this scandal is exposing the difference between the dogmatic version of science we all learned in the Scientific Method lecture back in 6th grade, and science as it is actually practiced in the real world. The Scientific Method lecture says that all results must be reproduced by experts. In the real world, the computer programs involved are outrageous hacks, the data are practically never seen after publication (and go missing with disturbing regularity), and the peer review concentrates mostly on making sure you cite the right people and scratch the right backs. Nobody is ever going to try to actually duplicate the math it took to transform your raw data into your massaged data -- that is hard work and have you ever met an engineer who would reliably leave detailed instructions to exactly reproduce 6 months of labor, in a form where they would be instantly comprehensible to anyone in the field? I would be astounded if scientists routinely actually committed to paper an exhaustive list of all the software installed on the computer they did the math on. (What version of which library, etc etc. So you can reproduce bugs.)
Then there is the "Frustrate anyone who opposes us at any turn, destroy data in response to FOI requests, get our nemeses expelled from their institutions, and ruin journals which publish competing work" angle, which takes it from "science as it is practiced in the real world" to just farce.
If anything, the blame should then go to the state departments responsible for funding the research. If they are being blinded by such crap, it is a problem.
That said, I think succeeding in academia requires good skills in politics, I have a lot of respect for some people who manage to do well both in research and in pulling the right strings.
This shouldn't be necessary, of course. The raw data and a description of the algorithms (with sufficient latex-y equations) should suffice. If not, I don't think the paper is worth much.
I think, though, this is what you meant by:
"Nobody is ever going to try to actually duplicate the math it took to transform your raw data into your massaged data -- that is hard work"
But that is exactly what must happen for anyone to claim they have reproduced the original results. If you just get a VMWare instance of the original author's computer, run his program and get the same results you still are no closer to knowing whether he made a mistake or not.
Defending bad science because you agree with its conclusions is not a good idea. I'm sure there's good climate science out there, and I suspect it can show a connection between human endeavors and the actual climate. But if it turns out the evidence just isn't there yet, at least with this bunch at the wheel, well then let's suck it up and accept that, rather than circle the wagons and defend "science" when it's clear the scientists in question have not acted in accord with their professed principles.
Basically there is already existent a lot of evidence that many scientists disagree strongly about Climate Change. People who have doubts should not need to lower their arguments to a shrill 'Aha I see a smoking-gun'.
This morsel of evidence is tiny by comparison to more eminent disagreement.
---
For more weighty disagreements see amongst many more:
S McIntyre and R McKitrick 'Corrections to the Mann et a. (1998) proxy data base and northern hemishpere average temperature series' [1]
as Nigel Lawson[2] says: 'it turned out ... that crucial data for the period before 1421 was based on.. one single pine tree'
-----
[1] http://www.uoguelph.ca/~rmckitri/research/trc.html
[2] p126 note 39 of A Call to Reason: A Cool Look at Global Warming by Nigel Lawson. The 2009 release has lots of up to date information in a foreword
I think many misrepresents the level of conflict. There are plenty of people that think man made impact on Climate is negligible. However, most people with a strong and relevant scientific background acknowledge, the heat island effect around city's and or desertification which are less controversial man made impacts on local Climate.
CO2's absorption spectrum is different than other atmospheric gasses and it's low over all percentage in the atmosphere magnifies it's importance. And there is little debate over the quantity of CO2 released by human activity's. But, quantifying the effect of CO2's impact on global climate with high precision is hard, so there is debate about the magnitude of it's impact.
The book mentioned has quite up-to-date info.
Maybe some of the climate change scientists got caught up in the politics of the situation and inappropriately exaggerated matters. But consider this: every single national academy of science in the world agrees the basic theory of global warming is correct. That's pretty overwhelming.
So there are two possibilities: it's all a big hoax by scientists, or it's nearly certain, but an unwarranted level of doubt has been fueled by, well, the industries that want to keep burning fuel... So what, really, sounds more likely?
Let's say there's only an 80% chance the theory is correct. If it's not, then we're screwed anyhow. If it is, we might just be able prevent a whole lot of trouble if we act soon. How certain do we have to be?
That said, finding this was a big deal: http://camirror.wordpress.com/2009/11/26/new-the-deleted-dat...
The linked article also claims that the NOAA "deleted" the post-1960 data from a data set (briffa2001jgr3.txt) put together for an article that doesn't use these data. In other words, what really happened, is that Briffa and his collaborators created a data set only containing the data they used, and this derived data set was archived by NOAA. That's hardly nefarious or even the least bit worrisome.
Now to the heart of the issue: Why was the post-1960 not used after 1998/2000? Why do all of the e-mails talk about "hiding the decline" and the code talk about stopping in 1960 to "avoid the decline"? The answer, it turns out, is right in the very Briffa 1998 paper your link mentions (but fails to cite). It's title is Reduced sensitivity of recent tree growth to temperature at high northern latitudes and it's about how temperatures reconstructed using Briffa's MXD (maximum latewood density) data diverge from actual temperature readings after 1960. The implication of the word "reconstructed" is very important to understand: Briffa's MXD data don't show an actual decline in temperature. They show a decline in a particular measurement of tree-ring density that had, until 1960, been a good indicator of actual temperature. All that's going on here is that data that is known to be inaccurate is not being used.
It's no big deal.
This was the first time that the data was available. As McIntyre points out in the top of his post, the decline was discussed in the 1998 and 2001 Briffa papers.
Of course, scientists could not actually evaluate this because the data was not released. But that's not the issue.
The issue is, as McIntyre also points out, that the IPCC 2001 graph (and later graphs) simply truncated the climate reconstruction data in 1960 and pasted the (more accurate) direct temperature station measurement data onto the graph after 1960.
While this was done it plain sight, it's still lousy scientific practice. Why? After all, shouldn't more accurate data be used? The problem is that comparison of temperature reconstructions with direct measurement tells you a lot about the accuracy of your reconstruction.
In this case it tells you a lot about the inaccuracy of the reconstruction and the sorts of signals that are missed.
I'm sorry, the paleoclimate reconstructions available in the literature are not convincing. They often use dodgy statistical techniques, mixing data gathering steps with data correlation steps to increase apparent agreement with the instrument record. There have been huge problems with releasing data and there are often bizarre choices made in proxy record selection.
Without good paleoclimate reconstructions, it's hard to know how far outside (or inside) the historical standard deviation for temperature we are. I happen to think it's an important question to have answered. A trillion dollar question.
Any data that you receive from another scientist with the intention of reproducing their experiment is fundamentally tainted and violates the scientific method.
Please don't argue against straw men, it's annoying.
Scientists are normally obligated to release all data. It's the standard in most other fields (e.g., biology, econometrics).
Running the same data analysis on existing data doesn't count as reproducing the experiment. Running a different analysis on the same data is, however, doing new science.
In contrast, failing to release the data does count as "violating the scientific method".
Something completely unknown is making tree rings too small after 1960. And if you correlate your tree ring data to post-1960 temperatures it makes historical temperature look way too high.
Bizarro-Briffa sent a similar letter to Bizarro-Nature warning of an unknown divergence problem and saying not to correlate tree rings to pre-1960 temperatures because it makes historical temperature look way too low.
Since we do not know why the temperature has not correctly correlated over the last 50s years, we can't really know when else that has been the case, can we?
So why do we rely on tree rings for any period?
They are proven to be unreliable over a signifcant period, and we don't know if or how often that's also true of the past.
Thus, either (a) it has already been done, or (b) it is more difficult than you would expect for non-obvious reasons. So what is your point? Have you even bothered to find out if anyone has done it or not?
The standard of evidence to show fault is yours. Assuming that papers published in freaking NATURE have sound statistics is a reasonable thing for a layman to do.
I thought that's what you were asserting. My assumption is that scientists have correlated tree ring data with other sources and found it to be unreliable. But of course, neither my nor your assumptions about what's possible and impossible in this area are worth a damn, because we __don't know what we're talking about__. That is my point.
Also, all this talk about fallacies, ad homonim, etc. etc., is very "internet". Especially since you only seem to have a vague grasp of what these words actually mean. Do you think we could maybe leave all that stuff out and have a serious discussion?
That or you go look for the papers or the reason the papers have not been written.
This game is old. The standard of proof to demonstrate the method is flawed based on poor correlations is on your side. The method to do the correlation is public knowledge. The data to correlate is public knowledge. R is open source. Go forth and correlate.
You're confused about something: the post you're replying to clearly wasn't accusing an entire scientific field of engaging in "completely terrible" statistics. The poster was just questioning whether the comparison you were suggesting was feasible given our current (rather imperfect) knowledge of temperature changes over the last few hundred years. It may not be a matter of just getting two sets of figures and comparing them.
This is a clear claim that climate science researchers have not done basic statistics when presenting new models. This claim has been substantiated by nothing except you requesting more and more detailed analysis from non-experts, which you discount every time we provide it because you assert that you would only listen to an expert.
Ironically, you are actually failing to respect the authority of experts, because the discussion is fundamentally an accusation that experts are failing to do basic statistics. You are accusing us of not being experts so you can't listen to us in the defense of the experts you're attacking by proxy.
I seriously fucking doubt you understood any of this. But it doesn't matter, I won't return. I have no tolerance for people who argue by asking questions and petulantly demanding unreasonable proof while providing absolutely nothing of their own.
Thanks for ruining hacker news.
My problem here is that I'm not sure which metaphorical goalposts you are referring to. By the way, constantly talking in terms of "fallacies" (especially these informal ones, as opposed to actual fallacies of propositional or predicate logic) just makes you sound like a petulant 15 year old. (Ad hominem -- I'll do that one for you.) Don't refer me to a Wikipedia page on the informal fallacy of the month; address what I'm saying instead. Otherwise, you may be guilty of the fallacy of Missing the Goddamn Point Because You're Looking For Fallacies Instead of Reading What the Other Guy has Written.
Getting back to the issue at hand, I can only suggest that you (a) reread the thread and (b) calm down. I am not accusing any scientists of failing to do basic statistics, and neither was anyone else. I am merely expressing my awesome ignorance of all things related to climate science, and suggesting that perhaps some of the other posters here are about as ignorant of these issues as I am, but lacking a suitable degree of caution in their statements.
Good luck with leaving HN. Sadly, I very much doubt that you will find any discussion sites worthy of your superior intellect.
Climate research is research. It is not some know-it alls claiming one thing or another. It is making up theories and testing and aligning them with available data. It is trying to get new sources for data. And so on. It is an ongoing process.
Science never KNOWS anything. It only assigns weights and likelihoods.
We don't know if all history of earth is just an illusion, because we could just be a computer simulation that started 10 seconds ago.
So we will never KNOW if tree ring data at all times before 1960 correlated with temperature. All we can do is to collect evidence, make up theories and compare them with reality.
Disclaimer: I am not a climate researcher myself.
Trees are complex organisms whose growth relies on many factors in a complicated relationship. Just as a simple example, if you have a long, hot drought, do you expect tree growth to be more affected by the temperature or the lack of moisture? The scientists must control for these kinds of factors to isolate the temperature signal. But when you're dealing with the interactions of complex systems, new factors can emerge and the methods need to change to keep up.