Climate science has incentives that seem unusually pointed away from truth-discovery compared to other fields, and there seems to be evidence of climate scientists (rationally!) responding to those incentives.
This isn’t an issue I care much about, so I hope you won’t mind a quick treatment. The 5-second version is that more is on the line for the researchers personally with climate change research than other sorts of science, that scientific reports have consequences for politics which forces the scientists to be more involved in politics than other sorts of scientists, and that this corroding influence of politics adds friction to the processes that pump errors out of science. Here’s an incomplete list with more details:
The most visible one is media pressure, which rewards more shocking possibilities. (Did you ever hear about clathrate gun hypothesis? Did you hear that further research didn’t pan out?) In most other fields, the media rarely if ever comes calling, and so more of the incentives come from other researchers. (I remember a historian, who was called up because Sarah Palin had made a correct but non-obvious comment about Paul Revere, who pointed out that he had never been interviewed until Palin had mentioned Revere.)
Relatedly, the funding and visibility of climatology depends in large part on how much people care about climatology. This is true for many fields- NASA seems to find asteroids whose paths take them close to Earth whenever Congress debates NASA’s funding- but seems particularly true for climatology.
There is also a lot of money and influence going against climate change research. The decision to not fly DSCOVR, for example, was almost certainly political and did significant harm to our understanding of the climate and its dynamics. There are claims of data suppression during the Bush era which are harder to evaluate, but are suggestive of politicization of research.
The existence of climate deniers acting in bad faith poisons the well for normal scientific bug-checking. The Lenski affair happened in evolution, but the same sort of dynamic can show up between climatologists and skeptics. There is no social benefit in providing data to people acting in bad faith, and so it’s easier to just not give them your data, even though people acting in bad faith can identify scientific errors that you would need to correct. But this prevents normal criticism from taking place, and (as a general human tendency) scientists are more likely to see others as acting in bad faith if their work is being criticized. (This is pretty neatly captured by Climategate. Most critics stupidly focused on things that were benign- like the phrase “trick”- while the emails included evidence of data withholding and stonewalling, which are not benign.)
There is no social benefit in providing data to people acting in bad faith, and so it’s easier to just not give them your data
I am not sure I agree with that. I think I would phrase my position as “There is social benefit in providing data to absolutely everyone without any prequalifications”.
For example the Freedom of Information Act says government must provide certain kinds of information on request. It doesn’t get to peek into the requester’s mind and decide whether there was “bad faith” or “good faith”. I think it’s a very good thing that it doesn’t.
Similarly, I think science should provide supporting data to everyone who asks (subject to reasonable limits tied to the costs of providing data). In particular, scientists should stand ready to provide data to their worst critics regardless of what they think of their theories, competency, or fashion sense.
And yes, Climategate brought into the open a rather large fail in that respect on the part of pro-warming crowd (among other things).
Similarly, I think science should provide supporting data to everyone who asks (subject to reasonable limits tied to the costs of providing data). In particular, scientists should stand ready to provide data to their worst critics regardless of what they think of their theories, competency, or fashion sense.
Agreed that this is the most truth-seeking policy. I specified “social benefit” because I don’t think the two line up.
The “trick” has not benign, given it consisted of pretty blatant data cooking.
DISCLAIMER: literally the only thing I know about this is the code on the linked page.
I’m not sure thats true. The code you linked to seems to be implementing this correction pretty much directly from the literature:
Briffa, K.R., Schweingruber, F.H., Jones, P.D., Osborn, T.J., Shiyatov, S.G. and Vaganov, E.A., 1998
“Reduced sensitivity of recent tree growth to temperature at high northern latitudes.”
Nature 391, 678-682 (R)
(I found it by taking the name of the file briffa_sep98 and googling for a few minutes). I know nothing about tree-rings, but its a nature paper, so I assume the authors know about tree-rings. I can’t verify how correctly its implemented because I’m not going to trace the code leading into this routine to check the units for the tree ring thickness being adjusted..
The code in question seems to output a line on a graph labled “Northern Hemisphere MXD corrected for decline.”
So as far as I can tell there isn’t a lot of scientific misconduct going on- they took a data-correction from the literature (anyone know if its a standard correction?), coded it up, and labled the line on the graph as implementing the correction. The uncorrected line is also output by the code on to the same graph.
Or are we considering that the Briffa correction itself is cooking the data? If so shouldn’t the discussion center around the published paper and where it goes wrong?
The “trick” (“Mike’s Nature trick”) refers to a technique for plotting reconstructed temperatures (from tree rings) together with recent direct measurements, apparently simply by drawing them in different colours. The code you linked above is a separate technique (labelled “hiding the decline”), a correction for an anomalous decline in tree ring growth, which before 1960 had been closely correlated with temperature. It is speculated that the decline in tree ring growth is due to air pollution or other anthropogenic causes. See here for more detail.
Really? What kind of incentives?
This isn’t an issue I care much about, so I hope you won’t mind a quick treatment. The 5-second version is that more is on the line for the researchers personally with climate change research than other sorts of science, that scientific reports have consequences for politics which forces the scientists to be more involved in politics than other sorts of scientists, and that this corroding influence of politics adds friction to the processes that pump errors out of science. Here’s an incomplete list with more details:
The most visible one is media pressure, which rewards more shocking possibilities. (Did you ever hear about clathrate gun hypothesis? Did you hear that further research didn’t pan out?) In most other fields, the media rarely if ever comes calling, and so more of the incentives come from other researchers. (I remember a historian, who was called up because Sarah Palin had made a correct but non-obvious comment about Paul Revere, who pointed out that he had never been interviewed until Palin had mentioned Revere.)
Relatedly, the funding and visibility of climatology depends in large part on how much people care about climatology. This is true for many fields- NASA seems to find asteroids whose paths take them close to Earth whenever Congress debates NASA’s funding- but seems particularly true for climatology.
There is also a lot of money and influence going against climate change research. The decision to not fly DSCOVR, for example, was almost certainly political and did significant harm to our understanding of the climate and its dynamics. There are claims of data suppression during the Bush era which are harder to evaluate, but are suggestive of politicization of research.
The existence of climate deniers acting in bad faith poisons the well for normal scientific bug-checking. The Lenski affair happened in evolution, but the same sort of dynamic can show up between climatologists and skeptics. There is no social benefit in providing data to people acting in bad faith, and so it’s easier to just not give them your data, even though people acting in bad faith can identify scientific errors that you would need to correct. But this prevents normal criticism from taking place, and (as a general human tendency) scientists are more likely to see others as acting in bad faith if their work is being criticized. (This is pretty neatly captured by Climategate. Most critics stupidly focused on things that were benign- like the phrase “trick”- while the emails included evidence of data withholding and stonewalling, which are not benign.)
I am not sure I agree with that. I think I would phrase my position as “There is social benefit in providing data to absolutely everyone without any prequalifications”.
For example the Freedom of Information Act says government must provide certain kinds of information on request. It doesn’t get to peek into the requester’s mind and decide whether there was “bad faith” or “good faith”. I think it’s a very good thing that it doesn’t.
Similarly, I think science should provide supporting data to everyone who asks (subject to reasonable limits tied to the costs of providing data). In particular, scientists should stand ready to provide data to their worst critics regardless of what they think of their theories, competency, or fashion sense.
And yes, Climategate brought into the open a rather large fail in that respect on the part of pro-warming crowd (among other things).
Agreed that this is the most truth-seeking policy. I specified “social benefit” because I don’t think the two line up.
The “trick” has not benign, given it consisted of pretty blatant data cooking.
DISCLAIMER: literally the only thing I know about this is the code on the linked page.
I’m not sure thats true. The code you linked to seems to be implementing this correction pretty much directly from the literature:
Briffa, K.R., Schweingruber, F.H., Jones, P.D., Osborn, T.J., Shiyatov, S.G. and Vaganov, E.A., 1998 “Reduced sensitivity of recent tree growth to temperature at high northern latitudes.” Nature 391, 678-682 (R)
(I found it by taking the name of the file briffa_sep98 and googling for a few minutes). I know nothing about tree-rings, but its a nature paper, so I assume the authors know about tree-rings. I can’t verify how correctly its implemented because I’m not going to trace the code leading into this routine to check the units for the tree ring thickness being adjusted..
The code in question seems to output a line on a graph labled “Northern Hemisphere MXD corrected for decline.”
So as far as I can tell there isn’t a lot of scientific misconduct going on- they took a data-correction from the literature (anyone know if its a standard correction?), coded it up, and labled the line on the graph as implementing the correction. The uncorrected line is also output by the code on to the same graph.
Or are we considering that the Briffa correction itself is cooking the data? If so shouldn’t the discussion center around the published paper and where it goes wrong?
The “trick” (“Mike’s Nature trick”) refers to a technique for plotting reconstructed temperatures (from tree rings) together with recent direct measurements, apparently simply by drawing them in different colours. The code you linked above is a separate technique (labelled “hiding the decline”), a correction for an anomalous decline in tree ring growth, which before 1960 had been closely correlated with temperature. It is speculated that the decline in tree ring growth is due to air pollution or other anthropogenic causes. See here for more detail.
I hope nobody just starts listing incentives here, but also explains what’s unusual about them.
Yeah, that “compared to other fields” part confuses me too.