A recent article in Wired–BS Detector—described how DARPA is seeking proposals for “ways to determine what findings from the social and behavioral sciences are …credible.” The request for proposal states, “There may be new ways to create automated or semi-automated capabilities to rapidly, accurately, and dynamically assign Confidence Levels to specific … results or claims. … Help experts and non-experts separate scientific wheat from wrongheaded chaff using “machine reading, natural language processing, automated meta-analyses, statistics-checking algorithms, sentiment analytics, crowdsourcing tools, data sharing and archiving platforms, network analytics, etc.” Boiling this down to its essence, the article’s author suggests that DARPA is “trying to build a real, live, bullshit detector.”
It might seem that this is a far out notion except for two things. First DARPA has a solid reputation for innovation and solving the unsolvable. Its work on communications is now what we call the internet. Second, Artificial Intelligence technology is already being in business, especially in investing, where it is achieving above average returns, and in the developmental work on autonomous cars.
Machine learning is a more generalized application of artificial intelligence and may be the means by which DARPA achieves the ability to assign confidence levels to scientific results and claims. When that day arrives, hopefully not that far in the future, the proponents of an impending climate catastrophe will be in serious trouble.
If there was a working BS Detector a recent article in the Washington Post titled “We only have a 5 percent chance of avoiding ‘dangerous’ global warming might not have been published. Written by avowed climate activist Chris Mooney, the article is based on two peer reviewed articles published in Nature Climate Change. One analysis is based on the hypothesis that additional warming is embedded in the climate system through an energy imbalance from past emissions but has ”not yet arrived.” The other analysis applies past statistics to the Kaya Identity model to estimate emissions and warming in 2100. The Kaya Identity is a simple way to project future carbon emissions based on changes in population, GDP per capita, energy intensity, and carbon intensity. The problems with this article are so obvious that Nature should have rejected publishing it. The Kaya Identity model can be grossly right for near term periods but is useless for making estimates decades into the future. No one can accurately estimate carbon intensity or energy efficiency 83 years into the future. It is foolish to try. In addition, this article implies an unjustified accuracy in climate sensitivity—the warming from doubling CO2. The IPCC estimate has changed over time with the lower bound being reduced. The current estimate varies by a factor of 3.
The media assigns too much credibility to peer review. The peer review process has been misused in all fields of science and its shortcomings well documented. In part, peer review fails because reviewers are not blinded, are too busy to invest enough time to do a thorough review, and most important the essential steps of data validation and replication are not performed.
If DARPA’s initiative is successful climate catastrophism will be exposed for what it really is, advocacy. Fact Checker gives Pinocchios; maybe the BS Detector will assign piles of dung.