27. An expertÕs estimates
which are not based on strict calculations
cannot serve as a measure of real probability
Unlike in the stock markets, where the
average estimate of the best experts is used as a forecast of market behavior,
we cannot select our experts and average them based on their track record of
predicting human extinction, because there is no track record of such an event.
If it had happened, we would all be dead, and quite incapable of predicting
anything.
Slovic, Fischhoff,
and Lichtenstein (1982, 472)21, as cited in
Yudkowsky (2008)22 observed:
A
particularly pernicious aspect of heuristics is that people typically have
great confidence
in judgments based upon them. In another followup to
the study on causes of death, people were asked to indicate the odds that they
were correct in choosing the more frequent of two lethal events (Fischoff, Slovic, and
Lichtenstein, 1977) In Experiment 1, subjects were reasonably well calibrated
when they gave odds of 1:1, 1.5:1, 2:1, and 3:1. That is, their percentage of
correct answers was close to the appropriate percentage correct, given those
odds. However, as odds increased from 3:1 to 100:1, there was little or no
increase in accuracy. Only 73% of the answers assigned odds of 100:1 were
correct (instead of 99.1%). Accuracy ÒjumpedÓ to 81% at 1000:1 and to 87% at
10,000:1. For answers assigned odds of 1,000,000:1 or greater, accuracy was
90%; the appropriate degree of confidence would have been odds of 9:1. . . . In summary, subjects were frequently wrong at
even the highest odds levels. Moreover, they gave many extreme odds responses.
More than half of their judgments were greater than 50:1. Almost one-fourth were greater than 100:1. 30% of the respondents in
Experiment 1 gave odds greater than 50:1 to the incorrect assertion that
homicides are more frequent than suicides.
The point of this quote is to
illustrate that experts are consistently overconfident, often ridiculously so.
From Parkin's Management Decisions for Engineers23:
Generally, people have a displaced
confidence in their judgment. When asked general knowledge or probability
questions, experimental subjects performed worse than they thought they had (Slovic et al., 1982). Calibration experiments that test the
match between confidence and accuracy of judgment, demonstrate that those
without training and feedback perform badly. Lichtenstein et al. (1982) found
that from 15,000 judgments, when subjects were 98% sure that an interval
contained the right answer they were wrong 32% of the time. Even experts are
prone to some overconfidence. Hynes and Vanmarke
(1976) asked seven geotechnical gurus to estimate the height of a trial
embankment (and their 50% confidence limits), that
would cause a slip fracture in the clay bed. Two overestimated the height and
five underestimated. None of them got it within their 50% confidence limits.
The point estimates were not grossly wrong but all the experts underestimated
the potential for error.
Simply put, ÒexpertsÓ are often wrong.
Sometimes their performance is equal to random chance, or to that of a person
pulled off the street. Statistical prediction rules often outperform experts24.
This creates trouble for us when we rely on experts to evaluate the probability
and nature of catastrophic global risks.