vanesch
Staff Emeritus
Science Advisor
Gold Member
- 5,109
- 20
Bored Wombat said:The majority of the work (but not the majority of these papers) is from GCMs, which of course analyse the cause effect relationship by modelling it.
The statistical and neural net papers look for constraints on the parameter climate sensitivity. I guess they could find a zero or negative one if there is not a causal relationship.
Needless to say, they don't.They don't estimate a correlation coefficient.
I don't know if you see the difficulty to distinguish between both. In a dynamically coupled system, in fact, what is "cause" and what is "effect" is sometimes extremely difficult ; it can be hard to determine causal relationships without having experimental access and be able to change the cause at random and see if the effect is correlated. I have the impression, when I read those abstracts, that people take it for granted that the drive is the CO2, and then try to find what is the dynamical system that gives temperature. I don't see them try the opposite, for instance. I don't see them try to establish by how much CO2 could rise given a temperature change.
I don't see them take as a starting point "let us assume that CO2 doesn't affect temperature (and, say, temperature is driven by other factors, such as albedo change, ocean current change,...), and see where that leads us". No, they take as a starting point that there is a "radiative forcing", that its effect is temperature change, and then go on estimating how much it is, *starting from that hypothesis* that the main drive is radiative forcing. Then they find a value that can explain data. Fine, but that didn't prove the starting hypothesis.
Let me tell you again why all these things bother me. As I said, I'm a proponent of nuclear power, and it is of course an attractive discourse to say that we need nuclear power for reasons of AGW. However, if you make that link too strong, and AGW turns out not to be true, then this will backfire, and I think there are also *other* reasons than AGW to promote nuclear power. So before I want to link the advantages of nuclear power to AGW, I want to know whether the case for AGW is strong enough so that *no possibility exists that it will turn out, 20 years or 30 years from now, that it turns out not to be there*. Because, I don't know if the AGW crowd realizes this, we WILL eventually find out. It is not like a futile debate about, say, the interpretation of quantum theory or something, or string theory or whatever, where there will be always a way to wiggle out, or where there won't be any way to know for sure. No, with AGW, there is no *scientific suspense*: we will find out. We will have serious indications in 20 or 30 years and we will know for sure in 50 - 200 years. So before wanting to link the fate of nuclear power to the fate of AGW, I wanted to find out how strong the AGW case is, so I started playing the devil's advocate. I said: I'm going to read the AGW material (mainly from the IPCC) and I'm going to try to find loopholes in it. Given that I'm not a climate scientist (but on the other hand, I'm a physicist and I think I know enough about stuff such as radiation transport and so on to know the basics), I shouldn't be able to find ANY LOOPHOLE in the argument.
And that's where my disappointment came from: there's no logical case at all ! It is just a pile of suggestive evidence, and a lot of logical errors ! Suggestive evidence, to me, is observations which are more or less in agreement with what one would expect if one willingly takes on as hypothesis the statement to be proved. Logical proof to me is a set of contradictions one finds when one takes on most if not all thinkable hypotheses that contradict the to be proven hypothesis: it is falsifying the reasonable alternatives.
Now, a bad sign to me was that arguments used as "smoking gun" only 10 - 15 years ago as proof for AGW had to be reviewed in such a way that, to me (playing the devil's advocate), undid it of about all the smoke they displayed back then. That doesn't mean that they aren't still suggestive, but they've LOST their convincing power they had back then.
- You may read into the "hockey stick" plots whatever you want, you cannot deny that the display of the 2001 hockey stick is far far more convincing than the different 2007 hockey sticks without the current instrumental data glued on them.
- The 600 year lag between temperature rise and CO2 rise spoils a bit the evident "cause-effect" relationship that CO2 causes a temperature rise
Also, the arguments put forward to prove the two main theses of AGW, namely that the atmospheric CO2 content is fossil-fuel driven and that CO2 is a strong greenhouse gas seem to fail. I didn't say that this isn't true, I'm saying that the *arguments put forward* don't conclusively lead to the conclusion they want to demonstrate (like in mathematics, where there's a hole in the proof).
The 3 points: O2 decrease correlated with CO2 increase, isotopic composition of atmospheric CO2, and correlation between CO2 increase and fossil fuel exhaust, while of course suggestive, do not lead to the LOGICAL conclusion that the CO2 LEVEL is determined by the fossil fuel exhaust, given that the atmosphere is coupled to large carbon reservoirs (ocean, land mass) which have a dynamical relation with temperature, and with the CO2 level.
I jumped up when I read those arguments in the IPCC report (the physical basis, 4AR), because this was the first thing that came to my mind. I didn't see that point addressed at all.
Again, these 3 points are suggestive evidence. If you take as a starting point, the thing you want to prove namely that "atmospheric CO2 is rising because of fossil fuel combustion"(because you are somehow a priori convinced that this is true), then of course the 3 arguments are "in resonance" with what you think already: "Yeah, given that we are already sure that atmospheric CO2 is human driven, it's no wonder that CO2 increases with fossil fuel combustion, and that consumes oxygen, and then see, it was the same CO2 that we find in the atmosphere as the one that was consumed". That's suggestive evidence indeed. It is compatible with the suspected cause.
But it doesn't PROVE it. And to show you that it doesn't prove it, I simply have to find a counter example where the same LOGIC is used (show you that the observed increase of A is correlated with the output of source B, show you that the isotopic composition of A corresponds to that of source B, show you that some other quantity diminishes with B) and nevertheless the inferred causal relationship is NOT true in this case. That was the heavy water and the shore.
Again, this doesn't say anything about what I think about where the CO2 comes from. I also think it is sensible to think it comes from fossil fuel combustion. But it is not because I also think that Fermat's last theorem is true, that I think that your proof of it is correct!
This is so an elementary logical error that I found it extremely worrisome that it wasn't obviously addressed, and what I found even more so, was Bored Wombat's reaction which I'm afraid illustrates this: yeah, ok, with your nitpicking you found maybe a hypothetical logical error, but then the proof wasn't needed in the first place because we knew already of course that it was fossil fuel output that set the CO2 level. I don't even see why we should bother...
This illustrates the cavalier attitude towards scientific inquiry, and not Bored Wombat's, but the whole AGW crowd I am afraid.
If this is really true, then you don't present evidence as logical proof with a logical flaw in it. You simply say that you have to take it as an article of faith that CO2 levels are "of course" fossil fuel driven, and that if you take on that article of faith, then there is circumstantial evidence that comforts that idea.
The next point is: "CO2 is a strong greenhouse gas". I haven't seen any proof of that either, that doesn't work in the same circumstantial way. Most arguments presented follow the same kind of argument: "take it as an article of faith that CO2 is a strong greenhouse gas", and then, starting from that hypothesis, we can explain this and that. And if one asks then if there's ANOTHER way of explaining this or that, then the answer is: not needed, we KNOW that CO2 is a strong greenhouse gas. Ultimately, the argument is: we know the OPTICS OF CO2, it tells us it is a greenhouse gas. Yes, but the optics tells you it is a WEAK greenhouse gas.
In the papers you cite, I have the impression that one doesn't try to PROVE that it is the CO2 that was the cause and one doesn't try to prove that it is the CO2 that is working as a greenhouse gas. It is simply taken as a hypothesis (especially all the statistical estimators! I've done neural network simulation also, in a totally different context as black box models for microwave electronic components when I was working with HP: I can tell you that no neural network is ever taken seriously as a cause-effect dynamical model: it is always a black box).
So when digging into all this, I was appalled at the lack of criticism towards the basic hypotheses on which the whole AGW theory is built. It is as if one has to take them as articles of faith, and then we get a huge pile of suggestive evidence which "makes sense - more or less" when one has taken them already as truth. This HAS some value of course. But it is far from establishing the original hypotheses.
Also, what is worrisome, even more so, is the lack of critical inquiry in those pieces of evidence that don't seem to fit perfectly into the puzzle.
So although there is of course suggestive evidence (if you read it with the "right" mindset), I am far from convinced that one has an "air-tight case". I wouldn't want to convict anyone on such shaky evidence alone, and I'm amazed that this could even be there. After looking into all this, I'm less inclined to promote, say, nuclear power on such a shaky basis, because I take the possibility very very real that all this may turn out simply not to be true, and that it will seriously backfire if ever that's the case.
On the other hand, I don't find much evidence either that AGW isn't true, and in as much that it might be catastrophic, even if it has a 30% chance of being there, one should consider fighting it (the expected damage is then one third of the estimated AGW damage, which is still very consequential). After all, there IS a lot of suggestive evidence. There is a lot of data that doesn't contradict the hypothesis, and that's of course also a way to "prove" things - on the condition of being honest ! In that case, one has to *try to find - even a single case - evidence which might *contradict* the working hypothesis. In as much as one *really tries* to falsify one's working hypothesis, after sufficient time, and if EACH TIME the falsification didn't work, this also builds confidence in the working hypothesis. But there's a danger here. There is the danger that one selects only those cases where the falsification doesn't work, and *one might be inclined to neglect those cases where falsification might work*. That's what a lawyer does, on purpose. And that's the trap Feynman warned against.
And the reaction I see to small indications that there might be some details that "don't fit" scare me. The point Andre brought up should worry climate scientists. What Andre did was taking the cavalier "explanation and proof" that the lagging CO2 in the paleodata act as a positive feedback mechanism, literally. He set up a very simple model with a simple positive feedback - as one does in system dynamics 101 - and looked at the result. But he didn't need to do that: simple positive feedback is known to lead to multistability. So if the explanation was indeed so simple and "evident" as given by the AGW crowd to make the lagging CO2 nevertheless "causal", then this multistability should explain most of the data. And in fact, it does explain part of the data, but it is in contradiction with certain passages such as Andre indicates. And in dynamics, the devil is in the details, and Andre points out correctly that you cannot have that kind of behaviour in a multistable system with positive feedback. Of course his model is way too simple. That's not the point. But if the explanation were so easy (evident proof of the CO2 as causal agent), then the model would be easy to. So this is more complicated.
The comment of "yes, but in reality it is more complicated" is of course acceptable, ON THE CONDITION THAT IT IS UNDERSTOOD.
So my question is: has there been a dynamical model using exactly the proposed causal relationships, which PREDICTS CORRECTLY exactly the passage Andre illustrated ? Has one identified in that passage exactly what was then the cause of the LOWERING of the temperature when CO2 was rising slowly ? Because this is going to be difficult in a positive-feedback model, so this is exactly the kind of relevant detail which needs to be adressed. THIS is the kind of falsification exercise which is worthwhile: we seem to have a piece of data which CONTRADICTS at first sight the working hypothesis, this is an ideal case to test falsification on. Has this been done ?
Last edited: