Crichton is simply bright, deep, and realistic about the history of science. He chooses two examples; let me not tell you which ones because his text is written in such a way that they appear as a surprise. The social circumstances surrounding these examples have striking similarities to some of the presently popular scientific hypotheses and their political support.
Some readers don't understand a basic point about this text by Crichton. This text is not about any particular question related to the climate. In fact, it uses very different examples; nevertheless, it indicates something if the proponents of the currently fashionable theories are able to recognize themselves. Instead, Crichton's text is about the scientific method itself. It is about the huge difference between a scientific argument and an argument supported by political power.
In contrast to a comment that Greg Kuperberg is going to write, Crichton does not argue that 70% of the climate scientists - or what's exactly the percentage of the "alarmists" - must be wrong because some other scientists were wrong in the past. But he certainly does say that the question whether an opinion is backed by a majority of those paid as scientists is irrelevant for the scientific search for the truth and we have an overwhelming evidence that majorities as large as 80% do not mean much and are still consistent with the 50:50 odds for both answers.
The majority of climate scientists who - at least officially - endorse the global warming paradigm is equal to the majority of the string theorists who have voted against the anthropic principle. I definitely do not think that this vote is one of the main reasons why the anthropic principle goes in a wrong direction. (The likely fact that it goes in a wrong direction has very different reasons.)
Sometimes people are right, sometimes they're wrong. The same thing holds for the scientists. There is no universal rule that a majority must be wrong or right - and it is pretty easy to manipulate a majority.
I wonder how it's possible that someone with IQ above 90 can still misunderstand this crucial point in the 21st century, after so many historical examples involving the church, various totalitarian regimes, and the majorities that these systems controlled in which brute power of manipulation had tried to neutralize the scientists who were so often right. And in many cases the regimes succeeded (temporarily). It's just amazing that someone still misunderstands that the scientific truth and the political influence are two very different things. Maybe they're below 90 after all.
What the climate is gonna look like in 2030 is mostly unpredictable with the current tools accessible to science, and I don't think that it is a terribly important question because the climate is guaranteed to be similar to the present one; and also guaranteed to be slightly different; and the local differences will definitely be more important than some "global" appraisal. But if one actually wants to develop science about the average temperatures in 2030, it must follow the usual scientific standards. Avoiding bias is completely crucial especially because of the large fluctuations (and noise) that obscure (or artificially enhance) any potential signal (much like it was the case of lysenkoism).
If one looks at the bias in the newspapers and elsewhere and the systematic intimidation of the people who generate or defend politically inconvenient data ensembles or theories - and I've seen just far too much of this thing - he can estimate that even if the bias were reduced to 10% of its present value, the bias would be enough to change a prediction of the warming to a prediction of the cooling if its sign were reverted. There are so large variations in the data that filtering out 5% of the data ensembles is typically enough to change the predicted trend entirely.
Some of us are pretty strict about the standards of scientific integrity - and if someone says that a piece of data can't be accepted because it would damage someone's political goals (or uses "oil companies" as a part of the sentence that is supposed to justify something), the scientist is more or less neutralized in my eyes; be sure that this neutralized list includes very many people who study the climate and who may enjoy the respect of others but certainly not mine. Of course that one can't be quite dogmatic. Someone may be slightly biased towards an answer. Whether or not his bias is significant may be evaluated quantitatively as long as we ask quantitative questions.
Estimate how many "inconvenient" ensembles of data a scientist is ready to reject because of his or her bias (and how many "desirable" ensembles will he or she overblow). Equivalently, in the case of computer models, how many potential physical effects contributing to a physical observable are typically included just because they lead to the "right" answers and how many other effects are omitted.
Also, estimate how many statistical and other tricks that have the capacity to change the final result the person is willing to insert to his or her calculations. Invent a generic meta-ensemble with roughly the same statistical characteristics as the data from the real world and a calculational procedure with the same number of tricks. Eliminate the appropriate percentage of the most inconvenient data ensembles and choose the appropriate fraction (majority) of the tricks that twist the outcomes in the "right" way. Compute the averages and apply the procedures. Compare with the unbiased answers.
When you do this exercise, you will see that the bias of the semi-quantitative newspaper articles is roughly 50 times as big as the signal; for example, it means that if you compare the articles about various places that are cooling and various places that are warming, you may use them to derive that the newspaper-based warming is currently 50 times as big as the observed one.
In scientific literature the bias is smaller than the newspaper bias but it is still bigger than the actual signal. There are many particular ways to derive this conclusion; they differ in details but the qualitative outcome is the same in most cases. Just try to do one of these things quantitatively. Of course, the numbers above are very fuzzy.
This inequality means that the actual climate science, as done in the real world of the early 21st century, is closer to non-science than science. Of course, this is a statistical statement that does not mean that every particular paper is non-science! But the overall set of all articles is closer to a conglomerate of noise, bias, dirt, and politics than to scientific results. The desired signal - usually a catastrophic human-driven climate change - probably does not exist if the things are done properly and if the contamination by bias is eliminated - much like the ESP phenomena disappear if you do the experiments right. Its appearance is due to bias from 98%. It's just too bad. Moreover, these pernicious tendencies may start to spread to other fields of science which would be really sad.
