---------- Forwarded message ----------
Date: 11 Dec 98 10:35:18
From: Roberto Verzola <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
To: [EMAIL PROTECTED],
[EMAIL PROTECTED]
Subject: [interdoc-y2k 17] Modularization vs globalization
I suggested in my earlier message that the shift from a
risk-minimizing strategy to a gain-maximizing strategy is one major
flaw which led to the Y2K problem.
A second flaw, I suggest, is the shift from a "modular" to a
"globalist" approach, which converts a network of relatively
independent but interconnected systems into a single tightly-coupled
complex system.
Below is an article on this topic which I had submitted earlier in the
GKD/Y2K discussions but which was not posted. The article provides a
theoretical argument for modularization, based on the experiences of
systems designers.
Roberto Verzola
---------------------
Our Economic System: Badly Designed?
by Roberto Verzola*
The international financial crisis which struck Asian countries in
1997 and continues to cause widespread damage this year is a perfect
example of what systems analysts call "the side-effects of global
variables."
Take the most complex systems ever designed -- like the Apollo
spacecraft system which took men to the moon and brought them back, or
computer chips that are made of tens of millions of components, or a
complex operating system with one hundred million lines of code. They
work as designed because the system designers followed certain rules
of design which time and again have been proven correct.
Follow the design rules, and you get a system that is robust and
reliable. Violate the design rules, and you get a system that is
unreliable and crash-prone.
One of the most important rules that good designers will never violate
is modularization: breaking up a complex system into relatively
independent modules, which are isolated from each other except for a
few well-defined interfaces. This design rule can be found in all
engineering and computer science texts. It is true for hardware and
software designs. Most complex systems that violated this rule ended
as miserable failures, while those which tried to implement it showed
much better rates of success.
The reason for the rule is simple: as the number of components in a
system increases, the number of possible interactions between
components rises exponentially. Normally, all possible interactions
must be checked for the possibility of unintended and undesirable
results, called "side-effects." But beyond a certain number of
components, it becomes impossible to double-check or even to trace the
results of every possible interaction. Because these potentially
undesirable side-effects increase at a faster rate than the number of
components, they eventually bring the whole system crashing down.
Designers had earlier argued against modularization because it was
"inefficient." Modular designs tended to use more components; a lot of
thought and effort had to go into the interfaces between modules; some
level of redundancy was required among the modules. But the loss in
efficiency was gained in reliability. Modular designs failed less
often (the average time between failures is a standard measure of
system reliability); and when they failed, errors were corrected
faster.
The history of systems design is replete with crashed spacecrafts and
crashed computer operating systems that drove home the point: complex
systems must be broken up into smaller, more managable, independent
modules; otherwise, you get an unreliable, failure-prone, or
unworkable design.
The opposite of modularization is globalization. It is true: that
favorite word of World Bank and IMF economists is an absolute no-no
among systems designers. Open any respectable textbook on computer
science or system design, and one of the first design rules you are
going to meet is: avoid anything that affects the entire system
globally. Break up large systems into smaller modules. Protect your
modules from interference by other modules. Isolate your modules from
each other. Hide information. Build firewalls.
Most of all, avoid global variables.
In an economic system, a global variable would be anything that can
affect many portions of a large system. Global corporations, because
they operate worldwide, would be a good example. The IMF, the World
Bank, and the World Trade Organization (WTO), because they intrude
into almost every economy in the world, would be another. Their moves
and decisions affect many other economies in the world, and result in
consequences and other interactions, that are so numerous that it
becomes impossible to anticipate and correct for undesirable
side-effects. These side-effect then proliferate; eventually, they can
bring the whole system down.
Unfortunately, most economists appear to have little understanding of
system design. (When I was in college, many of those who failed our
engineering subjects shifted to economics.) Instead of following good
principles of design, our economists repeat the most common mistake of
amateur programmers: they rely on global variables.
Instead of building protective firewalls around our economy, they tear
down existing walls of protection. Instead of strictly regulating
those global variables that breach the walls that remain, they launch
a perverse program of "deregulation," enlarging instead of restricting
the impact of global variables. All those legal infrastructures which
in the past protected us from the side-effects of global variables --
such as protective tariffs, foreign exchange controls, regulatory
mechanisms and others which would have dampened the impact of global
side-effects on our economy -- are being torn down.
Instead of blocking IMF, WTO and World Bank interference, they kneel
and bow before them. Instead of relying on local variables and local
interactions, which are manageable locally, they put greater reliance
on global markets and global players, touting this reliance as "sound
economic fundamentals."
It is interesting that neo-liberal economic theory conflicts with
systems theory, though both of them claim to be a science. Real
science, however, anticipates reality better than pseudo-science.
Looking at the current global financial crisis, tt should be obvious
which is which.
Until we learn the basic lessons of systems design, and apply these to
our own economy, we will be saddled with an unreliable, crash-prone
economic system, one which will cause us endless suffering.
There is another lesson we can learn from successful designs of the
past. If a system is badly-designed, and suffers from too many global
variables, any attempt at modification will likely produce even more
unintended side-effects. Often, it is better to junk the misdesigned
system altogether and to start again from scratch.
Saddled with a system that embraces globalization and leaves us at the
mercy of its side-effects, this is perhaps what we should also do.
* Roberto Verzola is an engineer who specializes in computers. Funded
by the Philippine government, he designed a computer system in 1981,
the first Filipino to do so. He also designed the software for first
online systems used at the Philippine Senate and House of
Representatives in 1991. He is also an activist, and is the
coordinator of Interdoc, a loose international network of NGOs
tracking the social impacts of new information technologies.
