|
IT looks like it’ll be another boom year for the prediction
industry. Adjusted seasonally, it seems that there are more
consumers of forecasts this month than there were in January last
year.
Astrology seems to have lost market share. Feng
shui or geomancy is ascendant. Astrologers now cherry-pick
astronomical charts to divine what the months ahead hold. This
selective use of science, however, has failed so far to restore
their craft to its former pre-eminence.
Soothsaying is one of the oldest professions.
Among the duties of a priest during the time of the pharaohs was to
foretell the future. Divining the future became a distinct
occupation in Delphi where oracles were consulted by the anxious,
distraught or confused. Roman haruspicy was an acquired taste that
did not win wide acceptance.
Today, those in the prediction business have all
but shed their reliance on religion and magic. They are now mere
artisans in the service sector. A fortuneteller’s shop is not
unlike a beauty parlor. Geomancers are hired, like plumbers, to help
orient a house or rearrange the furniture. Probably it’s
apochrypal but a columnist friend of mine swears that the front page
of a national daily was designed by a feng shui master. Talk of
market penetration!
It’s not unlikely that one of these days the
Dictionary of Occupations will have an entry for this category of
service worker with a standard title, job description, and 5-digit
code.
All this is harmless fun. But not so harmless
are the predictions of economists, bankers, and sociologists.
Investment decisions, government targets, even trade flows are more
often than not influenced positively or negatively by forecasts that
are not dissimilar to those made by the oracle at Delphi.
Isn’t there a better way to make economic and
social forecasts? V.S. Subrahmanian of the Institute for Advanced
Computer Studies of the University of Maryland in the US said in his
article in Science (Sept. 14, 2007) that “Accurate forecasts
depend critically upon the ability to build behavioral models of the
people and groups involved.”
The traditional method is to construct models
based on surveys or face-to-face interviews or by immersing oneself
in a community or group and then making hypotheses and testing
correlations by means of various statistical methods.
Not only is this approach expensive and
time-consuming, there are many situations where it’s difficult to
carry out. In conflict areas, for example, surveys are not easy to
do and, more important, almost impossible to update with the same
respondents.
Subrahmanian thinks that “computational social
models may offer the best solution in cases where conventional data
gathering is not possible.”
He suggest tools such as The Resource
Description Framework Extraction (T-REX) that use social, cultural,
political, economic and religious (SCPER) variables that can be
derived automatically from news sources, blogs, YouTube, social
networks, and wikis. The SCPER variables can include financial
activities, crimes and other violent events, and political
relationships. Automatic analysis of the data from these sources
could provide early warning of potential conflicts. Unlike the
traditional methods, “these methods do not require previous
knowledge of the groups being investigated.”
A software, Cultural Reasoning Architecture
(CARA), was able to predict the probability of suicide attacks by
the Hezbollah in Lebanon by including parameters like education and
propaganda. When these were used as part of Hezbollah’s strategy,
the probability of suicide attacks was around 47 percent; when they
were not, the probability rose to 80 percent. These were
automatically discovered by CARA with a data set called
“Minorities at Risk.”
In any real life situation, the number of
possible determining conditions is so large that human processing
results in judgments that are uncertain and sometimes even
capricious. T-REX, for example, generates as many as 45,000 pages of
data a day. Classification algorithms become indispensable to
discriminate desirable from undesirable conditions.
These methods can be used to model in real time
terror groups, political parties, regulatory bodies, intra-corporate
disputes and even family conflicts.
The ability to make predictions on the outcome
of initial actions of individuals and groups and to estimate the
probable sets of actions can be of enormous advantage in making
tactical and strategical choices. They can be used, to paraphrase
Subrahmanian, to stop terrorism or to improve corporate projects.
Astrologers, fortunetellers, and geomancers will
continue to thrive because they appeal to the irrational in human
beings. But economists, bankers, and sociologists who aim for
rationality in making predictions could use Subrahmanian’s ideas
to their and their clients’ advantage.
|
