There was a range of results based on given models. The natural varience was significantly wide of all of them. That's why, like Halley, I'm confident.
I'm trying to figure out what you're saying here but you make little or no sense.
Simple fact: the reconstructions had non-overlapping confidence intervals. They are incompatible; i.e., they are inconsistent.
Simple fact: the order of magnitude of inconsistency is the same as the difference required to make Halley's results insignificant
Simple fact: the inconsistency between reconstructions is a form of error.
Consequence: the magnitude of the effect measured by Halley is no larger than the error.
Direct quantification of the results show that the 4x (which isn't really even 4x, it should be 2.7x, due to unaccounted for uncertainty in the instrumental data) is not "enormous", is not "huge". It is just 4x (2.7x), and since there is a known (but unquantified) systematic bias present in the reconstructions, the conclusions you wish to draw are a stretch. You can hand-wave about how likely they are, but you can't quantify it, and if you can't quantify it, in science the claim is without merit.
And that's only with respect to the statistics. Again, none of this deals with any physical evidence, and there's a very real problem with what the world would look like were natural variance that strong.
Oh, but analysis of LTP is all about physical evidence. I appreciate you don't understand this.
But I'm curious how confident you are. You want to have a friendly bet on this matter? What do you think the odds are that current warming will be explained by natural variance? I'd say pretty high.
You see? You don't understand. I am not confident in the results. That's the point. The confidence is overstated. You are the one with the blinkered confidence. I can see large swathes of unquantified uncertainty that you are just ignoring.
But that is the problem. Because you ignore the uncertainty, you will never be a good judge of the situation.
Additionally, the difference between the models in question is well understood. There are clear factual explanations of how they yielded different results. There is no such evidence to support the idea that natural variance will ultimately be found to be many multiples greater.
Yes, there is, any system using variance matching (as all reconstructions do) in low signal/noise regimes will tend to systematically underestimate variance outside of the calibration period. This is very well understood and applies to all reconstructions.
What is unknown (as Halley correctly observes) is the magnitude of the underestimate. And without that magnitude, you are left with hand waving. Simply asserting there isn't a problem doesn't make it so. That isn't how science works.
The "unlikely" event that occured once did so in a range that the second "unlikely" event doesn't approach. It's illogical to hypothesize, absent any evidence, that because two models vary to a certain degree, natural variance must also have the same property.
We aren't absence of any evidence. The effect is known and documented in the press (Halley even cites an article for you). Variance matching suppresses variance outside the calibration period.
Yes, that was the section I linked. He says, "How much, we don't know precisely, but we do know that it being enough to explain warming with natural variance is unlikely."
He posed the question to dismiss the theory behind it.
Ugh. No, he didn't dismiss it. He clearly said we don't know the magnitude. He then ignores it. But how can you dismiss something you have no idea of the magnitude of?
Incidentally, Halley's work isn't terribly new. The points were largely covered in Koutsoyiannis 2007, in which he compares six proxy records to the instrumental series:
Thus, if one accepts one of the other four series as representative of the past climate, one can readily conclude that the observed temperature variation in the last years is not a result of natural dynamics. In other words, there is a statistical significance in the change of standard deviation, so no additional statistical test is needed. Furthermore, with simple statistical calculations with the standard deviation estimates shown in Table 1, we can easily classify the proxies in two groups (one is J98, M03, M05 and the other one M99, B00, E02), each of which contains series compatible to each other but the two groups are incompatible to each other. This makes unrealistic the possibility to use all series simultaneously in a global statistical approach and highlights once again the uncertainty involved in the use of proxy series.
Koutsoyiannis (more correctly) notes the disagreement is enough to make the natural dynamics claim, but it is clearly absurd to do so when two things measuring supposedly the same thing enable you to reject the possibility of natural dynamics. That is clearly an absurd result. I'm aware the existance of such an absurd result is no problem to you. But it is to people who follow the scientific method.
Ref: Koutsoyiannis, D., and A. Montanari, Statistical analysis of hydroclimatic time series: Uncertainty and insights, Water Resources Research, 43 (5), W05429, doi:10.1029/2006WR005592, 2007.
