The Meaning of ClimateGate

Last week thousands of emails, graphs, and computer programs from a British university were released by an unauthorized source. This was originally reported as the work of a hacker who filtered through who knows how much data. Later, many people including Junk Science decided that the emails, etc. had already been assembled for a Freedom of Information Request (FOI) and someone copied them from that source. Regardless, there are several implications.

Some skeptics have combed through the emails looking for a smoking gun. They found references that imply that researchers were manipulating the data. The warming true believers insist that these quotes are being taken out of context or were poorly worded. This is a dubious interpretation but I am going to accept it. There is no smoking gun here although there might be some smoke.

There are three other points that are very important. Two cast serious doubt on the famous scientific consensus and the third casts doubt on the base figures.

A chief argument against the warming skeptics is that there is a wide consensus in peer-reviewed journals. The emails show that this is a manufactured consensus. In one case a skeptical article was accepted for publication. The warming-believers managed to bend the submission rules so that a rebuttal could be published in the same issue instead of in the next one. In a couple of cases, an editor allowed a skeptical article to be published. The response was a discussion on how to have that editor fired (which happened in both cases). When one publication showed itself open to skeptical papers, the emails discuss how to marginalize the entire publication.

All of this creates a false consensus where dissent exists but is suppressed, then the lack of public dissent is given as proof.

Good science is reproducible. Twenty years ago a peer-reviewed journal accepted an article on cold fusion. Once this was published, other scientists tried to duplicate it but failed. In climate science, reproducibility means making the raw numbers and the methods of compiling them available to third parties for independent verification. There are emails discussing how to avoid doing this. There are cautions to delete emails so that they cannot be requested through FOI. Some scientists say that they would rather destroy their work than allow a skeptic to see it.

Climate science is a complicated field. You can't simply plug figures into a spreadsheet. Most of the final numbers have to be coalesced through custom programs. There is no off-the-shelf product that can do this and the climate researchers frequently have to invent new calculations. This opens them to errors. By hiding their work, they are implicitly admitting that there may be serious flaws in their calculations that they don't want the world to see.

Which brings me to the final point - the computer programs. Someone dug through the comments in a large program. A talented programmer was trying to get the program to work to reproduce some previous series and add new figures to them. He discovered numerous problems. In one case a function that returns the square of a number returned a negative (this is impossible in the real world but in a computer usually means that a result was too large for the field holding it). He tried rolling back the most recent changes without success. Eventually he concluded that the only way to get the program to work would be to go back to the earliest version. This was unacceptable because he didn't know how many problems the original code had that were corrected in later version. The conclusion was that the program was too poorly written and documented to be usable.

This is not the first time that this has been discovered. Several years ago Michael Mann (who is well-represented in the emails) released his results of a temperature reconstruction using tree-rings. This showed a long, stable temperature range until the 20th century when it rose so rapidly that it was nicknamed the "Hockey Stick". Later analysis showed that Mann's math was off and any set of random numbers would produce a hockey stick graph.

So, new, dissenting research is suppressed, the inputs into published papers is hidden from skeptics, and the computer programs used to calculate them can be buggy and unreliable.

Before the scientific community asks us to reshape civilization they need to start adhering to scientific method including opening their research to everyone asking for help when dealing with something that is out of their expertise.

For more information, the entire collection is here. The program comments are here. You don't have to understand programming to see that there are major problems.