Information dynamics, selforganization and the
implications for management
IS4IS 2015 Vienna
John Collier
Philosophy, University of KwaZulu-Natal, Durban
4041, South Africa
E-Mail: collier@ncf.ca
Outline Part 1
1. Some major classes of information used in science form a hierarchy.
2. Within each level of the hierarchy are further hierarchies of
organization. These originate by self-organization (of two distinct
types, reorganization through dissipation and self-organization
through the promotion of perturbations).
3. The same basic systems principles apply across all levels, but new
properties appear at higher levels due to new possibilities created
at lower levels.
4. These principles are grounded in growth and self-organization, but
an important aspect is that there are natural system resonances
and it is easier (requires less energy) to form and maintain these
resonances than others. Preferably through self-organizing
processes. These resonances are typical larger scale and often
irreducible.
Outline Part 2
1. I will give some examples from physics on up through
biology. The explanation will be clearest for the physics
case, and I will apply it by analogy to other cases.
2. I will argue (speculate in the social case) that the same
principles apply for energy decoupled information
systems.
3. I will draw (speculate on) some consequences for social
level management.
4. The most significant of these is that the management
technique most likely to be effective if enough time is
available to use it is facilitation.
Information and Boundary Conditions
• In standard Hamiltonian mechanics of conservative or
near conservative systems, boundaries are not time
dependent (the systems are holonomic)
• This fails in many types of systems in which the
boundary conditions interact with system laws in a
mathematically inseparable way. These systems are
radically nonholonomic.
• I have argued elsewhere that such systems can show
emergent properties, but my concern here is that
boundary conditions deal with forms, and are best
described by information methods.
• So the dynamics of information become important for
systems with dynamical boundary conditions.
Energy and information budgets
• If the information budget of such systems is largely
decoupled from the energy budget, then we can deal
with information dynamics separately.
• This occurs notably in biological systems, but also in
such basic physical systems as black holes, the study of
whose boundary conditions led to the “it from bit”
view of nature, which extends an informational
analysis to the boundaries of elementary particles and
their consequent physical parameters, including mass.
• So information dynamics has certain general properties
that apply widely.
Major classes of information
Social
Cognitive
Functional
Hierarchical
Negentropy
It from Bit
How the different kinds of
information are related
• The kinds are nested, with the most basic kind in the middle.
• Each successive kind introduces further restrictions, which in turn create
new immediate possibilities. This in turn feeds inwards, e.g.:
1. The social requires communication between individuals, which creates new
possibilities for the individuals in addition to social possibilities. This can be
further iterated within the social kind.
2. Negentropy creates possibilities for self-organization, allowing new structural
information (a form of negentropy), which is new “its”. At the same time the
new constraints permit variant complexions within the constraints, leading to
an entropy of the structural information. This can be iterated within this kind.
• There is a general pattern of common principles applying to each kind
through the expression of new information within that kind, so each kind
has levels as well, constraining inwards. This basically reiterates the
situation between information kinds, but within the same realm of
science.
Viewed hierarchically, e.g.
Common
Principles
Across
Levels
(Systems
Theoretic)
Selforganization
creates new
information
within levels.
New forms of
interaction
produces
new kinds of
information.
Major
levels
Minor levels hierarchical information
(examples)
Molecules
Atoms
Basic particles
Minor levels Negentropy
“Clumped matter”
Kinds of matter – kinds of radiation
Matter-Radiation
Minor levels It from bit
Black hole horizons
Event horizon, particle horizon
Origin of universe (universe
boundary)
Universal (multilevel) processes
• Growth: Order and Entropy (disorder) can
increase together. Order = Smax - Sactual
• This “overhead” permits self-organization:
– reorganization to lowest energy state by dissipation –
noise dissipated, also natural selection
– spontaneous self-organization (state change,
bifurcation) through continuous dissipation of free
energy → new information
• Creates new possibilities for interaction (reduced
abstract possibility space constrains processes,
making them more immediately likely).
Entropy and information
• The basic physical relation is given by Boltzmann's constant in a
suitable form, leading to measures in entropy units or bits.
• Microstates, according to David Layzer, have microinformation. Part
of this is the information of the macrostate in which they are
contained, called macroinformation.
• Not all microinformation is accessible to a macroscopic entity (2nd
Law of Thermodynamics)
• Macrostates, to be real, must be cohesive, binding over space and
time.
• The same statistical principles apply to any system in which the
microstates are not constrained by anything but the macrostate, so
we can apply the related principles generally.
• In particular, we can talk of the macrostate of an information that
has informational microstates. An example is the gene pool of a
population or species. (Variation of the gene pool is taken to be
random with respect to the properties of the species – in fact this is
not always true and must be controlled for.)
Rhythmic Entrainment
• Rhythmic entrainment is a ubiquitous phenomenon that produces
large scale coordination either through the elimination of
interfering factors (re-organization) or through the emergence of
higher level order (self-organization).
• It produces new macrostates.
• In both cases the result is a sort of harmony at a larger scale while
permitting variations at lower scales. Both produce new
possibilities, as above.
• Entrainment can be produced either by forcing (constraining the
system at a large scale), or spontaneously.
• The former requires more work (energy expenditure) to produce
and maintain than the latter.
• It is worth noting that entrainment produces symmetries, and it is
the symmetries that reduce the amount of work required.
Externally forced systems are typically in more asymmetric states.
PART 2 EXAMPLES AND
APPLICATIONS
Physical systems
Forced Harmonic Oscillator
Some other physical cases
• Bénard cell convection and other vortices
• Planetary resonances – Earth-Moon, MercurySun, Jupiter’s major moons, Pluto and Neptune.
• Japanese satellite to the moon
• Various climate phenomena (el Niño, North
Atlantic Oscillation …)
• Old Faithful
• Geochemical differentiation
Jeremejevite Al6(BO3)5(F,OH)3
Chemical systems
Benzene
Chemical systems
Ethylene
Butadiene
Biological systems (evolution and
development)
• D. R. Brooks and E. O. Wiley, 1988, Evolution as Entropy, argued that far
from being close to equilibrium and under control of environment and or
genes, biological populations are far from equilibrium (panmixis), a
condition maintained by growth in their phase space (Smax ) faster than the
phase space is filled out (Sactual). The microstates are variant gene
distributions and the macrostates are the gene pool determined by
species cohesion (usually interbreeding connections).
• This allows further self-organization between the diversity achieved and
maximum possible diversity. This can lead to speciation, which increases
the system entropy (as defined).
• Furthermore, this allows the formation of layered or hierarchically
organized nested self-organizing systems.
• So Brooks and Wiley generalized to development, historical ecology and
other areas.
• Such systems are highly self-regulating, while remaining flexible, unlike
“optimally evolved” systems that are likely to end up on local maxima,
making further adaptation difficult of not impossible.
Biological systems (ecosystems)
• R. E. Ulanowicz, Ecology, the Ascendent Perspective, 1997, argued
that overhead, similar to Brooks and Wiley’s system diversity, and
ascendency, similar to Brooks and Wiley’s potential diversity,
combine to define ecosystem capacity.
• Ascendency corresponds to ecosystem stability at a given time. It is
produced by dominance of more efficient interactions between
nodes of the ecosystem.
• A healthy ecosystem must have a good deal of overhead, or else it
is too committed to present trends to be flexible in the face of new
factors.
• Ascendency in one factor (e.g., nitrogen) can be isolated to some
extent, but it is overall ascendency that matters.
• Importantly, the ecosystem ascendency rises due to internal
features of the system itself, so it is self-stabilizing towards the most
efficient exchanges.
Human systems (begin speculation)
• The general principles should apply to human systems,
especially social systems, but the problem is to find proper
analogues of dissipation, energy, information and other
properties essential to self-organizing systems.
• The basic issue is communication of ideas and behaviours
and how these can self-organize.
• In general, there must be excess (unconstrained) diversity
in the system and a sink (exit) for diversity in order to allow
self-organization.
• Forced order and organization (external constraint of
diversity) will be relatively costly in comparison to selforganization.
What is the currency?
• Economics gives a clear currency.
• Neoclassical economics assumes a sort of neoDarwinian
optimality leading to equilibrium in the market.
• However incomplete information, variations in trading
times in the market, “animal passions”, and most
importantly, unequal information, distort the market so
that the optimality assumption fails.
• In general, the market changes faster than it can
equilibrate. So self-organization should be expected. That is
all I can say right now about economics.
• But in general the currency of social systems is information,
and it is passed more or less efficiently by communication.
Money is a special case.
Management desiderata
• Productivity – the system should produce useful
results relatively efficiently. These might be:
– Sales (industry)
– Societal benefit (governments, social agencies)
– New ideas (think tanks, research units)
These will require different focus (on results) but it isn’t
clear that different management organization is
required. However, different time scales for returns
might justify different management methods.
• Minimization of unexpected (crises, disasters,
externalities producing antagonism)
What is to be managed?
• System organization (productivity is a result of
this, and cannot be ensured directly): determines
to some degree information flow and dynamics
• Information flow: each unit must have the
information it needs, and provide information
other units need. Some units can produce
information themselves, others cannot, and
others are mixed.
• Information flow cannot be demanded
effectively, it can only be encouraged.
Top down hierarchical
management
Usually a control structure where top
determines lower level behaviour.
Advantages
 Can get results quickly
 Perhaps good for emergencies if
lowest level is well trained (eg
emergency services, military)
 Focussed tasks work best
Disadvantages
 Results limited to constraints
central management can apply
 Unintended consequences likely
both internal and in externalities
 Ensuring results is difficult (and
typically costly)
 Enforcing the order is typically
costly
 Lack of flexibility
Central
Management
Local or task
management
1
Production 1
Production 2
Local or task
management
2
Production3
Problems with top-down control
as a form of governance
• Authoritarian and totalitarian states put a premium on top-down control.
• This will work better if there is a clear focus or a small set of ideological
goals. A sense of emergency also helps justification.
• Requires a large concentration of political, economic and social power,
since driving a system artificially always requires a lot of power and waste
of energy.
• A more efficient but less reliable method is propaganda and advertising,
which attempt to drive or create resonances through subtle forcing
(constraints on what is acceptable to do or say. (Chomsky)
• But this method still requires a concentration of power to exclude
competitors.
• Such a social system is always fighting a gradient towards more efficient
organizations.
• This is one of the primary social lessons we can derive from complexity
theory: forced control is unstable and expensive (wasteful).
What about “flat governance”?
• The idea is that everyone makes their own decisions, so
“flat”.
• A form of radical libertarian anarchism in its most extreme
form but also compatible with some forms of democracy.
• Not all democracy, since there might be a vote for a central
authority with no limits on its power and we are back at the
previous case.
• Representative democracy comes someplace between,
closer to the top-down model, and direct democracy with
votes required on all decisions comes closer to the flat
model.
• But it is unclear how any of these involve room for the
production of new social information, new ideas or visions.
Consensus?
• Consensus governance requires full agreement
on any group issue.
• It can work well with small groups with largely
common beliefs, but the worry is always what to
do about holdouts.
• Furthermore, common beliefs constrain possible
solutions, perhaps overly (e.g., the Amish).
• So consensus seems to be either overly restrictive
or else to difficult to achieve.
Facilitation rather than
control or chaos
• The management model that best fits the selforganized complexity model of social systems is
to encourage diversity with minimal top-down
control. I call this facilitation.
• Facilitators concern themselves with attaining
group agreement, and focus on the whole. There
role is to suggest, not demand.
• They need not be a single person or the same
people throughout. It is largely a voluntary role.
• When a facilitator finds themselves being treated
as an authority, it is best they step back.
Some features of
facilitative governance
• It is a variation on anarchism, but could be applied
within many different management systems.
• Facilitators should encourage diversity while still
helping to maintain focus, permitting self-organized
solutions. Too much diversity “saturates” the system,
whereas to little, or too little communication, makes it
unstable (recall Ulanowicz)
• Often the introduction of a novel point will lead to
reorganization of discussion around that point. This
tends to be temporary unless resolution is reached.
• A disadvantage is that its gentleness implies that it is
slow (recall Japanese satellite).
Thanks for your attention.
Questions?
John Collier
http://web.ncf.ca/collier