Daubert and the Admissibility of Climate Models as Evidence in a Court of Law

February 29th, 2012 by Roy W. Spencer, Ph. D.

I get involved in so many things, sometimes I forget to let people know when something significant happens.

A few months ago, a paper I co-authored with attorney Brooks Harlow appeared in Energy Law Journal entitled: An Inconvenient Burden of Proof? CO2 Nuisance Plaintiffs Will Face Challenges in Meeting the Daubert Standard

I had always wondered how climate models, which are the ultimate source of “evidence” of anthropogenic global warming, could meet the rather stringent guidelines that judges use when deciding whether expert testimony should be allowed in court. Then, one day I received a phone call from an attorney (Brooks Harlow) who had been wondering the same thing, and we agreed to collaborate on a paper for submission to a law journal.

According to Wikipedia, “The Daubert standard is a rule of evidence regarding the admissibility of expert witnesses’ testimony during United States federal legal proceedings.

While the Supreme Court’s decision in Connecticut v. AEP effectively shut the federal court doors to federal common law nuisance claims, about half of the states rely on Daubert, and CO2 litigation can be expected in the coming years in various states.

There are 5 relevant standards in the Daubert test of admissibility, but which the judge does not necessarily have to follow rigorously. Again from Wikipedia:

  • 1. Empirical testing: the theory or technique must be falsifiable, refutable, and testable.
  • 2. Subjected to peer review and publication.
  • 3. Known or potential error rate.
  • 4. The existence and maintenance of standards and controls concerning its operation.
  • 5. Degree to which the theory and technique is generally accepted by a relevant scientific community.
  • In my opinion, two of the standards above (#1 and #3) are not met by climate models. In standard #1, climate models are not testable in any rigorous sense in what they are being used for: prediction of future climate change. We will not know whether they are right or wrong until the future happens.

    And even then, we can’t be sure that warming which ends up being observed in nature occurred for the same reason the models predicted it. Just because they create average weather that is somewhat realistic does not mean they can tell us how average weather might change in the future.

    And, contrary to what you might have heard, there is no obvious “fingerprint” of human warming. Natural warming from a slight change in cloud cover would look the same.

    Also, the oft-mentioned missing “hot spot” of warming in the upper troposphere would be the theoretically expected result of ANY warming; its presence (or absence) does not demonstrate causation as to the source of the warming.

    After all, the twenty-something climate models are all over the map in their predictions, and if one or two happen to be correct, what about all the other climate models which are also based upon “physical principles”, but which were wrong in their predictions? At some point, we have to admit that given enough different model predictions, one or two models are bound to be close to correct just by chance.

    Standard #3, which I also believe is not met, is related to standard #1 just discussed: the methodology should have a known error rate. In other words, how likely is a climate model to be correct based upon its success in previous predictions?

    “Predictions” of the past (e.g. what happened in the 20th Century) don’t really count because those are more exercises in curve fitting than prediction. What if the modelers did not know what the 20th Century temperature variations looked like, and were asked to use models to “hindcast” what happened. Would they be able to? Not likely.

    And since future warming (if it occurs) is a one-of-a-kind event, we won’t know for many years whether any of the climate models are correct even once, let alone for a statistical sample of independent predictions.

    If the climate system is always changing anyway, as some of us believe, you have a 50% chance of being right in a prediction of a warming trend by just flipping a coin.

    While some might also argue whether one or more of the other Daubert standards are not met, I thought I would address the two most obvious ones.

    I yearn for the day when the “scientific consensus” defenders are cross-examined in court. No longer will they be allowed to get away with demanding they be believed just because they are experts. That kind of attitude does not get you very far in court.

    Comments are closed.