Edited September
17, 2004
Anticipatory Web
Wednesday,
May 26, 2004
The two views of
the future and
two
understandings of the term
“complexity”
Complexity as
descriptive of computer programs
The Challenge
of Complexity (Vanguard)
Complexity as
descriptive of natural systems
The Challenge
of Complexity (Peter Krieg, Pile Systems)
Brief comment on
the correctness of Peter Krieg’s statement
The Challenge
of Complexity (Paul Prueitt, BCNGroup/Ontologystream)
Computing
systems are inherently complex and growing more so. We are now close to hitting
the complexity “wall,” a wall that threatens to hamper growth. Our hardware and
software systems have become so complex and so hard to maintain, that it’s
nearly impossible to think about or envision them as a whole. Complexity rears
its ugly head both for systems put together from a few extremely complex
components and for those developed and deployed with very large numbers of simple
units. The resulting systems are increasingly brittle and respond to change in
highly unpredictable ways. These systems are also labor intensive and are
costly to maintain.
In complex systems, we understand the
individual components, but often cannot predict or control the overall system.
The future will require more performance and flexibility, thereby increasing
complexity. How will this play out when we aren’t coping well with the
complexity we’ve already created?
New principles, tools and techniques are
needed. How can we redesign computer architectures, software, and systems to
create more robust systems? Systems of the future must be able to automatically
and autonomously adapt, maintain, repair and heal themselves. Increasingly,
we’re using models and metaphors borrowed from biology to find new solutions.
Will we be able to predict system behavior, and will we be able to get
synchronized behavior from non-synchronized events? Will we be in a position to
deal with emergent behavior?
The study of complexity must involve
mathematicians, social engineers, technologists, computer scientists,
economists, and biologists. As we build new devices that each contain a
computing environment, how can complexity be controlled? How can we
satisfactorily integrate new systems into our everyday lives?
(introductory text for a conference of the
Vanguard group, 2004)
Computing systems today are
inherently complicated, not complex.
Their problem is exactly this restriction to complication. The “red brick wall”
of complexity that they are already beginning to hit (see Intel’s thermo
problems) is actually a “complication wall”. As long as we do not differentiate
clearly between complexity and complication, we will neither understand
the problem nor see the solution.
The difference lies in the logic: a
complicated system is a mechanism, an
analytical system restricted to deductive inference within one single logic
domain as its frame of reference. As already Kant has observed, nothing new
will come from such a system.
Complexity arises, when two or more
different frames of reference (or logic domains) intersect. This creates ambiguity
in any description that tries to map these frames of reference into one
single “polylogic” map capable to both integrate and differentiate
independent logic domains. A cognitive system is capable of doing just that: it
operates both analytically (“rational”) and synthetically
(“intuitive”); it synthesizes through analogical, intuitive reasoning new
ideas or theories, from which it then can deduct analytically.
Complex polylogic descriptions are
necessary to describe dynamic systems, but particularly all living systems, as
these descriptions have to integrate their “inner logic” with the “outer
logics” of time and environment. Quantum theory, moreover, teaches us that not only living system, but
all physical systems require complex (polylogic) descriptions.
A true “thinking machine” therefore
must be able to combine synthetic and analytical reasoning. The basic condition
for this is complex, polylogic mapping. Today’s computing systems are not able to
map complexity because their software architectures are still mono-logic. This
is the reason why, although modern computer chips are much faster than most
neural processes, even the largest, most complicated supercomputers are less
adaptive than an insect with its pin-size, yet complex brain…
To avoid hitting the “Red Brick
Wall” we need computers that break through the monologic complication barrier
into polylogic complexity.
http://www.pilesys.com/tech.htm http://www.pilesys.com/Red%20Brick%20Wall.pdf
The understanding that computer science is confusing us about fundamental issues can be nowhere more illustrated than in the use of the term “complex” by computer scientists, and information technologists.
I have more to say on the issue of complex science and knowledge management in the Preface to Foundations of Knowledge Science. The more difficult issues of science are developed in the book. Peter and I are in complete agreement in regards to the need for a polylogic mechanism to arbitrate when the machine ontology pushes us towards points of complexity (defined in exactly the way that Peter does)
How, the other misunderstanding in the Vanguard statement has to do with how we as a society might end up dealings with increasing complicated data. In other words, given that there is a lot of “simple” data how does one deal with this?
There are several new ways, all which converge on a concept called “fractal compression”. My work on Orbs and SLIP demonstrate how new data might be encoded with less and less effort as the data set gets larger. There are some other technologies that do the same. But the business community has not seen any of this as yet.
The Vanguard statement can be a good beginning but the statement needs to not reflect the common, and incorrect wisdom. We are interested in helping the business community understand the advent of anticipatory web technology.
We invite a discussion on this. Post to postal@ontologystream.com