Thursday, April 26, 2007

Brilliance in Mystery Novels

Am reading The Mysterious Affair at Styles. It's Agatha Christie's first book, written in 1920. This is very early for mysteries and the debt to Sherlock Holmes is evident.

I've always wondered how authors write mysteries. It must be close to computer programming. Working out the crime, the suspects movements, and then the order in which clues are discovered so as to hint but not give away the murderer. Then I came across this. Wikipedia has this "Agatha Christie told him that she wrote her books up to the last chapter, and then decided who the most unlikely suspect was. She would then go back and make the necessary changes to 'frame' that person."

What a brilliantly unorthodox approach! It's maximizes surprise because not even the author knows the murder while the book is being written. If the story has enough suspects and red herrings then going back and "framing" that person should not be too hard.

Another case of a brilliant idea being a simple one that tackles complexity by simply avoiding it in the first place.

Tuesday, April 24, 2007

The Goldilocks Problem

The IPCC uses climate models to make "projections" about climate roughly a hundred years from now. Recall that the IPCC never makes predictions; only projections. I find it strange that the models are considered accurate for this time range yet not other time frames. The models can't predict climate 10,000 or 1,000 years from now because a slight change in initial conditions or model dynamics can lead to vastly different outcomes over time. I wonder if the models have been run for a 1000 years; do they go runaway or get pinned in some final state?

Then there are short time frames, say 1 or 10 years. Down almost into "weather" vs "climate". Again we are told the models don't work at this range. The reasons? Well, there are too many unknowns in this highly dynamic system for the models to be accurate.

Notice I'm not arguing for extreme time frames such as 1 day and 1 million years. At extremes the significant processes in play probably change. But 1 year and 1000 years is reasonable. It would seem that the same reasons (such as Limited Scientific Understanding of climate) preventing the short and long time frame predictions also apply to 100 year predictions. Or is it some sort of climactic sweet spot -- "just right".

A skeptic would say that 100 years is the perfect time frame for scary global outcomes. Far enough way that we won't be around to see if they're correct, but close enough to affect our children or grandchildren.

Wednesday, April 18, 2007

The Problem With Climate Measurements

I've been reading a lot about climate change lately. One thing that has really become clear is just how poor our historical data is. The Earth's surface is 71% water but there are no weather stations in the ocean even now. Ships have measured SST (sea surface tempuratures) since the 19th century. However they used wooden buckets! Drop the bucket over the side, let it fill with surface water, then pull it back on board (by hand) and measure the tempurature. Holy inaccuracy! When it comes to weather we only care down to about one degree. But climate change is about 0.1 degree per decade! There's no way that measuring water in a bucket will be accurate to even 0.5 degrees. The bucket will either warm up (during the day) or cool (during night or winter) as soon as its pulled back on board. Evaporation in the bucket will cool the water. And in fact a strange jump in temperatures around 1940 caused the climate scientists to add 0.3 degree to all readings before 1940. They surmised that the USA's entry in WWII caused a change from using buckets to using engine intake temperature measurements. Yet a recent study shows that even in 1970 the majority of SST was still done using buckets!

This is crazy. Climate change of 0.6 deg over the 20th century and it turns out that 0.3 was (incorrectly) added to 40 years worth of measurements. Note that if removed, the SST numbers would show more temp change, so this isn't really a climate change denialist argument. It's a data quality argument.

Then there are tree rings. Before we had thermometers, in what's called the "pre-instrument" period, scientists have to use "proxies"; other things that can reveal what temperatures were. Such as ice cores or tree rings. With tree rings the idea is that when temp goes up the rings get larger. Except that there are multiple factors affecting growth: sunlight, humidity, rainfall, soil changes, etc.

The simplest thing for a layman to do here is throw out all measurements except satellite data. The satellite data started in the late 1970s and has shown warming. More importantly the data also shows cooling in the stratosphere and warming in the (near earth) troposphere, which matches a CO2-induced greenhouse effect. Changes in solar activity, for example, wouldn't produce this. However, the CO2 models say that the troposphere should be about 30% warmer than the surface. And the data doesn't show this. So either the satellite data is overly cool or the surface measurements are erroneously warm.

On the Michael Coren show today, two retired profs pointed out that yes temperature is rising, and yes CO2 is a greenhouse effect, but we don't yet have proof that the current rise is because of C02. Climate science is too young a science.

Tolkien's Children of Hurin -- Is It Tolkien?

A new book "by" Tolkien has been published, and reviewed here. Prof Drout gives a rebuttal here, especially about critisms of Tolkien's style.

I really like Tolkien. The Lord of the Rings is one of the most important books of the 20th century, not least for spawning a new genre. However this new book is not really a Tolkien book. JRRT was a perfectionist who took 17 years to write and rewrite the LOTR. He specifically did not publish Children of Hurin or the Silmarillion during his life. For the simple reason that they were not finished. So it seems rather unfair to criticize the good Professor's dense or archaic "style" in these half-finished stories.

Perhaps this style is what he intended for the finished work. He certainly had challenges with tone. He greatly regretted the overly cute tone of The Hobbit. In the LOTR he combined a more serious "adult" tone with a more accessible style. The hobbits, so modern that they owned umbrellas, provided a bridge for readers into the old ways. The Children of Hurin has no hobbits, and Tolkien in his letters acknowledged the challenge that this made in presenting his "lost tales" to modern day readers. I wonder if his inability to solve this challenge is one reason they were never published.

Friday, April 13, 2007

The Great C02 - Tempurature Lag Debate

As someone who is a skeptic of both denialists and AGW, I've been following one part of the debate with interest. Al Gore's film, An Inconvenient Truth, shows a 600,000 year graph of CO2 and of T, "showing" that rising C02 causes rising T. Later it came out, in places like the film "The Great Global Warming Swindle", that in fact what the ice core data shows is that CO2 rise lags T rise by about 800 years.

This has given rise to huge debate (here at RealClimate, or an excellent debate here) over what this means.

The AGW view is that something causes T to rise. There are lots of known somethings: change in solar output or changes in Earth's orbit. The T rise causes oceans to release some of their dissolved C02 (since warmer water can hold less). C02 is a greenhouse gas, so T increases even more. This is "positive feedback". After a while (around 5000 years on average), another something causes T to fall. C02 is re-absorbed into the oceans and we end up back where we started.

My problem is understanding why T ever falls back (eventually) after CO2 rises. If there is positive feedback (or converging feedback), that still suggests a one-way trip. The system would drive itself into a high-CO2/high-T state and stay there. And yes there is solar or orbital forcing going on. But it seems that a small forcing can kick off the T rise, but a huge forcing must be needed to drive the system back to a low CO2 state. I haven't seen this explained anywhere.

Many AGW skeptics get caught up on the lag as proof that significant AGW isn't happening. Their argument is that if C02 lags then it can't be the cause of global warming without repealing the laws of cause and effect. It's a good argument because it throws Gore's false claim back in his face. But the argument only works if there is no feedback; which is impossible given basic physics of water and gases.

They also wonder why positive feedback isn't runaway. If rising T causes rising C02 which causes more rising T, then we should end up with a climate like Venus. However this needn't happen if the feedback decreases. That is if the amount of feedback decreases as T rises then the system won't go runaway.