I Find This Argument Uncompelling

I am skeptical of some but not all global warming claims, but must admit that even as a skeptic, I find this argument by James Lewis uncompelling:

Now imagine that all
the variables about global climate are known with less than 100 percent
certainty. Let's be wildly and unrealistically optimistic and say that
climate scientists know each variable to 99 percent certainty! (No such
thing, of course). And let's optimistically suppose there are only one-hundred x's, y's, and z's
--- all the variables that can change the climate: like the amount of
cloud cover over Antarctica, the changing ocean currents in the South
Pacific, Mount Helena venting, sun spots, Chinese factories burning
more coal every year, evaporation of ocean water (the biggest
"greenhouse" gas), the wobbles of earth orbit around the sun, and yes,
the multifarious fartings of billions of living creatures on the face of the earth, minus,
of course, all the trillions of plants and algae that gobble up all the
CO2, nitrogen-containing molecules, and sulfur-smelling exhalations
spewed out by all of us animals. Got that? It all goes into our best
math model.

in the best case, the smartest climatologist in the world will know 100
variables, each one to an accuracy of 99 percent. Want to know what the
probability of our spiffiest math model would be, if that perfect world existed?  Have you ever multiplied (99/100) by itself 100 times? According to the
Google calculator, it equals a little more than 36.6 percent.

The Bottom line: our best imaginable model has a total probability of one out of three. How many billions of dollars in Kyoto money are we going to spend on that chance?

Yes, there is a point to be made that climate is really complicated.  However, I can still make correct and valid directional predictions without knowing the exact state of every variable.  For example, I can say with some certainty that, at least here in Arizona, that the temperature at 4PM is going to be higher than the temperature at 4AM, and probably by many degrees.  I can make this statement despite having no idea what the temperature at either time actually is.

I think one can say that the hypothesis is pretty strong that man-made CO2 is increasing atmospheric CO2 concentrations which in turn is causing some warming.  Where Mr. Lewis probably has a point is on the issue of positive and negative feedbacks.  Most of the warming in the estimates in productions like "An Inconvenient Truth"  relies not on just CO2-driven warming, but warming from a variety of feedback processes.  These feedbacks are really really complicated and not well understood.  I discuss this issue of feedbacks both here and here and here.

(HT Maggies Farm)


  1. Matt:

    Yes, you can confidently assert that it will be warmer at 1600 than at 0400 on almost any day, in almost any climate. But without more specific information than that, you don't really have knowledge that's useful in making practical decisions. If you narrow down the range to a particular location during a particular time of year, you have a lot more information. And if you're talking about what the temperature is going to be in Phoenix tomorrow, basing your guess on what the temperature was in Phoenix today is going to be even more accurate than that.

    But climate change scientists aren't trying to predict the weather tomorrow. They're trying to CONTROL the weather, and they're trying to control the weather 100+ YEARS FROM NOW. And in a chaotic system like climate, if you guess any of the initial factors wrong, the error will magnify continuously as time goes on.

  2. Ian Random:

    So let me get this straight, our weather forecasts are accurate to a week. I don't think anyone would trust a forecast beyond that. But if I say that average temperature will go up 1 C in the next 100 years, everyone acts like the sky is falling.

    I. Random

  3. Rob:

    For me the debate has never been about the climate changing and if man is currently causing it warm... it's been about the notion that we want to continually try to control the climate so that it doesn't have swings. Are we going to continuously look for excuses as to how humans are killing the planet everytime there is a swing in the climate. Let's say we reduce the man made gases and the planet continues to warm slowly... will be blame it on too many bodies producing heat, or too many cows farting (the ones we are feeding off of)... so how about we switch to a veggie diet... oops we ate too many plants which is causing the planet to cool down... it seems that the only solution is if the planet had no human beings....

    So I pose this question, what is the purpose of reducing man-made greenhouse gases? Isn't it to somehow control the climate change? Like the 'Matt' mentioned above, the chaotic nature of climate means that one action can exponentially multiple. It's dangerous to try and 'control' the climate. The only reason we aren't in another ice age could be because humans started warming the planet.

    The point is that just because we know (through consensus) that the planet is warming because of mankind, doesn't mean we know (or can predict) what will happen when we start trying to control it.

  4. markm:

    Ian: Very long-term forecasts (e.g., the average temperature over a century) have a possibility of being more accurate because you aren't trying to predict the chaotic events that constitute weather, you're just calculating the energy flows in and out. More energy coming in than going out = it must heat up, although the details of how the heat is spread around aren't predictable.

    The only problem with that theory is that the weather influences the energy flows. Water vapor that hasn't turned into clouds or precipitation yet is a far more powerful greenhouse gas than the CO2, and both evaporation and precipitation are hard to predict. Snow reflects sunlight back into space, cooling the earth. Clouds may either reflect sunlight back into space, or reflect radiated heat back to earth, so they can have a heating or cooling effect depending on just when and where they form and dissipate.

    The predictions of 1 degree or more temperature increase all assume a positive feedback loop, that when more CO2 raises the temperature just a little, it evaporates more water (a reasonable assumption so far), that some of the increased vapor stays in the air to add to the greenhouse effect, and that the net effect of clouds and snow cover doesn't offset the heating effect due to the added water effect. None of these assumptions can be confirmed by modeling at the scale they use. Furthermore, I know of no models that take into account how more CO2, water, and warmth will affect plant growth, which takes CO2 out of the air...

  5. T J Sawyer:

    The biggest problem with climate models comes not from variables known with 99% accuracy, but with relationships known with who knows what accuracy. Continuing the Arizona temperature example: suppose we build a model that embeds this relationship - the 4p.m. temperature equals the 4a.m. temperature times 1.5 Might be a good estimate for those 80 degree mornings in the summer. Turns out not to work so well for Minnesota in the winter when it's minus ten.

    The best thing that could be done for climate models is to document them on the Internet and let the incredibly smart public start picking at them. Bet we would soon lose all faith in climate modeling.

    This is precisely why existing models don't explain the "little ice age" etc.

  6. Greg:

    The fact that 0.99^100 is so small is only important if all 100 must be correct. For example, that's the right calculation for your chance of surviving 100 cliff dives, if each dive has a 1% chance of killing you.

    If it will only take getting 90 of the 100 points right for the models to reach the right conclusion, then your chance of reaching the right conclusion are a lot higher.