I am happy to have been granted permission by Russell Turpin to post
this very useful article. Included are comments on the original article
by Eric Pepke and Ken Arromdee, also posted with permission.
Listening to the frequent discussions over controversial empirical claims,
an unsophisticated reader could easily walk away with the view that only
tradition and prejudice separate the sparring factions. Such a reader might
think that most scientists cast a skeptical eye on paranormal phenomena,
the claims for homeopathic dilution, the idea that the earth is relatively
young, etc., merely because these scientists were taught opposing claims.
As one poster's signature would have it, such critics merely engage in "school
of thought bashing."
I think this view is wrong. I think it stems, in part, from an inadequate
understanding of how to evaluate evidence. The evidential claims for many
of these controversial notions exhibit common flaws. They are the kinds
of flaws that scientists recognize from many, many past failures. It is
this history of dead ends which seduced previous researchers with flawed
evidence that informs the way scientists evaluate the evidential claims
accompanying these controversial notions.
In this article, I will first list some of these evidential flaws and
then discuss errors in relating evidence to theory. Of necessity, this is
a short list that omits most such problems. It is largely biased by what
I have seen in newsgroup discussions. (A true survey would require a book,
of the order that David Fischer wrote for historians.) Finally, I will discuss
when mere mistakes (which plague every research direction) turn
In the foreground of such controversies are the various studies and experiments
published in journals or elsewhere. Various professional posters in the
science newsgroups often complain about readers who read all such studies
and experiments as if they were the same. The problems listed below are
a small sampling of the kinds of issues that the critical eye brings to
the reading of these studies and experiments. (I purposely omit particular
issues of experimental design and statistical analysis.)
SUBJECTIVE MEASUREMENT. There are unfortunately times when a study
or experiment must rely on the measurement of very subjective experience:
whether a patient feels better or worse, whether two drawings are similar,
etc. This element of subjectivity is notorious for introducing unintended
and subtle errors into the result. Studies that eliminate this element as
much as possible put the result on firmer ground. Thus, it is better to
measure the effect of a medicine through chemical or physical analysis or
other objectively measured symptom than through patient report, it is better
to compare discrete matches rather than drawings, and it is better to count
light flashes with a photodetector than with one's eyes.
SMALL DIFFERENCES. Studies and experiments that show a small difference
between the test and the control when the test result falls within what
well-established theory would predict are somewhat suspicious. This kind
of result begs for different experimental design, tighter controls, or investigation
of other possible causes.
TIGHTER CONTROLS TURN POSITIVE RESULTS NEGATIVE. If tightening
the controls in an experiment turns a positive result into a negative one,
this is virtually the death knell for the alleged phenomenon. Almost always,
this shows that the positive results stemmed from a phenomenon other than
the one the experiment is designed to detect. Future positive results are
viewed suspiciously unless a good explanation for this history is forthcoming.
CONTINUING NEGATIVE RESULTS. Negative results count more against
a claim than positive results count for it. This is especially true if negative
results continue over time as the alleged phenomenon is studied, even if
they are few in number compared to the positive results. The reason is simple.
If the phenomenon is real, those studying it should eventually reach the
point where they can reliably demonstrate it and where they can teach others
how to reliably demonstrate it.
It often takes a knowledge of the field of concern to evaluate these
issues. The history of forward steps, set-backs, or stagnation set a context
that underlies how a new study is received. This context usually is not
explicit in the article or report on the study.
The flaws above concern a particular phenomenon that is alleged to occur
and the experiments to evince it. The step from evinced pheonomena to theory
is also plagued by potential error.
NO DIRECT EVIDENCE. Perhaps the most severe flaw of an empirical
theory is that all evidence for it is very indirect. Sometimes this cannot
be helped. For example, all historical theories suffer this flaw, since
the past can only be observed through its effects on the present. (This
makes the study of history particularly challenging.) But theories of current
phenomena should admit fairly direct testing. For example, if the flow of
qi energy through the body and the existence of molecular patterns from
homeopathic dilution are true theories, those who study these things should
be able to find experiments that fairly directly measure qi and these molecular
NO DEEPENING EVIDENCE. Similarly, theoretical knowledge should
grow and become more detailed as experience increases. In the 1960s, molecular
biologists could only mouth vague claims about DNA guiding the development
of organisms. Now they can tell how this happens in more detail, and back
this discussion by (tens of?) thousands of experiments that evince these
details. Two centuries ago, Lavoisier described how oxygen combines with
other elements to release energy. Our knowledge of chemical reaction has
increased tremendously since then. But what has happened to the theoretical
underpinnings of homeopathic dilution in two centuries? Why does it remain
vague mouthings about "molecular patterns"?
PREDICTED PHENOMENA REMAINS SLIPPERY. As experimental and theoretical
work progresses, more evidence and more sound evidence for the
related phenomena should appear. If the phenomena predicted by a theory
remain plagued by evidential flaws as research progresses, then the theory
itself becomes very suspect.
POOR INVESTIGATION OF ALTERNATIVE EXPLANATIONS. Often the results
claimed for a novel theory are potentially explained by well-founded theories.
These alternative explanations need to be investigated, and such paths barred
by better controls in future experiments.
REVOLUTION WITHOUT SUPPORT. A theory becomes especially suspicious
when, in addition to suffering the above flaws, it directly conflicts with
a theory that measures well by the same criteria. Using again the homeopathic
theory of dilution as an example, if it is true, it will cause a revolution
in chemistry and biology that makes cold fusion look like small potatoes.
But its evidence remains far too indirect, too shallow, and too slippery
to succeed at such a revolution, despite two centuries of research in it.
All the problems above occur within conventional theoretical and experimental
investigation. Whether and how they are resolved help determine which theories
are accepted and which are rejected. Scientists live on the tension between
two poles. Driving them to the exotic is their eagerness to discover new
and revolutionary facts. Warning them away from quackery is a skeptical
eye informed by knowledge of the myriad errors that have misled others in
the past. Scientists looked at N-rays, slippery water, and cold fusion because
of the exciting potential to discover something new. They turned away from
these things because the evidence did not pan out. John A. Wheeler invited
parapsychologists into the AAAS because he thought there was beginning to
be some real science in what they did. Ten years later, he knew this had
been a mistake.
The attraction of the new and exotic is very strong, and its lure is
so bright that it sometimes causes people to lose their critical sense.
And some people, unfortunately, never develop a critical sense. Those who
have lost or never developed a critical sense create and join "schools"
where quackery is born from weak theories and mistaken notions becoming
instutionalized. These "schools" are full of the kinds of rationalizations
that people use to justify their views when nothing else is available. There
are far too many of these to list, but some of the more colorful signposts
are listed below.
"PARADIGM" TALK. "Paradigm" is perhaps the
most abused word in these discussions. Whenever a proponent of a controversial
empirical claim counters criticisms of the evidence by reference to a "paradigm
shift," it is time to put on one's hip-waders. To the extent that "paradigm"
just means a new theoretical view, it prevails because of -- not
in spite of -- sound evidence. The rise of quantum mechanics is frequently
referenced as the paradigmatic example of paradigm shift. But the discovers
of quantum mechanics did not have to philosophically argue their opponents
into making a paradigm shift before quantum phenomena were accepted. The
proponents merely presented ever increasing amounts of solid evidence.
To the extent that "paradigm shift" is used to describe something
about the social and historical process of how research is done, it has
little legitimate role in discussions of evidential quality. Most other
uses are so vague that no significant meaning can be attached.
THE WORD "SCIENCE" USED NARROWLY. A quack will often
reply that his ideas have evidence, just not the kind accepted by "science."
The problem with this is that science is no more and no less than sum total
of what we have learned about evaluating general empirical claims and their
evidence. (Its application to modern research and the need for a new word
such as "science" is merely because so much progress in this area
has been made in the last three centuries.) With regard to general empirical
claims, asserting that there is no scientific evidence is the same
as asserting that there is no good evidence. Quacks want to find
some room in between, but they cannot explain why we should accept the kind
of evidence in their case that has proven so bad in other cases. In essence,
they engage in a kind of special pleading that hangs on attaching some odd
meaning to the word "science".
"SCIENTIFIC PARADIGM." This phrase has almost no useful
meaning. (Peter Kaminski take note!) If it is used by someone defending
a controversial empirical claim, it is virtually guaranteed that the argument
MISCHARACTERIZATION OF THE STATE OF THE ART. Quack theorists often
distort the rest of science is in order to make their favored notions seem
more equal in comparison. Thus, "conventional" physics is sometimes
accused of ignoring the observer. (Hah!) "Allopathic" medicine
is sometimes described as based on non-holistic principles, as practicing
the notion of "one symptom, one diagnosis, one cure," etc. ad
nauseum. This is all bullshit.
"QUANTUM." Unless the writer is referring to physics
or chemistry, the use of phrases such as quantum, the uncertainty principle,
entropy, etc., are warning signs. If they are combined with other words
in novel ways -- e.g.: "quantum psychology," "democratic
entropy," etc. -- it is an almost sure sign of bullshit. (For Jeremy
Rifkin, the rule is reversed. His writings about entropy are bullshit especially
when he discusses physics and chemistry.)
CARTS BEFORE HORSES. Proponents of quack theories are full of
excuses for why they have such meagre evidence of their beliefs. These range
from "no one funds us" to "the conspiratorial and established
institutions ignore us for political reasons." These excuses would
not be needed if there were good evidence for the notions in question. The
fact that these excuses are offered is almost an admission that the proponent
believes despite a lack of good evidence. It it were otherwise,
the proponent would focus on the evidence and argue for funding
or institutional change because the evidence is so good, rather than excusing
the lack of evidence because of these other factors.
"MILLIONS OF CHINESE CANNOT BE WRONG." This excuse usually
comes in the defense of notions resurrected from older traditions, e.g.,
traditional Chinese medicine. In some sense, it falls under the "big
lie" tradition. In a few minutes, someone with a modicum of historical
knowledge should be able to think of several cases where millions of Chinese
(or Amerindians or ancient Hellenes or ...) and millenia of experience were
wrong. The fact is that we have learned a lot about how to perform and evaluate
empirical research in the last three centuries and that this gives us a
significant advantage over previous traditions. (One of the curious things
about the resurrection of older traditions is that foreign traditions are
more interesting that native ones. Thus, one hears arguments for qi and
traditional Chinese remedies, but almost never for the four humour theory
of disease and the frequent bloodletting and purges it prescribes.)
Once a "school" has developed around poor theories, it essentially
halts all useful progress by its practitioners until the "school"
is reintegrated with the larger scientific community. The institutionalization
of theories in an uncritical atmosphere and away from the larger scientific
community almost guarantees that there will be a continuing sequence of
"positive" results, sometimes for centuries, even though the phenomena
remain slippery, understanding remains vague, and discovery of new knowledge
is left to the rest of science. In short, a duck is born. Quack, quack.
That's an excellent summary! Here are a few thoughts I had while reading
it. Some of them overlap with things you have said, especially the first
one, which overlaps several of your categories, and the second, which overlaps
REVOLUTION WITHOUT SUPPORT.
MARGINAL RESULTS. When faced with marginal results, scientists
will attempt to refine or replicate the experiments until stronger and more
consistent results are found. When a researcher spends an inordinately large
amount of time interpreting and reinterpreting old data, or new data from
the same experimental setup, and relatively little time attempting to get
better data, the results are suspect.
MISESTIMATION OF EFFECTS. Quack researchers frequently misestimate
the effects their discoveries will have. While they may speak about grandiose
social effects, they frequently underestimate the scientific effects. One
example is homeopathy, which would cause a revolution in chemistry if true.
Yet the supporters seldom grapple with the idea of these effects. Another
example is the frequent claims for a carburetor or other gizmo which will
make an automobile get an incredible number of miles per gallon. Simple
calculations reveal that the engine needs to operate at higher than Carnot
efficiency. Personally, if I knew a way to run a heat engine at higher than
Carnot efficiency and thus ignore the 2nd law of thermodynamics, I would
have better things to do than waste my time building a carburetor factory.
SCIENCE AS INSTITUTION. Philosophers, psychologists, and anthropologists,
when they deal with science, currently view it as "that which scientists
do." Although this definition is possibly useful for what they are
trying to study, when it is used as the meaning of "scientific"
in "scientific evidence," trouble starts. The conflation of meanings
leads to the notion that all those things which any scientist does are valid
science. This results into a combination of appeals to authority and ad
hominem attacks which are wrongly presented as scientific inquiry.
ANALOGOUS THEORIES. Many scientific theories begin as analogies
to existing well established theories or as attempts to apply the results
of a field of study laterally to something new. Although this sometimes
produces theories which hold up well on their own, it frequently gives undeserved
credence to the new theories. Well established theories generally apply
to a specific well-defined set of phenomena, and the support for the theory
exists within that context. The analogy or lateral application discards
the context entirely. The result is a sort of informal belief that the new
theory is well supported, when there may be no reason to believe that the
two situations have anything to do with each other. An example of this is
Social Darwinism, whereby evolution by natural selection of organisms is
assumed to work as well to social institutions.
DEFENSIVENESS. It is a common human tendency to take criticism
of one's work personally and respond devensively. Scientists must constantly
be aware of this tendency and suppress it, because unchecked defensiveness
is the death of scientific inquiry. When a researcher consistently interprets
criticism of his or her theories, hypotheses, or data as personal insults,
they become suspect. The researcher falls into the trap of considering it
a personal conflict and naturally resists the kind of criticism that is
absolutely neccessary to test hypotheses. The first strong indication that
I had of the problems with cold fusion, back when it still seemed plausible
and exciting and everyone was trading speculations about mechanisms, was
a letter by one of F&P [Fleischmann and Pons -- whj] accusing all of their
critics as attacking them personally.
I'd like to add something else, mostly because I ran across it yet again.
"IT WAS ONLY TO GET YOU TO THINK" One common tactic
of crackpots is to dismiss disproofs of their claims with the excuse that
the claim was not intended seriously, but was meant only to get their opponents
to think, to argue properly, or some similar meta-reason. Until the crackpot
gives this excuse, it is not possible to distinguish between his serious
claims and his non-serious ones. Furthermore, the crackpot's claim may contain
factual errors, or sufficiently elementary logical errors, which are too
simple to be useful for encouraging thought,
Several possiblities suggest themselves, none of which indicates worthiness
of the crackpot's ideas.
One possibility is that the crackpot is working backwards from his conclusion.
If he does not work far enough backwards, he will come up with problematic
"support" for his claim; since he does not really believe
the result because of the support, but rather believes the support because
of the result, he uses this excuse to dismiss the problems. In his own mind,
the support is not evidence, but only a means to convince others of what
he already knows, so he doesn't consider this unfair.
Another possibility is that the crackpot's true claim is somewhat broader
than apparent at first glance. Talk of paradigms, comparisons to Galileo,
etc. may suggest a general dislike of the scientific method and of what
the crackpot considers the scientific establishment. When the crackpot disputes
some well-known scientific result, he mainly desires not just to disprove
that result, but to take scientists in general down a peg. He argues many
nonscientific positions not because he strongly believes particular ones,
but rather because he holds an anti-science meta-position; to him, his argument
is about scientists' ability to determine truth, not about specific truths.
Return to Bill Jefferys' Goodies Page