Lubos Motl, the Czech string theorist and blogger, noted that in Russia, they don't really care about global warming, and Russian scientists are not stupid. He thinks this is because they aren't influenced by Western political correctness, a nice contrast to the politically influenced Russian scientists during the Cold War. And of course its not just political correctness: the US gives about $2B a year to study global warming, which means a lot of objective scientists are paid to basically come up with studies that support the hypothesis. That's a lot of money, and it's naive to think this doesn't influence the findings.
I think that since 85% of Americans believe the standard global warming story (it's anthropogenic, and threatens the Earth), given the sheer stupidity of the average American, that's reason enough to be skeptical. Consider that only 75% of Americans believe OJ killed his wife, and only 67% of American believe the US was NOT involved in 9/11, I think this kind of majority for a much more nonobvious fact indicates the 'facts' are not dominating the public debate, because isn't OJ's guilt more obvious than Global Warming? That is, the models that generate the scary anthropogenically caused global warming scenario are so complicated, how could so many Americans be so certain?
Part of it is the simple story. CO2 is a greenhouse gas. Cars and such create CO2. Ergo, more CO2, more warming. It's like Y2K, which was so plausible because the story was simple enough to explain to 60 year old executive (who had never coded in his life). But CO2 is only 0.036% of the atmosphere, and that is only 15% of the greenhouse gasses (water vapor and methane being the most). As humans only produce 2% of the CO2 annually (most is from water evaporation), I think its reasonable to assume that the natural variability of CO2 makes it difficult to assert that human activity is directly causing the temperature increase. After all, historically, you can't really infer from ice core samples whether the CO2 temperature relation is driven by the temperature causing CO2 to rise (say, via a Milkankovitch cycle), or vice versa. Further, there is the natural variability in the other greenhouses gases. For example, a one percent change in water vapor does the same thing as doubling the carbon dioxide in the air, and water vapor can vary by a factor of 2 day to day, so I'm sure it varies at the scales that matter for these doomsday scenarios. Then there's the CO2 mechanism. CO2 absorbs all radiation available to it in 10 meters. More CO2 only shortens the distance. Either way, anything that can be absorbed by CO2 is already being absorbed.
There are lots of positive and negative feedback loops in an ecosystem like the Earth. Given the persistence of life over hundreds of millions of years, one should think there are many more negative feedback loops (dampening effects), because if we were always on this unstable inflection point, we should see massive extinctions every million years. But the global warming alarmists just assume that the system is mainly full of positive feedback effects (wikipedia mentions 7 feedback loops on their website; 6 are amplifiers).
Anyway, Skeptic Magazine has a neat article on Global Warming, and they note that the current Global Warming Models greatly understate their uncertainty, because the standard errors refer merely to the model's errors--assuming they are true. That is, no uncertainty due to parameter uncertainty, or functional form uncertainty. The author estimates that the uncertainty in the part of the model that addresses cloud cover generates a 100 degree uncertainty band over 100 years. That's right, 100 degrees. With this much uncertainty inherent in these models, for this one aspect, one could reasonably assume these models are pretty worthless for long-run forecasting.
Indeed, to give an example of the uncertainty in the system they are modeling, consider that global temperatures have increased by 0.6°C this century. How much is due to human-based CO2? Maybe 0.2°C. Now compare that to natural variations. Seasonal variations are typically 20°C. Localized effects vary wildly between the equator and poles. Then there are random variations in such things as the Gulf Stream in the Atlantic Ocean which heats Europe, and El Nino which affects the US. But supposedly, out of that maelstrom, climate scientists can deduce the human-created CO2 is sufficient to take us to a tipping point. It's like saying, if you eat that French Fry, you're going to die. Uh huh.
Models are maps of reality, not reality. Often they capture a select subset of what is going on, and so have little overall relevance. For example, the major 7 money center banks in the US had annualized 99% VaRs of about $1.8B in 2007, but still managed to lose an average of $42B in market value over the past 12 months. Now, either their models were really wrong, or, more likely, they weren't that bad, they just excluded a lot of really important stuff (you should read the fine print by the asterisks). Such models are often useful in piecemeal application, but until one can demonstrate realistic, out-of-sample correlations on meaningful applications, they should be applied and discussed piecemeal.