June 13, 2010
On Global Warming
I've been meaning to do a post like this for a while...
When people talk about global warming, the legitimacy of the science backing it, etc., they're talking about several different issues and questions all rolled into one. They are these:
1) There are difficulties with temperature measurements. We don't have the luxury of being able to directly measure temperatures from a hundred years ago, or fifty years ago. (Or even five minutes ago, though we do a pretty good job of direct observation these days, so our recent records can be taken as essentially accurate.) Early temperature records were often taken inconsistently, with poor-quality instruments, over only a fraction of the area we might be interested in. Beyond that, we're forced to rely on proxies, such as ice cores and tree-ring data, that don't correlate all that well with actual temperatures.
2) On top of that, the data sets that we've been using for global warming analysis have been normalized in processes that are, well, shall we say not entirely transparent and well-documented? Now, this isn't necessarily instant evidence that they were falsified by crazed global warming activists who want to seize control of all industrial activity (they won't show up for a few sections!) But it does limit the usefulness of those data sets when it comes to analyzing the climate. We're studying a very complex system with relatively poor tools. We know that there are cycles that we only poorly understand, and we suspect that there are additional cycles that we don't understand at all (and we are absolutely sure that there's one big ice age-temperate cycle that we can't even describe properly). Data that's been normalized, especially data that's been normalized by hand, reduces our ability to accommodate questions regarding these cycles. Oh, we need to take into account El Nino-based warming? But do we need to apply that correction across the board, or have some of our observations already been Nino-corrected? Double-correcting is as bad as ignoring it; worse, because then we may think we have reached a solid conclusion when we have not.
3) There are limits to the resolution and utility of computer modeling of a complex system. I'm regularly astounded by encounters with individuals who believe that the computer models are modeling actual, physical processes that are understood in their entirety by the scientists creating the model; in short, they don't understand the difference between the sort of observation a physicist might make in a lab, and the output of a statistical model that takes topographical and temperature data as inputs to approximate weather processes that we can't model in the micro sense, much less the macro sense. In this sense, the models' failure to correctly model the relative temperature stability of the last decade should at least be troubling. Yes, okay, I'm willing to grant that the last decade might be an anomaly in the context of a gradual increase of temperature, but it's not like the models were predicting the anomaly; even now they can't really account for why the expected warming didn't happen, without applying a pretty big fudge factor to account for it without actually explaining it. To the extent that reality doesn't conform to your model, that suggests your model is faulty!
4) On top of that, the models that predict catastrophic warming in the next century do so mostly through the assumption of certain positive feedback loops (event A promotes B, B promotes A, A promotes more of B, etc.) The problem that you run into here is that positive feedback loops don't tend to result in broadly stable systems; if warming tended to suddenly result in a lot more warming, we have to explain why things haven't spiraled into disaster areas when those events have occurred in the past. (Yes, we've been warmer in the past. One can argue that we haven't had as much atmospheric CO2 in the past to go with that warmth, but the CO2 isn't generally involved in the feedback loop. Rate of change arguments don't really have a lot to do with this either - those have more to deal with habitat than climate.)
5) The computer models use pretty loose statistical standards of evidence (mostly 90% confidence levels). Granted that it's not like they can fire up their office mini-Gaea to run experiments. But if you move the error bars up to 95% or so, your error bars get big enough to include the null hypothesis (i.e. they suggest but don't prove warming). Keep in mind that those error bars have nothing to do with the previous sections - they reflect the chance that, if the underlying models are sound, that the statistical analysis is just flat wrong.
6) CO2 emissions can't be limited by country; they call it "global" warming for a reason. Emissions which occur in China or India have the same effect as emissions from your backyard. Thus, taking action country by country to limit emissions is worthless - to the extent that industry can relocate to regions without a CO2-emission legal limit, it does. This has certainly been the experience of Europe under the Kyoto treaty. In fact, arguably it could even make the problem worse, if the industry that relocates is then powered by less-efficient hydrocarbon-burning plants, characteristic of developing nations. (Does nobody remember the fuss with Beijing, smog, and the '08 Olympics?)
7) The chances of getting India or China to agree to an international system of emissions control is significantly less than zero. Well, it's no surprise. India and China have tremendous populations still mired in subsistence farming; these populations need access to energy and industrial goods in order to modernize. For them, the worst-case scenarios presented by global warming activists aren't nearly bad as "just do what you're doing now for the next 50 years". (Subsistence farming -sucks-!) Should we come up with a magic-bullet solution, that allows us to generate the power we need without setting fire to a hydrocarbon, doubtless they will agree to go along (probably asking for significant subsidies to help them make the conversion, but meh.)
8) It's not a magic bullet, but we do have a method of producing significant electric power without burning hydrocarbons; it's called nuclear fission. It works with current technology, it has a good safety record (especially if you discount Russian engineering design, and if you compare it to the radioactive emissions caused by, say, coal plants). France has embraced nuclear fission as a source of the majority of its domestic electricity and had no problems for decades. The failure of global warming activists to embrace nuclear energy is, well... illustrative, and has a lot to say about the objectives of a lot of those who fall under that umbrella.
ALL of these segments are independent of each other, more or less. We could perfectly resolve the political situations, but if the original data is sufficiently crappy, so too the science based on it. The data could be perfect, but fed into statistical models that just aren't up to the modeling job. The models could be predicting today perfectly, but the positive feedback mechanisms that they've posited for the future might simply not work that way. The science could be 100% flawless, but good luck convincing a Chinese peasant that because of that, his family should starve in the next famine.
Sorry that I went on a bit of a political/scientific jag there. Holiday from Haruhi DVD subtitle formatting (damn you, Minoru Shiraishi!)
Comments are disabled.
Post is locked.
When people talk about global warming, the legitimacy of the science backing it, etc., they're talking about several different issues and questions all rolled into one. They are these:
1) There are difficulties with temperature measurements. We don't have the luxury of being able to directly measure temperatures from a hundred years ago, or fifty years ago. (Or even five minutes ago, though we do a pretty good job of direct observation these days, so our recent records can be taken as essentially accurate.) Early temperature records were often taken inconsistently, with poor-quality instruments, over only a fraction of the area we might be interested in. Beyond that, we're forced to rely on proxies, such as ice cores and tree-ring data, that don't correlate all that well with actual temperatures.
2) On top of that, the data sets that we've been using for global warming analysis have been normalized in processes that are, well, shall we say not entirely transparent and well-documented? Now, this isn't necessarily instant evidence that they were falsified by crazed global warming activists who want to seize control of all industrial activity (they won't show up for a few sections!) But it does limit the usefulness of those data sets when it comes to analyzing the climate. We're studying a very complex system with relatively poor tools. We know that there are cycles that we only poorly understand, and we suspect that there are additional cycles that we don't understand at all (and we are absolutely sure that there's one big ice age-temperate cycle that we can't even describe properly). Data that's been normalized, especially data that's been normalized by hand, reduces our ability to accommodate questions regarding these cycles. Oh, we need to take into account El Nino-based warming? But do we need to apply that correction across the board, or have some of our observations already been Nino-corrected? Double-correcting is as bad as ignoring it; worse, because then we may think we have reached a solid conclusion when we have not.
3) There are limits to the resolution and utility of computer modeling of a complex system. I'm regularly astounded by encounters with individuals who believe that the computer models are modeling actual, physical processes that are understood in their entirety by the scientists creating the model; in short, they don't understand the difference between the sort of observation a physicist might make in a lab, and the output of a statistical model that takes topographical and temperature data as inputs to approximate weather processes that we can't model in the micro sense, much less the macro sense. In this sense, the models' failure to correctly model the relative temperature stability of the last decade should at least be troubling. Yes, okay, I'm willing to grant that the last decade might be an anomaly in the context of a gradual increase of temperature, but it's not like the models were predicting the anomaly; even now they can't really account for why the expected warming didn't happen, without applying a pretty big fudge factor to account for it without actually explaining it. To the extent that reality doesn't conform to your model, that suggests your model is faulty!
4) On top of that, the models that predict catastrophic warming in the next century do so mostly through the assumption of certain positive feedback loops (event A promotes B, B promotes A, A promotes more of B, etc.) The problem that you run into here is that positive feedback loops don't tend to result in broadly stable systems; if warming tended to suddenly result in a lot more warming, we have to explain why things haven't spiraled into disaster areas when those events have occurred in the past. (Yes, we've been warmer in the past. One can argue that we haven't had as much atmospheric CO2 in the past to go with that warmth, but the CO2 isn't generally involved in the feedback loop. Rate of change arguments don't really have a lot to do with this either - those have more to deal with habitat than climate.)
5) The computer models use pretty loose statistical standards of evidence (mostly 90% confidence levels). Granted that it's not like they can fire up their office mini-Gaea to run experiments. But if you move the error bars up to 95% or so, your error bars get big enough to include the null hypothesis (i.e. they suggest but don't prove warming). Keep in mind that those error bars have nothing to do with the previous sections - they reflect the chance that, if the underlying models are sound, that the statistical analysis is just flat wrong.
6) CO2 emissions can't be limited by country; they call it "global" warming for a reason. Emissions which occur in China or India have the same effect as emissions from your backyard. Thus, taking action country by country to limit emissions is worthless - to the extent that industry can relocate to regions without a CO2-emission legal limit, it does. This has certainly been the experience of Europe under the Kyoto treaty. In fact, arguably it could even make the problem worse, if the industry that relocates is then powered by less-efficient hydrocarbon-burning plants, characteristic of developing nations. (Does nobody remember the fuss with Beijing, smog, and the '08 Olympics?)
7) The chances of getting India or China to agree to an international system of emissions control is significantly less than zero. Well, it's no surprise. India and China have tremendous populations still mired in subsistence farming; these populations need access to energy and industrial goods in order to modernize. For them, the worst-case scenarios presented by global warming activists aren't nearly bad as "just do what you're doing now for the next 50 years". (Subsistence farming -sucks-!) Should we come up with a magic-bullet solution, that allows us to generate the power we need without setting fire to a hydrocarbon, doubtless they will agree to go along (probably asking for significant subsidies to help them make the conversion, but meh.)
8) It's not a magic bullet, but we do have a method of producing significant electric power without burning hydrocarbons; it's called nuclear fission. It works with current technology, it has a good safety record (especially if you discount Russian engineering design, and if you compare it to the radioactive emissions caused by, say, coal plants). France has embraced nuclear fission as a source of the majority of its domestic electricity and had no problems for decades. The failure of global warming activists to embrace nuclear energy is, well... illustrative, and has a lot to say about the objectives of a lot of those who fall under that umbrella.
ALL of these segments are independent of each other, more or less. We could perfectly resolve the political situations, but if the original data is sufficiently crappy, so too the science based on it. The data could be perfect, but fed into statistical models that just aren't up to the modeling job. The models could be predicting today perfectly, but the positive feedback mechanisms that they've posited for the future might simply not work that way. The science could be 100% flawless, but good luck convincing a Chinese peasant that because of that, his family should starve in the next famine.
Sorry that I went on a bit of a political/scientific jag there. Holiday from Haruhi DVD subtitle formatting (damn you, Minoru Shiraishi!)
Posted by: Avatar_exADV at
01:29 AM
| Comments (7)
| Add Comment
Post contains 1255 words, total size 8 kb.
1
Great post! A couple questions/comments...
To start, a general comment on the first five points collectively- I may be mistaken, but your comments give me the impression that you believe that anomalous temperatures in the last century are the primary, or even sole, piece of evidence for climate change. But even if we couldn't measure modern temperatures as all, we would have cause to be concerned-
Firstly, Human activity has increased the concentration of greenhouses gases in the atmosphere. That's pretty straightforward to measure and I don't think anyone disputes that.
Secondly, Greenhouse gas concentrations change the radiation balance of the Earth, which in turn changes the surface temperature of the Earth. This is pretty basic physics which has been understood for over a century- you can look at the absorption spectrum of these gases and see that they transmit visible light but absorb the kind of infra-red radiation given off by the surface of the earth.
So it's pretty well understood that some level of global warming should occur, and in fact was predicted as early 1896. It's only in the past couple of decades that the effect has been strong enough for us to directly measure- but those measurements should be taken as confirmation of the predictions of physics and not as our first indicator of a problem.
That's not to say that the measurements aren't importantly- firstly, it's good to have further evidence in support of that theory, if the temperature record were in stark contrast to what we'd predict we'd want to reevaluate the above argument to see what we might be missing. Also, it's important to understand feedback mechanisms- if the direct effect described above were all that was happening, we'd expect the effect to be much smaller than we are seeing, and measuring temperature changes helps us understand these mechanisms.
On to some of the individual points-
1,2- You bring up some excellent points about the difficulties involved in taking accurate temperature readings, not much to add here.
3- Again, good points about difficulties in computer simulations. But you say that you are concerned that we haven't been able to predict future global temperatures accurately, while I would say there are some very good reasons to think that it would be impossible to do so. There are completely unpredictable events which have major forcing effects on the global temperature. For a dramatic example, see if you can find a plot of the temperature record around the end of the 19th century. There is a pronounced dip around 1883 when Krakatoa exploded- the temperature was depressed for a good four years following this event. Without knowledge of these and other forcing effects I don't think it's possible to predict global temperatures, at least not on a year by year basis- the best one can do is check after the fact whether the temperature record is consistent with one's model. That's not nearly as strong evidence in favor of a model as prediction would be, but I think it's the best we can do.
4- You argue here that systems with positive feedback loops are generally unstable, but that is not, in fact, correct. Certainly systems with very strong feedback loops can be unstable, but that's not a general quality of feedback. Consider a simple system where a temperature of x results in a forcing of f*x. The feedback effect of this term will be of order f^2*x, and higher order terms will give the series x*(1+f+f^2+f^3...). Now for f greater than or equal to 1 this gives the kind of runaway effect you describe, but in general the total effect can be any arbitrary multiplier- for example if every degree change resulted in another half degree, the total effect would be to double the size of any change. Certainly there is no reason to conclude from positive feedback effects that things will spiral out of control. I think the concern is more that we would have difficulty living outside of a very narrow temperature band- if feedback effects result in the temperature changing a couple of degrees, we'd have problems coping, not because it would continue to compound but just because that would be enough.
5- I don't think any computer simulation can prove global warming with 100% confidence. But don't you think 90% confidence would be plenty to worry over? If the Earth had a 90% chance of being hit by a meteorite that would change the climate several degrees I think it would be well past the time to be concerned.
6,7- Largely in agreement with you about the political challenges ahead.
8- I'm very much pro nuclear power. Even if we can develop cleaner energy sources down the road, I can't imagine that happening quickly and we need something during the transition period. I'm as upset as you are that it hasn't been embraced.
To start, a general comment on the first five points collectively- I may be mistaken, but your comments give me the impression that you believe that anomalous temperatures in the last century are the primary, or even sole, piece of evidence for climate change. But even if we couldn't measure modern temperatures as all, we would have cause to be concerned-
Firstly, Human activity has increased the concentration of greenhouses gases in the atmosphere. That's pretty straightforward to measure and I don't think anyone disputes that.
Secondly, Greenhouse gas concentrations change the radiation balance of the Earth, which in turn changes the surface temperature of the Earth. This is pretty basic physics which has been understood for over a century- you can look at the absorption spectrum of these gases and see that they transmit visible light but absorb the kind of infra-red radiation given off by the surface of the earth.
So it's pretty well understood that some level of global warming should occur, and in fact was predicted as early 1896. It's only in the past couple of decades that the effect has been strong enough for us to directly measure- but those measurements should be taken as confirmation of the predictions of physics and not as our first indicator of a problem.
That's not to say that the measurements aren't importantly- firstly, it's good to have further evidence in support of that theory, if the temperature record were in stark contrast to what we'd predict we'd want to reevaluate the above argument to see what we might be missing. Also, it's important to understand feedback mechanisms- if the direct effect described above were all that was happening, we'd expect the effect to be much smaller than we are seeing, and measuring temperature changes helps us understand these mechanisms.
On to some of the individual points-
1,2- You bring up some excellent points about the difficulties involved in taking accurate temperature readings, not much to add here.
3- Again, good points about difficulties in computer simulations. But you say that you are concerned that we haven't been able to predict future global temperatures accurately, while I would say there are some very good reasons to think that it would be impossible to do so. There are completely unpredictable events which have major forcing effects on the global temperature. For a dramatic example, see if you can find a plot of the temperature record around the end of the 19th century. There is a pronounced dip around 1883 when Krakatoa exploded- the temperature was depressed for a good four years following this event. Without knowledge of these and other forcing effects I don't think it's possible to predict global temperatures, at least not on a year by year basis- the best one can do is check after the fact whether the temperature record is consistent with one's model. That's not nearly as strong evidence in favor of a model as prediction would be, but I think it's the best we can do.
4- You argue here that systems with positive feedback loops are generally unstable, but that is not, in fact, correct. Certainly systems with very strong feedback loops can be unstable, but that's not a general quality of feedback. Consider a simple system where a temperature of x results in a forcing of f*x. The feedback effect of this term will be of order f^2*x, and higher order terms will give the series x*(1+f+f^2+f^3...). Now for f greater than or equal to 1 this gives the kind of runaway effect you describe, but in general the total effect can be any arbitrary multiplier- for example if every degree change resulted in another half degree, the total effect would be to double the size of any change. Certainly there is no reason to conclude from positive feedback effects that things will spiral out of control. I think the concern is more that we would have difficulty living outside of a very narrow temperature band- if feedback effects result in the temperature changing a couple of degrees, we'd have problems coping, not because it would continue to compound but just because that would be enough.
5- I don't think any computer simulation can prove global warming with 100% confidence. But don't you think 90% confidence would be plenty to worry over? If the Earth had a 90% chance of being hit by a meteorite that would change the climate several degrees I think it would be well past the time to be concerned.
6,7- Largely in agreement with you about the political challenges ahead.
8- I'm very much pro nuclear power. Even if we can develop cleaner energy sources down the road, I can't imagine that happening quickly and we need something during the transition period. I'm as upset as you are that it hasn't been embraced.
Posted by: Dudest at June 16, 2010 08:25 PM (/OnPc)
2
Well, keep in mind that "greenhouse gases" and "CO2" are not equivalent. CO2 is a greenhouse gas, but there are others, and in fact other gases have a much stronger greenhouse effect (relative to their concentration in the atmosphere) than CO2 does. Methane, for example, has a very strong greenhouse effect associated with it; it doesn't get a lot of play because it has a very low concentration to start with, and most of that comes from the intestinal tracts of agricultural livestock. (Granted that people talk about released methane deposits from the sea floor as a possible global-warming related problem...)
The real kicker, though, is that numero uno when it comes to greenhouse effect is water vapor. But water vapor's inherently tied up with weather; the atmospheric content of water vapor is not a phenomenon that can be measured or understood properly except in the context of weather, and not climate-average weather but is-it-raining-tomorrow weather at that. Which means that it's not modeled well, because the models can only approximate the effects of weather, not actually model it.
But that's potentially huge, and yeah, it could go either way - if climate warming resulted in an increased amount of water vapor in the air, that would suggest a positive feedback event. But if the increased water vapor caused increased cloud formation, that would tend to increase atmospheric reflection, affect the albedo, and become a negative feedback event. And... you see how it rapidly becomes complicated.
Honestly, the idea that a couple of degrees of warming would inherently cause problems is bunk. Too much heat is bad in an "oh god it's hot, I want shade and a cool drink" way, but it's not the civilization-killer the way another ice age would be. It's something we could cope with until such a point as better options than setting fire to a petrochemical become available...
Your comment regarding statistical probability is a little odd. No computer simulation could prove global warming to 100%, simply because of the uncertainties present elsewhere in the data. The 90% means that, if everything else is accurate (i.e. if the temperature measurements are accurate, nothing's being double-accounted for, and the models are 100% accurate representations of reality), there's still a 10% chance that the model's results have nothing to do with reality. For that sort of thing, you generally DO want a 95% statistical chance or better; certainly you'd demand it from the FDA or from an engineer building a bridge.
The real kicker, though, is that numero uno when it comes to greenhouse effect is water vapor. But water vapor's inherently tied up with weather; the atmospheric content of water vapor is not a phenomenon that can be measured or understood properly except in the context of weather, and not climate-average weather but is-it-raining-tomorrow weather at that. Which means that it's not modeled well, because the models can only approximate the effects of weather, not actually model it.
But that's potentially huge, and yeah, it could go either way - if climate warming resulted in an increased amount of water vapor in the air, that would suggest a positive feedback event. But if the increased water vapor caused increased cloud formation, that would tend to increase atmospheric reflection, affect the albedo, and become a negative feedback event. And... you see how it rapidly becomes complicated.
Honestly, the idea that a couple of degrees of warming would inherently cause problems is bunk. Too much heat is bad in an "oh god it's hot, I want shade and a cool drink" way, but it's not the civilization-killer the way another ice age would be. It's something we could cope with until such a point as better options than setting fire to a petrochemical become available...
Your comment regarding statistical probability is a little odd. No computer simulation could prove global warming to 100%, simply because of the uncertainties present elsewhere in the data. The 90% means that, if everything else is accurate (i.e. if the temperature measurements are accurate, nothing's being double-accounted for, and the models are 100% accurate representations of reality), there's still a 10% chance that the model's results have nothing to do with reality. For that sort of thing, you generally DO want a 95% statistical chance or better; certainly you'd demand it from the FDA or from an engineer building a bridge.
Posted by: Avatar_exADV at June 16, 2010 09:04 PM (pWQz4)
3
I certainly hadn't meant to imply that CO2 was the only greenhouse gas, and I'd agree that the effects of water vapor are incredibly important. Another good example of the kinds of input that are impossible to predict ahead of time, making predictions on a year by year basis impossible. As you say, the final effect of increased greenhouse emissions depends in a detailed way upon the feedback mechanisms.
I think you are dramatically underestimating how bad more than a couple degrees change would be for much of the world. My understanding is that a four degree change would result in sea levels rising by about half a meter, which is enough to displace about 1.5 million people. I'd also be concerned about the effects on our food supply.
We're agreed that computer simulations estimates can only be accurate up to the degree that the assumptions built into them can be trusted. But why assume that any flaws in the assumptions mean it overestimate the risk of climate change, when it is equally likely that any existing flaws cause us to underestimate risks? Uncertainty just makes things more variable- it certainly doesn't make them any safer.
As for how much risk we should tolerate, I have to say I find your reasoning odd as well. The accuracy we expect from the FDA and bridge builders stems from how much risk we are willing to tolerate- we will not accept a 5% risk that a drug is unsafe, or that a bridge will fall down during use. We would never say that a drug must be shown to be more than 95% fatal before we would ban it, or that a building must be shown to be more than 95% likely to collapse before we would condemn it. So why would we say that not only is 5% risk of global warming not enough to take action, as we would with an unsafe drug or building, but that even a 90% risk is not enough, and that we must wait until we have shown with over 95% certainty that we are taking a risky action before we become concerned?
I think you are dramatically underestimating how bad more than a couple degrees change would be for much of the world. My understanding is that a four degree change would result in sea levels rising by about half a meter, which is enough to displace about 1.5 million people. I'd also be concerned about the effects on our food supply.
We're agreed that computer simulations estimates can only be accurate up to the degree that the assumptions built into them can be trusted. But why assume that any flaws in the assumptions mean it overestimate the risk of climate change, when it is equally likely that any existing flaws cause us to underestimate risks? Uncertainty just makes things more variable- it certainly doesn't make them any safer.
As for how much risk we should tolerate, I have to say I find your reasoning odd as well. The accuracy we expect from the FDA and bridge builders stems from how much risk we are willing to tolerate- we will not accept a 5% risk that a drug is unsafe, or that a bridge will fall down during use. We would never say that a drug must be shown to be more than 95% fatal before we would ban it, or that a building must be shown to be more than 95% likely to collapse before we would condemn it. So why would we say that not only is 5% risk of global warming not enough to take action, as we would with an unsafe drug or building, but that even a 90% risk is not enough, and that we must wait until we have shown with over 95% certainty that we are taking a risky action before we become concerned?
Posted by: Dudest at June 16, 2010 09:43 PM (/OnPc)
4
You're still conflating statistical certainty with probability. The 90% isn't "this has a 90% chance of happening". It's "there's a 10% that even if we're totally correct in our methodology, understanding of climate, data handling techniques, etc., that we're still completely wrong - that the data set that we have is insufficient to the point where it doesn't say anything about climate change at all."
It's not done at 90% because we're incapable of running the calculations at 95% statistical probability. It's because the error bars at 95% are so large that they include the null hypothesis. (Remember, when you're dealing with a statistical model of an event of unknown probability, you're just saying "the mean is somewhere within the error bars"; you're NOT saying "it's smack dab where I think it is with a normal distribution of chance either way", and it's a common misconception to assume that.)
Over 100 years, possibly displacing a million people isn't of a whole lot of consequence. That's maybe a quarter Afghanistan or half a Sudan. I mean, not great, to be sure, but we regularly tolerate refugee-producing events of a similar scale, but which would be vastly easier to solve. If you're worried about a refugee problem, spending money on carbon is the least efficient way to go about it. (The same goes for third-world health outcomes.)
Food supply? We've got large amounts of land which are currently of only marginal value as agricultural land, because it's too cold; a few degrees of warming turns big swathes of the north US, south Canada, and endless miles of Siberia into arable land. For there to be a serious issue with food supply, there would have to be widespread dessication events of the type which alarmists are fond of alluding to, but which the scientists are not predicting (their climate models just don't include those details, after all). Sure, there's some costs involved in repurposing various types of agricultural land, but in a world of agri-subsidy, the overall risk is pretty minimal.
A lot of the above arguments aren't indictments of the motives of the people involved in advocating for carbon emission controls. The models aren't deficient because of a conspiracy to lie about the results of the models; they're deficient because what they're trying to do is really hard. But it's still possible that they can be, simultaneously, the result of the best science available to us, and not sufficient to prove the case. I'd go further than that, actually. They're definitely not sufficient to prove the case.
That said, the argument about nuclear is pretty much a direct indictment of the motives of those advocates. You're correct that we know enough that there should be some concern. We have a potential solution, or at least something that we can do that would both help ameliorate the problem, and can be justified on grounds having nothing to do with climate change. But as we agree on this point, I won't belabor it.
It's not done at 90% because we're incapable of running the calculations at 95% statistical probability. It's because the error bars at 95% are so large that they include the null hypothesis. (Remember, when you're dealing with a statistical model of an event of unknown probability, you're just saying "the mean is somewhere within the error bars"; you're NOT saying "it's smack dab where I think it is with a normal distribution of chance either way", and it's a common misconception to assume that.)
Over 100 years, possibly displacing a million people isn't of a whole lot of consequence. That's maybe a quarter Afghanistan or half a Sudan. I mean, not great, to be sure, but we regularly tolerate refugee-producing events of a similar scale, but which would be vastly easier to solve. If you're worried about a refugee problem, spending money on carbon is the least efficient way to go about it. (The same goes for third-world health outcomes.)
Food supply? We've got large amounts of land which are currently of only marginal value as agricultural land, because it's too cold; a few degrees of warming turns big swathes of the north US, south Canada, and endless miles of Siberia into arable land. For there to be a serious issue with food supply, there would have to be widespread dessication events of the type which alarmists are fond of alluding to, but which the scientists are not predicting (their climate models just don't include those details, after all). Sure, there's some costs involved in repurposing various types of agricultural land, but in a world of agri-subsidy, the overall risk is pretty minimal.
A lot of the above arguments aren't indictments of the motives of the people involved in advocating for carbon emission controls. The models aren't deficient because of a conspiracy to lie about the results of the models; they're deficient because what they're trying to do is really hard. But it's still possible that they can be, simultaneously, the result of the best science available to us, and not sufficient to prove the case. I'd go further than that, actually. They're definitely not sufficient to prove the case.
That said, the argument about nuclear is pretty much a direct indictment of the motives of those advocates. You're correct that we know enough that there should be some concern. We have a potential solution, or at least something that we can do that would both help ameliorate the problem, and can be justified on grounds having nothing to do with climate change. But as we agree on this point, I won't belabor it.
Posted by: Avatar_exADV at June 17, 2010 03:46 AM (mRjOr)
5
"conflating statistical certainty with probability"- I don't think I am, but perhaps I'm drawing an inference from what you stated that you didn't intend. You've stated several times that if the error bars were extended to allow a 95% confidence level that they would include the null hypothesis that climate change was not occurring- and I inferred from this that at a 90% confidence level the studies you are considering do not have this possibility within their confidence range. On the other hand if those models predict climate change with less than 100% probability we have to further discount that rate by 10% to account for the chance that we have a false positive, the issue of statistical certainty you brought up. Could you clarify whether you in fact meant for the possibility of no climate change to fall within the 90% confidence interval, because I'm getting a different impression from your later comments than I did from your initial comments.
I still think you are being incredibly blase about the risks involved in climate change. We're basically engaging in reckless geoengineering by changing the chemical concentrations in our atmosphere, and our best guess as to the consequences is that they will be dangerous to us. If we are highly uncertain about the consequences, that should make us more cautious, not less. You say that our climate models don't go into the details of the effects on our food supply- you shouldn't take that to mean that we will definitely be able to handle any problems, it means that we are going in blind and thus that this is extremely risky behavior. Also, why do you assume we have 100 years to deal with any population displacement? And could you provide some evidence that tackling global warming, for example by switching to nuclear power, would be the least efficient way to handle a refugee problem?
I'd agree that this is an incredibly hard problem, but I'd also say that the scientists working on it are aware of the issues we're discussing here and have been careful to avoid making any claims stronger than the evidence supports. I'd also agree that we don't have absolute proof of climate change, and I don't see how we could ever be completely certain without the sort of "office mini-Gaea" you jokingly referred to earlier, but I think the preponderance of evidence is in support of this theory, and that even if we think there is only a small chance of it occurring, a cost-benefit analysis would support taking steps to prevent it. I think you are rightfully skeptical of our ability to predict future temperatures accurately with current models, but I think this is not our sole or even primary piece of evidence for climate change.
(Lastly, this is kind of an aside, but- why the concern with the possible motives of people in favor of taking steps to curb climate change but against nuclear power? I mean, yes, there are some very silly people in the world, and some of them will take a knee-jerk stance of whatever they perceive to be the green solution, and there are also some people who will take a knee-jerk stance against doing anything about climate change for equally silly reasons. But why should we worry about either camp for a second, when we can just focus on the actual evidence? Supposing some people come to the right answer for the wrong reason, should that make us hesitate at all if we independently came to the same answer?)
I still think you are being incredibly blase about the risks involved in climate change. We're basically engaging in reckless geoengineering by changing the chemical concentrations in our atmosphere, and our best guess as to the consequences is that they will be dangerous to us. If we are highly uncertain about the consequences, that should make us more cautious, not less. You say that our climate models don't go into the details of the effects on our food supply- you shouldn't take that to mean that we will definitely be able to handle any problems, it means that we are going in blind and thus that this is extremely risky behavior. Also, why do you assume we have 100 years to deal with any population displacement? And could you provide some evidence that tackling global warming, for example by switching to nuclear power, would be the least efficient way to handle a refugee problem?
I'd agree that this is an incredibly hard problem, but I'd also say that the scientists working on it are aware of the issues we're discussing here and have been careful to avoid making any claims stronger than the evidence supports. I'd also agree that we don't have absolute proof of climate change, and I don't see how we could ever be completely certain without the sort of "office mini-Gaea" you jokingly referred to earlier, but I think the preponderance of evidence is in support of this theory, and that even if we think there is only a small chance of it occurring, a cost-benefit analysis would support taking steps to prevent it. I think you are rightfully skeptical of our ability to predict future temperatures accurately with current models, but I think this is not our sole or even primary piece of evidence for climate change.
(Lastly, this is kind of an aside, but- why the concern with the possible motives of people in favor of taking steps to curb climate change but against nuclear power? I mean, yes, there are some very silly people in the world, and some of them will take a knee-jerk stance of whatever they perceive to be the green solution, and there are also some people who will take a knee-jerk stance against doing anything about climate change for equally silly reasons. But why should we worry about either camp for a second, when we can just focus on the actual evidence? Supposing some people come to the right answer for the wrong reason, should that make us hesitate at all if we independently came to the same answer?)
Posted by: Dudest at June 17, 2010 05:07 AM (/OnPc)
6
A quick explanation of how the confidence intervals work - the size of the error bars and the level of confidence are inversely related. At a greater level of confidence (that is to say, the percentage chance that the statistical data underlying the model is sufficient to indicate that the results of the model are significant), we have greater error bars - we're more sure that we're barking up the right tree, as it were, so long as we're willing to expand the definition of "right tree" to a bigger clump of trees, any one of which could be the one we're supposed to be barking up. We can narrow down that clump of trees (our error bars), but at the same time, we have relatively less chance of being barking anywhere like in the right area at all.
It's not so much that I'm blase about the risks, as that I recognize that there's a lot of terrible things going on in the world. Amazingly, a lot of those things don't have anything to do with global warming. If the suffering of refugees is what moves you, there are far more refugees being created by other causes than global warming (and causes that might, just might, be amenable to change with less than a mythical concerted effort world-wide). If poverty is what moves you, there's a great amount of poverty in the world as it is, and to put it bluntly, almost all efforts to prevent global warming will make it harder for impoverished nations and peoples to industrialize and enjoy the fruits of modern society.
You mention a cost-benefit analysis; the UN did one as part of a development conference a few years ago, to see what sort of expenditures in poor African nations would promote the most overall well-being. If I recall correctly, water sanitization and mosquito netting came in at 1 and 2, or possibly 2 and 1, with a whole raft of development projects coming in after those; global warming amelioration was dead last by over three orders of magnitude (which is to say, each dollar you spent on mosquito netting was worth more than $1000 you spent on CO2 reduction).
The reason I mention motive is that there are several potential solution spaces, if you'll allow me to use that term; a lot of people seem to have already come to the conclusion that the correct solution space is to limit industrial activity to generate less CO2 (as opposed to expanding nuclear power, or direct amelioration through geo-engineering, or "just deal with the effects"). Given the limited utility of the reduce-industry approach, given the geopolitics and economics involved, you'd expect it to be the least popular based on merits alone. Certainly, some of the anti-nuclear bias is just plain anti-nuclear bias, but to a degree, you have to ask, is it just that some people aren't interested in solutions that don't result in turning off the lights and shutting down plants? Certainly some people have been advocating for just that, long before global warming was an issue...
It's not so much that I'm blase about the risks, as that I recognize that there's a lot of terrible things going on in the world. Amazingly, a lot of those things don't have anything to do with global warming. If the suffering of refugees is what moves you, there are far more refugees being created by other causes than global warming (and causes that might, just might, be amenable to change with less than a mythical concerted effort world-wide). If poverty is what moves you, there's a great amount of poverty in the world as it is, and to put it bluntly, almost all efforts to prevent global warming will make it harder for impoverished nations and peoples to industrialize and enjoy the fruits of modern society.
You mention a cost-benefit analysis; the UN did one as part of a development conference a few years ago, to see what sort of expenditures in poor African nations would promote the most overall well-being. If I recall correctly, water sanitization and mosquito netting came in at 1 and 2, or possibly 2 and 1, with a whole raft of development projects coming in after those; global warming amelioration was dead last by over three orders of magnitude (which is to say, each dollar you spent on mosquito netting was worth more than $1000 you spent on CO2 reduction).
The reason I mention motive is that there are several potential solution spaces, if you'll allow me to use that term; a lot of people seem to have already come to the conclusion that the correct solution space is to limit industrial activity to generate less CO2 (as opposed to expanding nuclear power, or direct amelioration through geo-engineering, or "just deal with the effects"). Given the limited utility of the reduce-industry approach, given the geopolitics and economics involved, you'd expect it to be the least popular based on merits alone. Certainly, some of the anti-nuclear bias is just plain anti-nuclear bias, but to a degree, you have to ask, is it just that some people aren't interested in solutions that don't result in turning off the lights and shutting down plants? Certainly some people have been advocating for just that, long before global warming was an issue...
Posted by: Avatar_exADV at June 17, 2010 03:33 PM (pWQz4)
7
Your description of a confidence interval is exactly how I thought they worked, so I'm not sure where the confusion is coming from. Perhaps I can clarify my earlier comment- I wasn't saying that the model estimated a 90% chance that global warming was real and a 10% chance that it wasn't. I was saying that if it says global warming is real with a 90% confidence interval, then with a 90% chance we are "barking up the right tree" and global warming is definitely real, and with a 10% chance we are "barking up the wrong tree". If that is the case we have no idea what the chance of global warming being real is, but it has to be between 0% and 100%. Using conditional probabilities we can bound the chance of global warming being real between 90% and 100% chance in that scenario. Now, that's all assuming that the simulations predict global warming with certainty- I took you to be stating that but I'm not certain about that part.
That UN cost-benefit analysis you mention sounds very interesting- I'm going to try to track that down, and I may ask you for a link if I have trouble finding it. Frankly, it would be quite a relief to me if it were to turn out that the worst case scenario were not as bad as I had thought. I am under the impression that we could potentially be driving ourselves extinct, which is a possibility I find terrifying, and I would be very happy if I were to discover that this is not actually a possibility.
I guess I still don't see other people's motivations should be a concern here. I feel like the task before us is to find out how risky climate change is, do a cost-benefit analysis, and determine what the best course of action is. Suppose that as you say, there are people who want to limit industrial activity for it's own sake and are just using climate change as an excuse to push that agenda. I can't see how that would in any way cause us to adopt a different course of action.
That UN cost-benefit analysis you mention sounds very interesting- I'm going to try to track that down, and I may ask you for a link if I have trouble finding it. Frankly, it would be quite a relief to me if it were to turn out that the worst case scenario were not as bad as I had thought. I am under the impression that we could potentially be driving ourselves extinct, which is a possibility I find terrifying, and I would be very happy if I were to discover that this is not actually a possibility.
I guess I still don't see other people's motivations should be a concern here. I feel like the task before us is to find out how risky climate change is, do a cost-benefit analysis, and determine what the best course of action is. Suppose that as you say, there are people who want to limit industrial activity for it's own sake and are just using climate change as an excuse to push that agenda. I can't see how that would in any way cause us to adopt a different course of action.
Posted by: Dudest at June 21, 2010 08:59 PM (/OnPc)
38kb generated in CPU 0.0108, elapsed 0.1333 seconds.
30 queries taking 0.1262 seconds, 56 records returned.
Powered by Minx 1.1.6c-pink.
30 queries taking 0.1262 seconds, 56 records returned.
Powered by Minx 1.1.6c-pink.
