Well, my weekend spent wandering the web did produce some valuable insights. Thanks to a reference to Lorenz attractors I was drawn once again to the subject of chaos theory, and this time found some interesting material.

The Lorenz attractors are part of a set of mathematical equations developed by Edward Lorenz in his work on weather prediction. To cut a long story short, he discovered that very small differences in the initial data sets produce very huge differences in output. The result was his determination that weather was so responsive to minute variations that weather prediction over any substantial period of time was all but impossible.

Bear in mind that this discovery is one of the foundations of chaos theory. In fact the whole overused phrase "butterfly effect" comes from Lorenz's work. So, as far as I know, this work has not been disproved and no one has seriously challenged these assumptions.

And yet we hear daily how "computer models" are predicting the effects of anthropogenic global warming.

Now, I know there is a difference between predicting state at a specific moment and predicting overall trends for an entire system, so I will deal with both. Please do not think I am setting up a straw man here, there is a very good reason for my approach.

Now the gross models for AGW are simply models of the aggregate systems, they model the weather on a very simple level, allowing for very few confounding factors. Perhaps some allowance for increased plant growth, the oceans as heat sinks, maybe increased cloud cover. And all of those models, assuming the premises are correct, do predict global warming.

But such models are absurdly simplified. As we know from the massive impact of El Niño and La Niña, as well as recent discoveries such as heat funnels and other interesting phenomena, things such as winds, ocean currents, cloud movements, rainfall, and so on, all have a massive impact on temperature.

And the AGW advocates recognize that, which is why they have started using weather models in their proofs. They are trying to develop models to show the effects of warming on weather and then use those models to prove their theories.

But there we run into that pesky discovery by Lorenz, that a very tiny change in data produces massive differences. And since we don't even know all the factors effecting weather, we probably don't have all the relevant factors. But even with those factors we do measure, such as temperature and wind speed, the limits of our measurement may mask differences that make the system have different results from apparently identical starting conditions. For example, we may have the same temperature from point A for two different days, but the weather differs because the temperature 10 feet away is different. We simply cannot gather enough data to make realistic predictions.

And we have seen this in the computer models of AGW. The more realistic they become, the worse their performance. It has become common practice to modify the equations every day so that they can generate today's data using yesterday's conditions. The models are simply unable to accurately predict even a day into the future with any reliability.

All of which makes it difficult to take seriously those who argue that the debate is over. There are a  number of reasons to doubt even some of the basic premises of the AGW theories, but, supposing we grant those assumptions, they still have no reliable means to show what the result of those assumptions will be. Either they must use an absurdly simple model, or else they are unable to produce reliable results.

It does not make a strong case for the theory when their models can't even predict tomorrow, yet they ask us to believe their predictions about twenty or forty years in the future.

POSTSCRIPT

A similar problem exists in attempts to "scientifically" manage economic systems. The supposed supply and demand curves are impossible to determine, no matter how much data is gathered, as they are aggregate subjective valuations, which change by the second. All sampling, being historical, tells us nothing about the present. And even if it did, if we could sample everyone instantly, we would still have a curve valid for that one instant and no more. And if we can't even establish something as simple as a supply or demand curve, how can we scientifically manage the market?

An older essay entitle "Knowing Our Limits" relates to this. While it touches on many topics, from atheism to economics to global warming, it basically argues that there are boundaries to what we can know, and at some point we need to admit that data is simply insufficient for us to state anything with certainty. (And before any atheists get offended, all I argue is that those who say atheism is proved by evolution or other science are as misguided as those who find proof of G-d in science. Religious faith and science are unrelated and one cannot prove or disprove the other.)