Information Technology Reference
In-Depth Information
data, plugging current information into equations and following
charts and graphs. Their predictions may seem reasonably accurate
for a period of time, but so far each system has eventually failed—at
least from time to time.
In this regard, the basic problem is not a downfall of computers,
for computers are designed only to follow the instructions they are
given. Computers can produce many answers quickly, but the re
sults are always based on instructions. In the case of the stock mar
ket, people do not know what models to follow or what instructions
to specify. Think for a moment of all the factors that could influence
the stock market: politics, economics, environment, and the list goes
on. Some factors may even depend on human psychology—how
people feel about their lives and their surroundings—but no one
knows exactly how such factors will affect people's choices of indi
vidual stocks or investments. To date, we have only vague ideas
about investment behavior, and stockmarket predictions have some
qualities of looking into a crystal ball. Unfortunately, this lacks the
clarity required to produce an adequate algorithm. Thus, looking at
the results of computer computations only reinforces this conclusion
that people have not yet found an appropriate algorithm.
are unreliable, how I can tell when other computer
programs might be giving unreliable results?
Unfortunately, this is a very difficult question to answer. In pre
vious chapters, we have identified the stages of software develop
ment as specifications, design and algorithm development, coding,
testing, and maintenance. In this chapter, we have commented on
modeling and simulations. In practice, errors can arise in any stage
of a computer program's development. Here are some examples:
Specifications
: One can never truly know if the specifications
for a problem are complete and consistent. Various people can
review specifications in an attempt to find omissions or incon
sistencies, but this can only help refine the specifications—not
ensure they are correct. After all, an omission in specifications
could arise because no one thought of a particular scenario
that might happen; if no one thought of it ahead of time, then
no one will realize it was omitted until the situation actually
