Princeton physicist Freeman Dyson has been roundly criticized for insisting global warming is not an urgent problem, with many climate scientists dismissing him as woefully ill-informed. In an interview with Yale Environment 360, Dyson explains his iconoclastic views and why he believes they have stirred such controversy.
http://e360.yale.edu/content/feature.msp?id=2151
A lot of people are acting like he's outright denying climate change, which he is not. I think that he's simply stating that it isn't as urgent as an issue as some people think it is, and I have to agree with him. I've seen people suggest that global warming will lead to human extinction, and I just don't see how that's possible.
Dyson is a very smart guy whom I respect a lot, but being a physicist does not mean you can be effective at critically analyzing any field. This is a very easy trap to fall into. You need to have some education or experience in the field, or you are very likely to misunderstand or miss entirely the details, the techniques, the history of its development and how stuff was figured out. I encounter this all the time with physicists who are skeptical of topics in cosmology. Yeah, it's physics. But you need to understand a lot more than just physics to understand how it applies. :)
Anyway, I'm happy to discuss any particular topics he raises that people here might be interested in. For example, when he says that the strongest warming being in the Arctic suggests that it's not related to human CO2 emissions, which are uniform. Sounds reasonable, but it is wrong. Uniform distribution of greenhouse gases does not necessarily lead to uniform increase in temperature, and climate scientists have known this for a very long time. Global climate models predict that the Arctic should warm most rapidly. That is where positive feedback effects, like the ice-albedo feedback, are currently the strongest. We expect the Arctic to warm up faster than the rest of the planet.
Quote
I've seen people suggest that global warming will lead to human extinction, and I just don't see how that's possible.
Yeah, we should also take a moment to carefully examine the flipside of the climate change opinion spectrum. A lot of people accept climate change, but think it is way more serious of a problem than it actually is.
In particular, the idea that global warming will lead to our extinction... is surprisingly common. But it is not credible. I would actually expect someone to be laughed out of a scientific conference for even raising it, (or preferably very firmly corrected and educated). It a naive conclusion, it does not agree with current understanding, and it is not a productive viewpoint. The bulk of available research on climate change tells us a great deal about what impacts we can expect to see through the next century even under high emissions scenarios, as well as how we can deal with them. It isn't a pretty picture, but it is manageable.
Regarding urgency, it is very urgent. If we want to avoid more than 2°C of warming, then we need to start seeing reductions in our emissions extremely soon. Each year we wait will make the changes we will have to make more dramatic and potentially unfeasible, to the point of requiring net negative emissions. (Ah, the subtle menace of integrals...) Most realistically, we need to start seeing some dramatic decrease in emissions by 2020, or the option of easily avoiding this threshold will probably be lost, and we'll have to deal with worse impacts or even less appealing policies.
Dyson is a very smart guy whom I respect a lot, but being a physicist does not mean you can be effective at critically analyzing any field. This is a very easy trap to fall into. You need to have some education or experience in the field, or you are very likely to misunderstand or miss entirely the details, the techniques, the history of its development and how stuff was figured out.
And Dyson seems wise enough to state the limits of his knowledge. He says that his objection is mainly related to the behaviour and intolerance to criticism in climate science. If people rush to criticise him for that, they're just proving him right.
QuoteWatsisname ()
For example, when he says that the strongest warming being in the Arctic suggests that it's not related to human CO2 emissions, which are uniform. Sounds reasonable, but it is wrong. Uniform distribution of greenhouse gases does not necessarily lead to uniform increase in temperature, and climate scientists have known this for a very long time. Global climate models predict that the Arctic should warm most rapidly. That is where positive feedback effects, like the ice-albedo feedback, are currently the strongest. We expect the Arctic to warm up faster than the rest of the planet.
And we also know that the Arctic naturally goes through significant climate shifts. Many people fall into the trap thinking that if the Arctic is warming, it's due to CO2. If it's not, it's natural variability.
If there wasn't a CO2 change, we would still expect a significant warming of the Arctic in the period from 1980 to 2010 due to the multidecadal circulation patterns, though not yet very well understood. So how do we quantify the attributions? The fingerprints might not be as evident as some like to think.
If the topic wasn't so political, it would be possible to discuss these matters.
Princeton physicist Freeman Dyson has been roundly criticized for insisting global warming is not an urgent problem, with many climate scientists dismissing him as woefully ill-informed.
We here, today, are living in the time when such questions have long become a political and economical reality rather than scientific fact or hypothesis. There are several other "ecological" problems of such quality, such as - ozone layer depletion, overpopulation, deforestation, etc, and all of them have the same problems with bureaucracy and corruption in their basic structure. But "global warming" is the most popular one. As long as people just believe in the future that will resolve their problems by it's own "volition", there will be larger solutions that would rather create more problems than solve anything at all.
There is a difference between informed, constructive criticism of a discipline, and criticism framed from beyond one's own limits of knowledge. There is also a difference between criticism and questioning. You can question the experts on something that you don't understand or think is an error, or you can bypass them and present your differing ideas to everyone as if you know better than the experts. I think you are much more likely to encounter intolerance of the latter than the former.
And that goes for any field, not just climate science. More extreme example: anyone hear of the "electric universe theory"? People not involved in astrophysics who think that they know better than astrophysicists. And they complain that the astrophysical community dismisses them. (Actually, we did consider your model, but it turns out it doesn't work. Also, its motivation was pretty silly.)
Anyway, back to climate science:
Quotemidtskogen ()
And we also know that the Arctic naturally goes through significant climate shifts. Many people fall into the trap thinking that if the Arctic is warming, it's due to CO2. If it's not, it's natural variability.
In general, the resultant change in temperature in the climate system is due to a combination of external forcings, internal forcings, feedback effects, internal climate variability, and heat transfer between subsystems.
This sounds like it would be an impossible mess to figure out. How can we break into this complex interplay of different mechanisms acting to change the temperature? How can we be sure of attributions? It's a good question.
The answer is simple, but not easy. We recognize them in terms of properties and interactions that we can measure and model. It's physics meets planetary and atmospheric science. It works just like any other natural science.
For example, forcings are things that change the radiation flow between Earth, the atmosphere, and space. We know the physics of radiation transfer! We can run global climate models with and without the forcing to see what it does to temperature, not just as a global average, but spatially resolved over the surface. Therefore, we can see how much of the temperature change that we observe is related to a particular process. We can integrate them together and see where the residuals lie, and thus give clues to what we need to study and improve our knowledge of. (And there are some interesting residuals!)
With the current understanding of these various mechanisms, we have a very good picture of what is happening on global and on regional scales. GCMs run with the greenhouse forcings and the other internal processes that we have models for are necessary and sufficient to explain the observed pattern of temperature change. The same GCMs run without the greenhouse forcing do not come even remotely close.
The greenhouse forcing naturally leads to more rapid warming in the Arctic versus the rest of the planet (as well as over land versus over ocean, mountain tops versus lowlands, etc). The warming of the Arctic that we see fits well with what we expect from greenhouse forcing and its secondary consequences, from the models and from basic understanding of the physical mechanisms. It's really rather straightforward.
If we instead suppose the observed pattern is dominated by a mixture of effects we don't understand, then that's a very strange and difficult to explain coincidence. It also demands explanation for what we're getting wrong with what we think we do know.
For example, forcings are things that change the radiation flow between Earth, the atmosphere, and space. We know the physics of radiation transfer! We can run global climate models with and without the forcing to see what it does to temperature, not just as a global average, but spatially resolved over the surface. Therefore, we can see how much of the temperature change that we observe is related to a particular process. We can integrate them together and see where the residuals lie, and thus give clues to what we need to study and improve our knowledge of
QuoteWatsisname ()
The warming of the Arctic that we see fits well with what we expect from greenhouse forcing and its secondary consequences, from the models and from basic understanding of the physical mechanisms. It's really rather straightforward.
You're saying that we know that the models are right, so the difference between observations and the models are the results of other forcings that remain to be studied. Fair enough, but then you go on suggesting that the observations do a good job confirming the models.
With that mindset you will be a slave to your existing models and limit your freedom to explain the observations. And the observations will offer you no help in convincing anyone doubting the correctness of the models.
It's become too political. In the news fairly recently some experts did warn that there is a real possibility that the Arctic ice will expand again in the next decades. But guess what the context was? Whether or not to allow oil drilling in waters that have recently become ice free!
I have a somewhat agnostic view on these matters. I'd like to see 60-70 years of satellite records before making conclusions. It seems perfectly reasonable to me that the warming of the Arctic will now stop or even reverse until the 2040's. I've been to Svalbard almost yearly for the past 20 years and have seen the changes first hand and seen the traces of earlier changes, and I find the changing climate of the Arctic fascinating. The warming has been dramatic compared to what we experience at lower latitudes, but at around 80N such changes are precisely what the climate has been doing there for millennia.
No, I'm saying that we know the fundamental physics that is involved in the processes that the model is attempting to simulate.
We can never say that we know that a model is right. A model is simply a representation of reality. We only know if it agrees with observations, and the extent of that consistency. If the consistency is good, then you can have some level of confidence that the model is correctly capturing the dynamics of those processes -- that your model is a good model of reality.
Again, if GCM's do not correctly capture climate dynamics, if what's really happening with the warming is due to a different combination of processes than what we think they are, then it is an absurdly implausible coincidence of unknown factors that makes the model output consistent with observations. This would demand explanation, such as what is wrong with our understanding of physics, and what are these new dynamics that somehow evaded our detection.
Same idea applies to any other modelling. How can astronomers be confident in their models of stellar structure, for example? There are a lot of really complicated factors involved, and we can't even observe them all directly. They make a lot of simplifying assumptions, too. Why should we trust them?
Why don't you think about this one as a homework question?
If you interpret the differences between observations and your model as the things that your model doesn't cover, you're effectively assuming your model to be right.
QuoteWatsisname ()
Same idea applies to any other modelling. How can astronomers be confident in their models of stellar structure, for example? There are a lot of really complicated factors involved, and we can't even observe them all directly. They make a lot of simplifying assumptions, too. Why should we trust them?
We trust them in the absence of other plausible explanations. For instance, we trust that the heat of the stars mainly comes from fusion. One good argument for that is that it's the only known mechanism known to produce that kind of energy for so long. It cannot be a known chemical reaction, for instance (well, I haven't run the numbers, but someone probably has). The fusion hypothesis makes predictions, and so far observations agree.
As for the Arctic warming, we have two things to consider. Climate models predicting warming in line with the observed warming. And previous periods showing similar warming (before CO2 emissions were significant). So which one is it? Or how much of each, since the warming could be a combination of lesser warming from CO2 than expected and a weak warming cycle.
There's an alternative to improving on the model in the hope of producing a residual of natural variation: observation. Decades of observations. Not for your scientific work, but for the scientists of tomorrow.
Modern scientists have become impatient. They want to produce papers. Right away. Their reputation depends on it, and they can't wait for data. So there is this compulsive thought corrupting science: If data aren't available, there simply must be a way of modelling them instead.
There's a vast amount of scientific papers produced. Nearly everything will be forgotten in a few decades. If you devote your scientific work to careful observations, however, chances are better, I think, that your work will not be forgotten and still be used in 50 years.
So let's see what will happen in the Arctic over the next few decades. Then let's consider the climate models again.
I don't really like saying a model is "right" (though I sometimes may just because its more convenient). I prefer to say it "works" or is "a successful model". Perhaps "the best working model". It's semantics, but I think this is an important point to make. If you like to say it is right to mean we trust that it produces the correct results for the correct reasons, then okay. But a model is a model. It is a simplified mathematical representation of a complex reality with the purpose of having predictive and explanatory power of it. A model can be wrong, in the sense that it is not internally consistent, as like an equation with mismatched dimensions. A model can be "bad" in the sense that it has little predictive power or doesn't agree with data at all. And a model can be better than another model, in the sense that it agrees with observations more closely and/or over a wider range of conditions.
I would certainly say GR is our best working model of gravity, and is a better model than Newton's. But it isn't right. And in climate science, there isn't just one GCM. There are tons of them!
All right, let's talk about stars.
You can definitely argue that stars are powered by fusion reactions by considering their luminosity and lifetimes and lack of any other known mechanism. But that's not a very rigorous argument, is it? How can you be confident that there isn't some other process we haven't discovered yet? Maybe their power source is unique, and we'd never observe it elsewhere! Can you derive the conditions inside the star and show that they are consistent with fusion reactions? What about lifetime? Sure, you can argue the Sun is at least 4.5 billion years old based on geological evidence, but what if you could derive stellar lifetimes in general, from physics?
That's what stellar models are all about. They are not simply describing the power source, they are describing everything about their physical structure. Specifically:
The purpose of stellar models is to derive the central pressure, temperature, and density of a star, the functions which describe how those values change with radius from the center, and how these all change with time (how the star evolves.) (Sorry I didn't explain that earlier.)
The information gained from these models is critical to understanding practically everything about stars and their role in the cosmos, so it is very important that we can trust them. Astronomers have a huge amount of confidence in them. So... why is that? Do you honestly believe "we don't know of any other plausible power source" is sufficient? It's certainly not enough for me! And I'm the astronomer!
I would like for you to think about this some more. It is very important that you understand how modelling works and why we come to trust them or not, if you want to have any reasonable discussion of GCMs and climate prediction. I'll help you more directly next time I post if it is needed, but it'd be great if you can hit on it yourself. Do some research if you have to. I believe you're almost there at the critical insight.
Now let's talk about climate:
Quotemidtskogen ()
Climate models predicting warming in line with the observed warming. And previous periods showing similar warming (before CO2 emissions were significant). So which one is it? Or how much of each, since the warming could be a combination of lesser warming from CO2 than expected and a weak warming cycle.
If we suppose an error in our calculation of greenhouse forcing is compensated by error in handing some process in the Arctic, then this will not reproduce the global temperature distribution and its change over time. Again, we would have to invoke some phenomenal coincidence of multiple different processes being misunderstood. Furthermore, any change in forcings will affect the observed flux of radiation measured from orbiting satellites. So we have a strong independent test of this.
Finally, remember that GCMs can be run backwards. Not just forwards, or for the present time.
BTW, I agree with you on the problems raised by competition in science and the need to submit papers, but I think you are seriously overblowing that. Remember that scientists are trying just has hard to prove each other wrong as they are to prove themselves right! The biggest fear is not missing a deadline. It is of being proved wrong in front of your colleagues! Every paper they publish puts their reputation on the line!
Yes, we can say that models work. If observations show that the model has a problem, we might still want to use it if we don't have better options, either by acknowledging its limitation or by adding "epi-cycles" (consciously or not). For instance, if we only had Newton we could quantify its accuracy, or we could work out explanations for why observations disagree. Like, saying that planets that move slightly unexpected are influenced by some unseen dark matter in the solar system. We could say that we don't see it directly, but we know it's there because its the gravitational influence. But such ad hoc explanations can be dangerous because they lead us to think that there is no need to rethink the model. Because they still work.
QuoteWatsisname ()
That's what stellar models are all about. They are not simply describing the power source, they are describing everything about their physical structure.
I wasn't saying that stellar models are just about the power source. I simplified for the sake of an example.
I find it difficult to see how stellar models are relevant to Arctic warming. It's not about observations not well explained by the model (like, how exactly the sun's corona is so hot). It's about observations fitting the model! But fitting more than one model.
QuoteWatsisname ()
If we suppose an error in our calculation of greenhouse forcing is compensated by error in handing some process in the Arctic, then this will not reproduce the global temperature distribution and its change over time. Again, we would have to invoke some phenomenal coincidence of multiple different processes being misunderstood. Furthermore, any change in forcings will affect the observed flux of radiation measured from orbiting satellites. So we have a strong independent test of this.
Why don't we have a precise number for the climate sensitivity, do you think?
QuoteWatsisname ()
The biggest fear is not missing a deadline. It is of being proved wrong in front of your colleagues!
If they fear that, they're bad scientists! They should immediately congratulate their colleagues.
If they fear that, they're bad scientists! They should immediately congratulate their colleagues.
I'm speaking of fear of their work being discredited due to lack of rigor, not that they did good work but made a less trivial mistake or missed something more fundamental. I don't think that Newton would not congratulate Einstein. :) It takes a lot of time and effort to do a study and write up an academic paper to present it. This is wasted effort if you are sloppy, make a lot of errors, or don't back up your conclusions with appropriate evidence.
All right, let's have that big discussion on modelling. You haven't quite struck gold yet. You had some insight earlier on where confidence in stellar models might come from, but did not pursue it further, and now you are wondering what the whole point of the exercise is. So, I will review that. There is also an interesting thread on dark matter we may follow. I have written another section on that as well. Each section ends with a set of questions. I would like you to read both sections, and then choose a set of questions from one of them to answer.
Purpose: I have been asking you to make a small effort towards learning about the development of models of stellar structure so that we can discuss them and hopefully find insight on the matter of how we gain confidence in those models, and extend those insights towards modeling in general and the topic of climate modelling in particular. There are also some interesting parallels between the principles used to construct stellar models and for climate models. There is a reason I chose this particular comparison. I did not pull this exercise out of thin air or as busy work. :)
We've reviewed your example of the models predicting fusion reactions at the center, which is consistent with our lack of knowledge of another probable energy source. This is an example of the model producing output consistent with what we expect, which is indeed a good sign that we're constructing the model correctly. However, it is not adequate by itself. (If that is all we have, then astrophysics is in a very sad state!)
My counter was that there could be another energy producing process unique to stars, and concluding that the model got it right because it predicts fusion might prevent you from critically reviewing the model and discovering the true energy source. Another potential problem is that you don't necessarily know if the model is predicting those conditions for the right reasons, or if it is just a coincidence, and the equations describing the structure are very wrong. Then there's the question of whether it evolves over time properly.
So, there is more insight that we need. Where do we get it? Consider this situation:
Challenge question: Suppose two researchers, Aaron and Bob, have produced two different models for the same star. Their models yield the same overall mass, radius, and surface temperature, but Bob's model has a smaller radius for the hydrogen-fusing core. How would you test between these two models? What would give you confidence that one is better than the other?
Hints: How does one begin constructing a stellar model? What principles of physics apply? What essential equations must be satisfied? What else can you say about the two model stars from the information provided?
You did raise another point, explaining that when there is a discrepancy between the model prediction and observation, we must be careful when considering whether this implies the difference is due to a new phenomenon, or a flaw in the model. This is absolutely true, and I think it may be fruitful to discuss it further. Your example was a hypothetical of dark matter in the solar system, but it is also very relevant to the real and current situation of dark matter in cosmology. Would you prefer to follow this track over the one on stellar models? If so, here is a primer:
"Dark matter" is a key component of the concordance model of cosmology (LCDM model). We have enormous confidence that what we see as evidence of "missing gravitational mass" really is dark matter, a type of particle unknown to the standard model of particle physics.
If observations of celestial motions are inconsistent with Newtonian gravity, but are consistent with Newtonian gravity with additional matter, then there are the following possibilities. -The observations are erroneous. -The model of gravity needs to be modified. -There really is more matter, but it's the usual stuff we already know about except in a form which we can't detect easily. (MACHOS) -There really is more matter, and it is fundamentally different than the usual stuff. (WIMPS)
All of these explanations start out ad hoc, and any combination of them could be right. The way we progress forward is to test them. Scientists will examine all of these possibilities and see what works and what does not. If there is more than one possibility that can't yet be invalidated, then it is fair game to pursue and you can bet there will be researchers following that thread.
It has been shown that the observations taken as evidence of additional, unseen mass are robust and have been found through many independent methods. The "missing mass" problem is very real.
It has also been shown to be possible to modify our understanding of gravity to explain some of the observations (e.g. rotation curves, by supposing that the force drops as 1/r2 for small distances and as 1/r for large distances). However, this modification does not fit with all observations (e.g. the evolution of the universe's expansion rate.) What we find is that there is no modification that agrees with all observations simultaneously. Any modification ends up requiring the existence of more mass somewhere, anyway. Thus, the hypothesis that what we see as evidence of dark matter (DM) is actually evidence of a problem with our understanding of gravity... is not a very successful one.
Supposing the existence of DM within the framework of the current model of gravity, on the other hand, has been remarkably successful. It explains all of the observational problems it attempted to: the same abundance and properties of the hypothesized dark matter agree with the rotation curves, lensing, formation of the cosmic web, and cosmic expansion given by the Friedmann equations. This hypothesis also has good predictive power: in order to fit with those observations, DM particles must be non-relativistic (or they would not clump on galaxy-sized scales), lack electric charge (or they would interact with light and we would see them), and be weakly interacting with other particles and themselves except by gravitation (or they would cluster into objects much smaller than galaxies, and again we would easily detect them). This combination of properties makes predictions. For example, it should be possible for regular matter and DM to be separated during galactic collisions. Indeed, this is what we observe!
Challenge Question: How do we gain confidence in the attribution of the "missing mass problem" to actual additional matter in the universe, or to a flaw in our understanding of gravitation, or of a combination of other dynamics in the universe, such as proposed by the electric universe theory?
Quote
Why don't we have a precise number for the climate sensitivity, do you think?
We have never tested what the effect of doubling the atmospheric concentration of CO2 on hundred-year timescales is on Earth's temperature, so we do not know what the correct value of climate sensitivity is. We can only apply a variety of methods to try to predict it.
We can try to figure it out from previous climate states, but there is no perfect analogue. For example, if you derive it from a climate change where the ice sheets vanish, you'll obtain a much higher sensitivity than one where the ice sheets are preserved, because the melted ice decreased the albedo and raised the temperature further. This effect alone can make the difference between 3C and 6C for climate sensitivity.
We can try to figure it out from modelling (GCMs), but we run into a similar problem: we cannot precisely determine the future contribution from the ice-albedo feedback because we cannot perfectly model the dynamics of ice-loss, which is nonlinear and coupled with the ocean and atmosphere. This is an example of one of the more significant sources of uncertainty in climate sensitivity, but it is not the only one. We discussed another one a while back regarding the forcing due to convective clouds at low latitudes, which GCMs were found to under-represent.
We can try to figure it out from volcanic eruptions. Volcanic aerosols reduce the flux at the surface; a negative forcing and the opposite effect of global warming by greenhouse gases. So we can look at how the climate responds to a given decrease in forcing by eruptions, as well as its recovery. Pros: These changes are very fast (timescale of years), and so are easily distinguished from other causes of climate change such as internal variability. Cons: They're almost too rapid. They do not allow time for many feedbacks, which would be relevant to our present situation, to take a large effect.
And there are a variety of other methods as well, all independent of one another and all with their own strengths and weaknesses. We plot a distribution of results from them and find that they all predict a positive value for climate sensitivity, with a probable range of ~1.5C to 5C, and a most probable value of 3C.
With all of that, how can we be sure of the attribution of the current warming (with respect to global vs. Arctic and greenhouse forcing vs. feedbacks and variability)? Because, as an example, we know what the ice-albedo feedback is in this case; we can put that change in coverage into the GCMs directly, rather than have to predict how it will change in the future. More generally, we know from observations what the changes in the physical properties of the system have been, so we have a good handle on how the associated radiative forcings have changed.
There are other sources of confidence in that as well, which the above exercises will hopefully help you to key in on. Good luck!
ENG
New site
United States 
Norway 
Pirate 
Russian Federation 