Dr. Judith Curry is the President of Climate Forecast Applications Network and was previously Professor and Chair of the School of Earth and Atmospheric Sciences at the Georgia Institute of Technology. She is the author or co-author of nearly 200 published papers. She wrote a lengthy essay about how “confidence” is defined in well known climate reports, available at this link: National Climate Assessment: A crisis of epistemic overconfidence | Climate Etc.
In her extensive blog post she points out inconsistencies in how “Confidence” is applied to discussions of climate science. Major climate reports are known by abbreviations of their title, such as IPCC AR5 for Assessment Report 5 or NCA4 (4th National Climate Assessment).
Conclusions in these reports are given with an associated confidence in their findings. Here are the definitions used in AR4 (predecessor to AR5):
The text uses the wording at left to say something is “Medium confidence”, which sounds like “we are reasonably confident something might occur”. Yet its underlying definition is “About 5 out of 10 chance”, which means there is an equal chance our conclusion is wrong. Which means we have no confidence in our finding yet it has been labeled “medium confidence”.
The report uses text descriptions of “Low confidence” and “Very low confidence” when there is no confidence at all. Consider, a “Less than 1 out of 10 chance” means that we have “Very high confidence” (at least 9 out of 10 chance) that the opposite conclusion is true! Similarly “Low confidence” means we have “High confidence” that the opposite is true.
The bottom 3 elements of this “confidence” table are meaningless – and highly misleading. These “confidence” levels of “medium”, “low” and “very low” mean the author has no confidence in the conclusion and the author should say so.
The NCA4 report uses a similar set of definitions but drops the references to n out of 10:
“Medium” means “suggestive”, “limited consistency”, “incomplete”, “competing schools of thought” (in other words, we have no idea which is correct).
“Low” confidence means no confidence what so ever.
Most people who read a news article citing “high” or “medium” confidence will not know these words have been redefined from common usage – and do not reflect either “high” or “medium” in conventional discourse. Nor will they know that “low” and “medium” actually mean we have no confidence! The actual scale should read “high confidence”, “some confidence” and “no confidence”.
Consider how changing the definitions of common words can be used to convey a propaganda message and persuade you that scientific findings have “medium” confidence, when it means “we have no idea”.
Propagandists know that few people look at the details – yet details matter. When language is twisted as badly as it has been twisted, above, you are looking at a propaganda document, not a scientific document. This particular example is in the climate studies area but could equally appear in other areas. In fact, I have made a list of how language has evolved on a number of topics, but don’t know if I will ever publish it because some of the topics are controversial and people will focus on the controversial aspects rather than the issue of how language is modified to promote propaganda messaging.
This post is not about climate change – this is merely an example of how language is twisted to persuade others to adopt someone’s agenda. This is an excellent and easily understood example. Similar examples can be found in other endeavors such as nutrition, health policy, politics and more.