Thursday, February 24, 2011

Judith Curry vs Gavin Schmidt

Gavin Schmidt has been fighting against Judith Curry for some time (on discussion pages of Curry's blog - Part I and Part II - Real Climate, Watts' blog, and other places); see e.g. Climate Progress. The ecoterrorists have even established a special anti-Curry website which is totally incoherent but it is still a testimony how much those folks dislike her, a new heretic.

What has made various Gavin Schmidts upset was Judith Curry's endorsement of Richard Muller's statement that the hockey stick-like graphs were pieces of dishonest science. No doubt, your humble correspondent agrees with this obvious observation and it is more likely than not that you agree with it, too.

Judith Curry has been viewed as a "mainstream researcher", which has been identified with a "fearmonger" for quite some time. Steve McIntyre, on the other hand, has been a leader of skeptics. Those two fine folks actually have close opinions - they're also personally close to one another - despite their different backgrounds and former "affiliations". I think that both of them are kind of agnostics.

It's clear that as Curry observes that her colleagues in the system are not exactly fair people, she will be led to re-investigate many new things that she has previously believed. But even though your humble correspondent and many others have gone through a similar transformation (concerning climate change questions) in the past, I do think that her viewpoints will never quite coincide with ours.

But in one context, Curry has become a more "clearcut" skeptic than Steve McIntyre himself. And it's the question whether there's dishonesty behind all those mistakes that "accidentally" always happen to skew the results and predictions in the same direction. Steve McIntyre has avoided the term "dishonesty" for many years.

But it's just not plausible that all those problems occurred by chance. If they had occurred by chance, only about 50% plus minus 1/sqrt(N) (times 100%) of the error would make the result look "more convenient for the alarmists", while the rest would spuriously indicate that "the alarmist case is [even] weaker than it actually is". However, because N is very large and the percentage of the results that has been skewed in the alarmist direction is very close to 100%, it is implausible that the errors occurred by chance.

This argument is statistical in character. So much like the statistics of weather events, it is not enough to attribute individual events to dishonesty - much like individual weather events can't be attributed to overall trends. However, in the case of the dishonesty hypothesis, the statistical argument allows one to say that there have to be a large percentage of researchers who are simply being dishonest. There has to be a trend. (The conclusion of statistics of various weather events isn't necessarily the same.)

John Houghton, the ex-boss of the IPCC scientific assessment working group, answers the question whether the hockey stick has ever been important for the IPCC, a point that the alarmist deniers love to deny, too

The hockey stick graph has been one of the most painful - and potentially most costly - published results in the history of science. It's known that the actual available data don't allow us to conclude that the 20th century climate change was "qualitatively" faster than the climate change in the previous millennium. In fact, more impartial analyses make it much more likely that the temperature change in the 20th century didn't dramatically differ from the temperature change in many previous centuries.

Moreover, we actually know that those graphs were deliberately produced to suggest a certain result that was pre-decided. Michael Mann has used a statistical method that effectively create a graph that looked like the low-variability average tree graphs before the year 1900 - but as a sharply varying function after 1900. He used a seemingly complicated "principal component analysis" and we may have thought that his methodological mistake - he used a method that produces hockey sticks even out of red noise (as explained on this blog many times, including Mathematica notebooks) - was a mistake.

However, the ClimateGate e-mails have demonstrated that it wasn't a mistake. In fact, his colleagues such as Phil Jones were totally aware of the fact that Mann's trick was de facto equivalent to splicing graphs obtained by two totally different methods - low-variability tree reconstructions before 1900 and thermometers after 1900. Jones wouldn't even pretend that one needs a legitimate method of principal component analysis to get the result. He realized that Mann's method is de facto about the splicing of the graphs.

By splicing graphs, you may always create the feeling of a discontinuity where there's no one. The two methods that you merge have different systematical errors. In this case, one of them, the trees, clearly underestimates the variability. They do so not only in the past but also in the present era. The "divergence problem" is a demonstration of the fact that the trees don't even see the recent warming that is claimed to be fast.

If you consistently draw the graph for the trees only, you won't see anything special happening at the beginning of the industrial revolution or the early 20th century. You get random graphs. The same is true for any other method that has some tools to reconstruct the temperatures from the pre-industrial era.

So those folks were deliberately cheating in order to create a "message" that they have evidence that the climate recently began to behave very differently in recent decades or in the 20th century - even though they manifestly have no such evidence. This situation has been analyzed and discussed for years, it's crystal-clear, and everyone who denies that Mann's and similar graphs are just piles of dishonest people's feces should be immediately removed from any decent discussion about the climate. Those Manns simply don't belong to it.

The same comments apply to other tricks, such as the cherry-picking of the initial years where various reconstructions begin. There's no problem with having a beginning: but if you want to claim that an unprecedented event occurred during the interval you study, then it is unforgivable for you to omit the pre-history that could actually falsify (or does falsify) the label of "unprecedentness". Of course, if you're really not making any claim about the distant past, before the beginning of your graph, and if you're making no statement about the comparison of the present and the pre-history, you may start wherever you want. But if you do make such a claim, cherry-picking of the beginning is an example of dishonesty.

Judith Curry is so peaceful and eager to build the bridges that she doesn't even emphasize that Mann et al. should spend years in the jail or a few seconds on the electrical chair, which they definitely should. Still, her ecoterrorist colleagues are dissatisfied even with her peaceful approach.

To preserve a basic degree of moral hygiene on the Earth, Gavin Sch(m)idt, much like any Schidt, should be splashed into the toilet.

Wired for extinction

An amusing bonus: alarmist journal Grist has asked the question why the Americans don't belong among the most obsessed alarmist nations of the world. Grist's answer is that Americans are wired like deers, for extinction. A nice scientific discovery! :-)

People in the Scientific American discuss the very same question: why are Americans so ill-informed about the looming climate doomsday? They don't even have a deer theory to offer. ;-) Let me tell you: the Americans are actually more well-informed than others.

No comments:

Post a Comment