Coby, even though I have had nothing so far to comment on this excellent series of yours -- in part because I am not a scientist, in part because it is highly unlikely I shall ever engage in polemics with a global-warming denier -- , please know I am grateful to you for writing these posts, and to Gristmill for sharing them.

Your use of "predictive" in this post on the "hockey stick" graph, however reasonable it may be in climate science, strikes me as a bit too defensive, over against serious attempts to "predict," with greater or lesser certainty, in the several historical disciplines.  E.g.:

History.   Where is Iraq going to go next?  Plainly, there is no way to predict that with much certainty.  Nevertheless, it is reasonable to think that some predictions are more reliable than others, if they take fully into account such historical data as the major schism between Sunni and Shia in the third generation of Islam, the artificial creation of "Iraq" out of three distinct Ottoman provinces following WWI, Saddam Hussein's favoring of his own Sunni Arabs and his persecution of Sunni Kurds and Shiite Arabs, and the millennia-long contentions between Mesopotamians and the peoples of the Iranian Plateau.

Biology.  Why do species go extinct?  The pressures on the grey wolf, anthropogenic, are well documented, so we understand well enough why it finally disappeared from the Lower 48 around the mid 20th century.  If the bison had similarly disappeared, the reasons would have been clear enough.  The near disappearance of the California condor is a bit harder to understand, but not by much.  Same with the ivory-billed woodpecker, so that its alleged re-appearance a couple of years ago in the wetlands of eastern Arkansas was not inconceivable.  Hence, we have a very good understanding of how environmental pressures can endanger the continuing viability of species.  The "practical" extinction of the Baiji was quite predictable, and similarly the extinction of the northern population of the African white rhino is predictable.  The well-documented fate of the passenger pigeon, so incredibly numerous in the early 1800s and then extinct within a century, was a shock; but it turns out its life cycle required breeding in continent-sized populations, and the reduction to merely island-sized populations could not be endured.  So, its extinction is related to a great deal that we have learned about "island ecology," and therefore has been instructive to biologists in predicting the grave dangers faced by animal and plant species whose original range has been chopped up by human development and habitat destruction into several "islands."

Paleontology.  Why do some lineages thrive and others fail?  It is interesting that during the Triassic Period, both the first mammals and the first dinosaurs appeared, from already long-distinct reptilian lineages.  If mammals are supposed to be such a superior life-form, then why did they remain in the shadows for the next 170 million years, while dinosaurs became the dominant terrestrial taxa?  A suggestion I like, made by the British paleontologist David Norman, is that during the Triassic, all the continents happened to be in close proximity, forming the super-continent known as Pangaea.  Climate conditions were likely to have been hot and arid: and that would have suited the basic diapsid poikilothermic metabolism of the dinosaurs' archosaur ancestors much better than the homoiothermic metabolism of the mammals' synapsid ancestors.  Hence, if higher temperatures and aridification establish themselves as a normal direction of climate, whether regional or global, we might predict both the extinction of many "warm-blooded" terrestrial vertebrates, and the diversification of the "cold-blooded" reptiles, both anapsid turtles, and contemporary diapsids: lizards, snakes, crocodilians.

Medicine.  Can a sickness be prevented?  Can a sickness be cured?  Can a particular patient be diagnosed with a particular sickness?  In this (partly) historical discipline above all, predictability is of supreme importance.  The use of statistics and probability -- i.e., the collection and interpretation of data from the past -- is fundamental, both for diagnosis and for preventive care.  To say nothing of the importance of "family medical history" as a diagnostic tool.

My suspicion is, you all in climate science would like to be able to make reasonable predictions.  (Certainly, the meteorologists have been much mocked and abused by the misunderstanding public for a very long time.  "They said it was going to be a major hurricane, so I got out of town, and it turned out to be barely a drizzle!"; "They were predicting a blizzard, I was expecting a day off from school, I did not finish my essay on 'The Scarlet Letter,' and all we got was a little dusting!"; "Forecasting the weather is very simple: If I take my umbrella, it does not rain; if I do not take it, it rains.")  But you realize that your ability to explain past data is still too weak to allow you with confidence to predict what lies over the horizon.

Oh well.  Hang in there, I have confidence at least that you will get it before long.  Not that civilization has too many years left ... : )

Chickens are our cousins! So are other sensitive animals! Enough is enough! No more factory farms!

It seems perfectly clear that the various disciplines that consider data from the past will not be able to use such data to the same effect.  If it is not possible to gain a satisfactory understanding of ancient climates, and of past climate change, from such data as can be retrieved from ice, marine sediments, air bubbles entrapped in amber, etc., well, there is nothing embarrassing about that.  Frustrating, perhaps, but not embarrassing.  And it would certainly be a fallacy --- of which I am not guilty -- to believe that a variety of plausible interpretations of climate-related data from the remote past must convince us to have uncertainty regarding present climate patterns.  I am impressed by the high standards of measuring and monitoring "today's ocean-atmosphere system" as you describe them, and have the highest confidence in your conclusions.

In the November post to which you referred me, you wrote this clause: "although we are far from that elusive Perfect Understanding (tm)."  Why the capital P and U?  And what does "tm" mean?

Chickens are our cousins! So are other sensitive animals! Enough is enough! No more factory farms!

"Perfect Understanding(tm)" - tm is trademark.  It is just a bit of humour, making light of the idea that perfect understanding even exists.  I guess I got that by modifying a similar phrase I have read often on usenet discussions about programming, "the Right Thing(tm)".

I think that is pretty revealing of geekiness...here I thought everyone new that! ;)

Thanks for clarifying my misunderstanding of your first comment.

Invent a clever saying, and your name will live forever! -- Anonymous

Capitalized Thing (tm) is sort of like "scare quotes"--it's like saying, "The so-called (insert meme)."  Or sometimes it's more that the thing, whatever it is, is sort of a trademark characteristic or behavior of whoever or whatever creates it.  Like, my Mad Procrastination Skillz (tm) are causing me to be sitting here typing this despite not being done with the paper that was due an hour and a half ago.  Is that clear as mud?

The central "evidence" for global warming presented by the IPCC in their Third Report (Houghton et al. 2001) is the "hockey stick" graph which supposedly proves that current globally averaged surface temperatures are "unprecedented" for the past 1000 years.

The original graph was published in two papers by Mann, Bradley and Hughes (1998, 1999). It purports to show globally averaged temperature anomalies for the Northern Hemisphere for the years 1000 to 1998. Since widespread thermometer measurements are only available since 1860, most of the graph consists of averaged "proxy" measurements, based on tree rings, sediments, corals, ice cores and other indicators of past temperature change. The supposed 95% accuracy is shown in the graph, so that one can deduce that the temperature in 1998 was above previous levels, to that level of probability.

This graph contradicted the opinions expressed in the first IPCC report (Houghton et al. 1990) which claimed that there were higher global temperatures than those shown today during the "medieval warm period" from about 1100 to 1250 AD and that there was a "little ice age" from about 1550 to 1700 AD. The  opinion was expressed that the temperature rise shown from 1870 to 1940 may have been a delayed recovery from this Little Ice Age.

Soon and Baliunas (2003a, 2003b) gathered together many "proxies" and listed them. They concluded, firstly, that the coverage of data, even for the Northern Hemisphere, was not sufficiently representative to justify the deriving of an "average" which could be considered reliable. Their second conclusion was that both the medieval warm period and the little ice age were sufficiently frequent in the observations that they must have existed. Also, there was evidence that temperatures during the medieval warm period were frequently higher than those found today.

Von Storch (2004) questioned the assumptions of "variability" used for proxy measurements in the hockey stick. He showed that the low accuracy of the proxy measurements implies a much larger amount of "noise" which meant much higher figures for inaccuracy.

The most devastating attack on the "hockey stick" comes from papers by McIntyre and McKitrick (2003, 2005). They set out to see whether they could recalculate the Mann/Bradley data and were initially surprised to find that the data were not available and had not even been supplied to the journals publishing the work -- the papers had been published, and believed, without any check on their validity.

After a long period of wrangling they managed to get hold of most of the original data. When they carried out the calculations, however, they found serious errors which, when corrected, changed the whole conclusion that had been attributed to them. They found that they got a higher temperature in the year 1400 than is claimed for today. They found that the shape of the curve had been automatically predetermined. The small amount of actual data before 1550 led to the excessive use, including extrapolation, of several measurements that are not considered reliable by others. Holland (2007) has documented the IPCC's determined resistance to accepting these facts.

Loehle (2007) questioned the reliability of tree-ring measurements, which apply only to summer and are influenced by precipitation. Increased temperature lowers soil moisture and the rings get thinner rather than thicker. When he used all the proxies except tree rings he got a modified record  which restored both the medieval warm period, the little ice age, and the lack of "unprecedented" character of recent temperatures.

The recent IPCC Report (Solomon et al. 2007) has abandoned the "Hockey Stick" graph, but they still will not accept any criticism of it.