[Phys-l] Re: "The Truth Wears Off" by Jonah Lehrer in The New Yorker Dec 13, 2010.

[Hugh Haskell <haskellh@verizon.net>]

At 10:03 -0600 01/01/2011, Folkerts, Timothy J wrote:
> I wasn't referring to any specific result -- I 
> was considering the general human nature.
>
> But here is an example -- consider the "hockey 
> stick" temperature reconstructions.  There has 
> been some discussion of the statistical details 
> used to reconstruct past temperatures from the 
> proxy data.
>
> * If Michael Mann is predisposed TOWARD AGW and 
> he finds support that now it is hotter than any 
> time in the past two millennia, he is likely to 
> accept the result.  And the IPCC is like to to 
> accept the results.
>
> * If Stephen McIntyre is predisposed AGAINST 
> AGW, he is likely to say to himself "maybe Mann 
> did the analysis wrong" and try a second time. 
> Now when his statistical analysis shows that the 
> hockey stick disappears, he is likely to accept 
> that result.
I agree that this is worth considering, but I 
think it has been done. The various temperature 
data have been analyzed many times, and the 
results have always been the same, even when 
different data sets are considered. As long as 
one get a true average temperature for the world 
(that is, not biased to urban heat islands, or 
the like), and the temperature scales are 
calibrated to the same standards, the trends 
always show an upward trend over sufficiently 
long periods, since the 1850s. There are several 
leveling off periods and a couple of small 
declines, that can be explained by such things as 
soot from WWII, or volcanic eruptions, and the 
like, but the trend is consistently up, and at a 
much greater rate than those upward trends that 
happened in the pre-industrial  or geologic ages.

It is the speed with which the modern change is 
happening that is the primary cause for alarm, 
and enables a reasonably certain attribution of 
the data to human activity. This is 
unprecedented. In all of the discussions of this 
I have read, those who claim contrary results 
have been shown to either be looking at 
uncalibrated data, limited data sets (such as 
using US-only temperature records), or 
"cherry-picking" the data they are 
examining--that is looking at the data for the 
previous 10-15 years, dominated by the 
extraordinary peak temperature due to a very 
strong El Niño in 1998, and concluding that the 
temperature rise has leveled off. But even in the 
absence of the 1998 El Niño, a 10 year period is 
simply insufficient to base a trend on, and 
extending that data back to 1900 or so, shows 
that the trend continues upward, and the current 
shallowing of the trend slope is probably 
temporary. It would be nice if this was in fact 
the beginning of a level period, but we can never 
be sure that the rise won't recommence at most 
any time.

The upshot is that we are seeing the effects of 
global warming in several independent lines of 
investigation, from the northward migration of 
climate zones in the northern hemisphere (and 
conversely in the southern hemisphere), to the 
accelerating ice melting in the arctic ocean and 
the Greenland ice cap), to the northern migration 
of tropical diseases, the mountain top isolation 
of several species that are being dramatically 
reduced by heating of their mountain top 
habitats, the increased acidification of the 
ocean and resulting death of many coral beds, 
among others.

I just don't think that the global warming issue 
falls within Lehrer's sphere of concern. The 
effect does not seem to be losing force. I would 
like to see some more rigorous explanation of the 
effect he reports on. I don't find the 
statistical explanations he ends with very 
satisfying. Perhaps its just because he doesn't 
explain the statistical issues very clearly, but 
I'd like to hear more about why a strict 
double-blind test isn't good enough to present 
the effect from happening if it really is a 
statistical phenomenon. I think we need to know 
more about how samples populations are chosen for 
these tests, and especially how they are chosen 
for follow-on replication tests. Maybe the 
sampling isn't as good as it should be; maybe the 
tests aren't long enough; maybe there is 
inadvertent or intentional cheating going on; or 
maybe it's just because the tests to that show no 
particular advantage or disadvantage of the new 
product usually never get published, so we are 
biasing the data from the beginning.

In any event, I don't know of the selective 
sampling problem in the physical sciences, 
primarily, because a repetition of a test is 
often done to improve the accuracy of the 
measurement, so there is little reason not to 
publish a result that gives the same result but 
to a higher degree of accuracy, or refutes the 
original finding.

It does occur to me that something like the 
phenomenon Lehrer reports may have been present 
in the "fifth force" and "cold fusion" events. In 
both those cases, the initial reports were 
followed by confirming follow-ups but then 
quickly, the disconfirming reports started to 
come in, and the effect faded rather 
quickly--within a few months. In fact, it was 
only about a year after the initial report of the 
fifth force was made that those authors reported 
that the effect they "found" was an artifact of 
their data-taking methods (if I recall 
correctly). I have always thought of these two 
events as shining examples of how and how not to 
do and report scientific results, with the good 
example being the fifth force and the bad one 
cold fusion. Even so, it appears that even these 
events had a "lifetime" of only months, compared 
to the several years Lehrer reports for the 
effects he talks about.

On the other hand, the discovery of "high 
temperature superconductivity" had a similar but 
happier early lifetime. The initial results were 
not only immediately confirmed but led quickly to 
significant advances in superconductor technology.

It might be worth the effort to revisit the 
"polywater" fiasco from the late 1960s, as 
another example of the phenomenon Lehrer talks 
about.

I know that Ludwik remains interested in the 
intellectual follow-on to the cold fusion 
phenomenon, and still hopes that something will 
come of it. I have no definitive opinion on the 
present status of that work, but I rather think 
that it remains in the realm of "fringe science" 
until some more definitive results can be 
presented.

I guess that, in general, I've talked myself into 
believing that Lehrer's work is mostly a 
phenomenon of the life sciences, where 
statistical sampling is the lifeblood of the 
scientific methodology. Although I'l like to hear 
if anyone else has some thoughts about its effect 
in the physical sciences.

Hugh
--

Hugh Haskell
mailto:hugh@ieer.org
mailto:haskellh@verizon.net

It isn't easy being green.

		--Kermit Lagrenouille