First, let's look at the indisputable fact that climate changes have existed in the past. These invariably coincide with natural variations in the composition of the atmosphere.
One of the most dramatic of these changes occured some 3-4 billion years ago, when tiny microscopic organisms started converting CO2 into O2, changing the atmosphere from a scorching, uninhabitable condition into one in which life could exist.
The first problem that this raises is this: One of the largest counter-arguments to Global Warming is the claim that individuals cannot make changes on the scale that natural catastropic events do. Ummm, that's probably correct, but we're also a great deal more powerful than even large numbers of single-cell organisms. Yet they were perfectly able to orchastrate changes on a global scale.
The second problem this raises is just as significant. It shows that life alters the climate. The mere EXISTANCE of life will alter the climate. This is the basis behind a theory I've yet to see any significant contradiction for, which is James Lovelock's "Gaia Hypothesis", in which he argues that organic and inorganic processes are constantly shifting the climate, and that the climate you observe is simply the "best fit" for all processes involved.
Second, let's also look at the equally indispitable fact that climate changes are NOT cyclic. They're "chaotic", in that they never repeat at the same interval, no two climate shifts are the same (or even remotely similar), and that it is impossible to predict the behaviour of the climate at any given point in time, based on past points.
This also raises two problems. First, chaotic systems (also called "non-linear dynamic systems") are very sensitive to conditions. One tiny difference in the values at one point can (not will, it's not predictable) cascade into a HUGE difference at another point. That cascade effect is extremely complex, and is the primary reason that even "simple" systems, such as hurricanes, can follow seemingly-bizare paths, even stopping, reversing, or spinning round some geographic feature.
Now, if you can't predict what a chaotic system will do, and it is extremely sensitive to values, then you don't need a huge cause to generate a huge effect. A single vehicle =can= (not will!) shift the path of a storm, if it's placed "just so", the engine temperature is in "this range", and the velocity is within "these bounds".
On these grounds alone, you might want to hesitate before assuming that a power station can't affect the climate, simply becase it's smaller than, say, a volcano. The power station might be in just the right point, or have just the right characteristics, to start a cascade. The volcano might, too. Or neither. The only way to know is is to stop assuming and start number crunching.
The second problem this raises is related. Because chaotic systems are so utterly unpredictable (though =some= may be self-similar), they cannot (obviously!) be predicted. There is simply no way to seperate out human activity from natural activity. The systems are too enmeshed.
What does this mean? It means that we are not able to tell if the globe really IS warming, or to what degree that warming is due to human activity. It's NEVER going to be provable. It's never going to even be "probable". The best you can ever manage is a vague "possible", and that's what we have. The strongest possible determination of Global Warming has been issued, and all it amounts to is a "maybe".
Conclusion: Global Warming may be real. It might not be. There is simply no way to tell. The effects are (just barely) deterministic. Sometimes. (You're on the borders of Quantum Mechanics, when you start dealing with atoms and small molecules. And QM doesn't give a damn about anyone's need to be sure.) But even when they're deterministic, the system is non-resolvable. It's simply impossible to know what the system will do, for any given set of values.
This goes into weather forecasting. It's usually not too bad, over a period of a few hours, but even then it's still no "certain", and you can be sure that the "exact" values they predict won't occur in the majority of the area covered. Why? Because the system is just too unstable, too variable, too dynamic to predict.
And that produces the final problem. We don't know if cutting CO2, SO2 and NO2 levels will resolve anything. There's a significant chance that it might make things worse! The only way to find out is to do the work. You can't know what will happen unless you do. And you can't know what any other path would have done, had you followed it.
If you like certainty, tough. Nature forbids certainty, and it's Nature that deals the cards. Or sells the science rag.
Do I think the Kyoto Accords should be followed? Yes. Because although I can't be "certain" as to the consequences, I believe that the risks are much lower. I also believe that a competent industry can turn the problem into a gain. Total Input = Total Output. Total Output = Useful Output + Waste. If you reduce Waste, and keep Total Input the same, then your Useful Output increases, so increasing profit.
Anyone in "industry" who opposes the Accords does so because they're too naive or too scared to seek the opportunity in the risk.