mead cohen berger shevtsova garfinkle michta grygiel blankenhorn
Predictive Problems
How Climate Models Warm the Planet
Features Icon
Features
show comments
  • mc

    After all the scandals, failures, and overselling climate modelers have subjected us to, one has to suspect that the current generation perceives a threat. It may be they’re not just afraid that a massive increase in computing power will produce the wrong answers; they might well fear a challenge to their priestly status as custodians of the temple of massive data processing. Until now they’ve been simultaneously wrong on many things and, as long as they controlled the machines, unassailable. This is obviously unsustainable, unless an artificial obstacle is erected to progress in the field. They won’t exit the stage as gently as their teachers did, in part because they know not to expect the kind of mercy they refused to grant others.

  • ljgude

    The reason double blind studies are required to prove the effectiveness of a new drug is because the researchers can unconsciously bias the results. Computer models encode the expectations and values of the modelers even if they are trying their best to be as objective as possible. Given such incidents as a couple of tranches of Cliamategate emails there is evidence that the researchers may not be unconscious when they ‘hide the decline’. But leave that aside, computer modeling is a form of high class speculation and the apparent precision of numbers gives a false aura of credibility and the Mana this output gains from employing the magical technology or our age (It just works, don’t you know) is every bit as effective as feathers and smoke in the hands of a competent medicine person. Look at the 4 or 5 possible hurricane tracks generated by computer models to see how loose the connection between computer modeling and reality is – even when trying to predict events a week in advance. There is a place for informed speculation in science, but it shouldn’t be confused with actual, much less, settled science. Putting numbers through a computer does not transubstantiate them into science.

  • Pait

    I think you did not understand the engineering compromises involved in building high performance computers.

    Additionally, the fact that high speed computers used in climate modeling use a lot of energy is ironic, as mentioned in the NYTimes article; however the amount if energy used in scientific computation is not conceivably of an order of magnitude that would impact global climate, as you imply in your piece.

  • Fat_Man

    The following is actually a comment I made on this site on Mach 24 of this year. I repeat it because it is equally apt here.

    The output of climate models is at best just fun with numbers. It is only to be used for entertainment purposes.

    If you want to understand how climate models are built and why they incapable of producing meaningful forecasts, you should spend a few
    minutes listening to this lecture by a professor of Applied Mathematics:

    “Believing in Six Impossible Things Before Breakfast, and Climate Models” A talk by Dr Christopher Essex – Professor and Associate Chair, Department of Applied Mathematics, University of Western Ontario, Canada. Published on Feb 17, 2015.

© The American Interest LLC 2005-2016 About Us Masthead Submissions Advertise Customer Service