380 John Woods
18.2.2 Software sensing
Many important features of the plankton ecosystem cannot be observed with sufficient clarity to permit an accurate empirical description in the sense of Bacon
(1605). So our knowledge of biological oceanography depends largely on mathematical simulation. The reliable scientific base provided by primitive equations
encourages us to believe simulated features that cannot be observed. Modellers
give the name "software sensing" to the procedure by which mathematical simulation is used to measure unobservable features of complex systems. Even when we
do have observations, the information obtained from software sensing is usualIy
more comprehensive. Such completeness and seeming realism are seductive. So
we must take care to control its quality by identifying and quantifying the uncertainties in our modelling technique (see below 18.17-18). With that note ofcaution,
we proceed on the basis that PE modelling will provide a scientificalIy sound basis
for software sensing. Models that are not based on primitive equations will tend to
be less satisfactory.
18.3 Biological primitive equations
Biologists often suppose that modelling with primitive equations applies only to
physical systems, like planetary motion, tides, ocean circulation and weather. My
aim is to show how it can also be applied to systems, which involve not only physics, but also chemistry and biology.
1 define Biological Primitive Equations (BPE) in the sense of Galileo (1638).
They are mathematical equations established by analysing data from reproducible
laboratory experiments designed to describe fundamentallaws of nature for living
organisms. Descartes (1664) was the first scientist to contemplate PE modelling of
biologic al systems. Rather ambitiously he thought of doing so for the physiology of
Man. Later, Turing (1950) and others seriously contemplated an even more ambitious challenge: PE modelling ofthe mind. We now know that the sheer complexity
of human physiology and mind, and of the brain which controls them, places such
goals beyond our reach for the foreseeable future. Neverthe1ess, there is growing
research activity in modelling complex systems involving people (Casti 1997).
18.3.1 Complex Adaptive Systems
It has became standard practice to divide complex systems into two categories:
(1) those that are based on mode1s comprising the immutable equations of physics,
and (2) those that are based on models with equations for the model agents (usualIy
people) that can adopt new rules when circumstances change. Weather forecasting
falIs into the first category, financial markets falI into the second. The latter are
called Complex Adaptive Systems, because they involve decisions by people who
can change their rules ofbehaviour. The equations remain conjectures because it is
not yet possible to perform reproducible laboratory experiments to describe mathe-
18.2.2 Software sensing
Many important features of the plankton ecosystem cannot be observed with sufficient clarity to permit an accurate empirical description in the sense of Bacon
(1605). So our knowledge of biological oceanography depends largely on mathematical simulation. The reliable scientific base provided by primitive equations
encourages us to believe simulated features that cannot be observed. Modellers
give the name "software sensing" to the procedure by which mathematical simulation is used to measure unobservable features of complex systems. Even when we
do have observations, the information obtained from software sensing is usualIy
more comprehensive. Such completeness and seeming realism are seductive. So
we must take care to control its quality by identifying and quantifying the uncertainties in our modelling technique (see below 18.17-18). With that note ofcaution,
we proceed on the basis that PE modelling will provide a scientificalIy sound basis
for software sensing. Models that are not based on primitive equations will tend to
be less satisfactory.
18.3 Biological primitive equations
Biologists often suppose that modelling with primitive equations applies only to
physical systems, like planetary motion, tides, ocean circulation and weather. My
aim is to show how it can also be applied to systems, which involve not only physics, but also chemistry and biology.
1 define Biological Primitive Equations (BPE) in the sense of Galileo (1638).
They are mathematical equations established by analysing data from reproducible
laboratory experiments designed to describe fundamentallaws of nature for living
organisms. Descartes (1664) was the first scientist to contemplate PE modelling of
biologic al systems. Rather ambitiously he thought of doing so for the physiology of
Man. Later, Turing (1950) and others seriously contemplated an even more ambitious challenge: PE modelling ofthe mind. We now know that the sheer complexity
of human physiology and mind, and of the brain which controls them, places such
goals beyond our reach for the foreseeable future. Neverthe1ess, there is growing
research activity in modelling complex systems involving people (Casti 1997).
18.3.1 Complex Adaptive Systems
It has became standard practice to divide complex systems into two categories:
(1) those that are based on mode1s comprising the immutable equations of physics,
and (2) those that are based on models with equations for the model agents (usualIy
people) that can adopt new rules when circumstances change. Weather forecasting
falIs into the first category, financial markets falI into the second. The latter are
called Complex Adaptive Systems, because they involve decisions by people who
can change their rules ofbehaviour. The equations remain conjectures because it is
not yet possible to perform reproducible laboratory experiments to describe mathe-