From the “anybody can make a hockey stick out of random numbers department” and Twitter comes this interesting exercise. Steve McIntyre and Ross McKittrick demonstrated this two decades ago. Of course Mann still insists he’s right, mainly because his super-sized ego won’t allow him to admit his errors.
@andy on X writes:
Did my own try on the Mann/Marcott/PAGES2K proxy screening routine and created an ensemble of 103 pseudoproxies consisting of random numbers between -2 and +2 for the time 0-2025 AD. To find the “temperature sensitive” proxies (as prescribed for this process) I tested correlation of the random proxies with the NOAA global temperature data set 1850-2025. The chart below shows what I received as average after discarding the negatively and non-correlating proxies. I created global warming from a bunch of meaningless casino numbers, pure noise.
The average of the whole ensemble of pseudoproxies shows no real trend, of course.
Simply flipping (changing the sign) of the pseudoproxies that have the worst correlation has a similar effect (chart below). That is also allowed in their view, since correlation is correlation and can be used, be it negative or positive.
Steve McIntyre weighs in:
Related
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.