I want to take up the topic I introduced in the last posting,
the ability to forecast social/political developments. My focus was the role that pundits play in
providing views of the future, an ability in which they have proven to be less
than proficient. My description of this
dynamic is based on the thoughts of Philip Tetlock,[1] a Canadian-American
professor and political writer. I tried
to stress the importance of pundits in our political culture in that they
provide the rationales for opinion-leaders among the general populous. Tetlock provides us with certain ideas about
how we, and more specifically pundits, can improve on this business of
forecasting.
To review his findings regarding pundits, he presents three
conclusions: their rate of success at
predicting is a little better than chance; despite this deficiency, they are
overly confident in their abilities to forecast and seldom do they express any
change of mind or heart when confronted with evidence that they were
mistaken. Central to this back and forth
system between pundits and their audience is the fact that the role pundits
play is not to inform the audience but to reinforce already held beliefs. My feeling is that what the pundits provide
is not so much information or expert opinion, but phraseology and cherry-picked
facts that support what has been established in their minds as being true,
good, or bad. Of course, reality about
the social and political world does not help.
After all, that’s what the forecasting business is in business to
predict, and one can very reasonably ask whether or not that world is
predictable to begin with.
Tetlock uses the notion of Nassim Talib of the “black swan”
approach to history. First, most pundits
on TV and columnists in the papers rely on history to provide the evidence they
use to justify their conclusions. Only
rarely do they cite social science literature for backing; instead, they recall
some historical incident or development that they present as analogous to the
conditions they are analyzing. I have
already in this blog voiced my bias for historical research over social science
research, but my bias is based on going beyond citing a case or two – a topic
for another posting. In terms of our
pundits, it is beneficial to adopt this black swan view. It encourages us to view history as a record
of social relations being in an equilibrium for a time and then being hit by
some highly disruptive event or development – the black swan. Affected institutions take into account the
swan and eventually create a new equilibrium.
Is this a way to view Donald Trump, for example, in our current
presidential primary season?
Possibly. But the Trump case
illustrates what this phenomenon is like; no pundit forecasted the Trump factor
in this political season. And yet, what
will be the consequences of Trump; what new equilibrium point will this
candidate help establish? Time will
tell, but I can assure you he will cause changes in the Republican Party.
One consequence of this process that might be overlooked is
that during periods of equilibrium, we seem to accept a new set of conditions
as “normal” fairly quickly. We,
collectively, seem to have short memories.
The new resulting social environment lulls us into a sense that we can
predict what’s coming. Patterns
establish quickly and within those patterns, the more informed and savvy
observers can up their accuracy rates of telling us what is going to
happen. But it is not whether a new
black swan will appear again, but when.
And as it appears, there goes all that ability to see what’s
coming. So the second consequence is
that our ongoing attempts at predicting do have a pattern: successful periods followed by almost
complete disarray primarily due to our undeserved hubris in feeling we know
what’s up until the arrival of a newer black swan.
So, one lesson is to be wary of any prediction that is stated
in longer-term time frames. Short-term
forecasting is what pundits should be about and even there the listener should
be leery. Or stated another way, view
this limited skill as being akin to the skills of a good poker player or hedge
fund manager. Seeing politics as poker
has advantages. For one, a “player”
defines the possible events as a range of possibilities. One can, under such thinking, view potential
acts as a sampling universe, not as well defined as with a deck of cards, but
more defined than the options entertained by the typical pundit one sees on TV. This view of possibilities is a landscape of a
type in which one’s preconceived assumptions are tested and worked out. The goal under such an approach is to
formulate reasonable probability estimates as a poker player visualizes
possible poker hands, and as with poker and an accompanying good sense of learning
from costly mistakes, this discipline provides one an avenue for
improvement. One can become better at
forecasting, although Tetlock warns it does take time – beware of the young
pundit.
As for what comprises the “deck” in politics, Tetlock
provides another analogy: that of a cloud – as opposed to a clock. The forecaster needs to disabuse him/herself
of any notion that politics resembles a clock with regular movements. The sense of a cloud, as it applies to
politics, is seen as this conglomeration of events Tetlock describes as follows:
If world politics is more cloudlike –
little wisps of clouds blowing around in the air in quasi-random ways – no matter
how theoretically prepared the observer is, the observer is not going to be
able to predict very well. … [T]he
cloudlike view would posit that the optimal forecasting frontier is not going
to be appreciably greater than chance or you’re not going to be able to do much
better than a dart-throwing chimpanzee.
One of the things we discovered in the [simulated studies] work was that
forecasters who suspected that politics was more cloudlike were actually more
accurate in predicting long-term futures than forecasters who believed that it
was more clocklike.[2]
Modesty is a factor in proficiency levels when it comes to
forecasting; that is, modesty is positively related to the ability to predict
long-term political developments.
Does that mean heuristics are useless? Not at all; the point is that when given the
time one should put in the work; figure the probabilities as best one can and
calculate what is likely to happen.
Heuristic hunches by the novice or uninformed are more in line with the
accuracy of the chimpanzee referred to above.
Heuristic hunches by ideologues are just a bit better. But heuristic hunches by those who have spent
a professional lifetime studying a field of knowledge prove to be fairly
accurate. But even they, if given the
time, do not rely on hunches; they put in the time. And the time among political experts seems to
be better spent in not looking at their field as this Newtonian world of
predictable phenomena, but as those clouds that come in and out, here and
there, and sometimes (with too regular a recurrence) they contain,
metaphorically, a black swan to potentially disrupt their best efforts at any
given time. Good luck to them; good luck
to us all.
No comments:
Post a Comment