Complexity & Interaction: Blurring Borders between Physical, Computational, and Social Systems. Preliminary Notes

Complexity & Interaction: Blurring Borders
between Physical, Computational, and Social Systems
Preliminary Notes
Andrea Omicini
andrea.omicini@unibo.it
with Pierluigi Contucci
pierluigi.contucci@unibo.it
DISI / DM
Alma Mater Studiorum—Universit`a di Bologna
ICCCI 2013
Craiova, Romania, 11 September 2013
Omicini, Contucci (DISI/DM, Alma Mater) Complexity & Interaction ICCCI 2013, 11/9/2013 1 / 47

Outline
1 Interaction & Complex Systems
Complexity
Computational Systems
Physical Systems
Social Systems
2 Perspectives
Coordination Models
Socio-technical Systems
3 Final Remarks

Interaction & Complex Systems Complexity
Outline
Complexity
Physical Systems
Social Systems
2 Perspectives
Coordination Models
3 Final Remarks

Complexity as a Multi-disciplinary Notion
Complex systems everywhere
The notion of complexity is definitely a multi-disciplinary one,
ranging from physics to biology, from economics to sociology and
organisation sciences
Systems that are said complex are both natural and artificial ones
Natural vs. artificial complex systems
We observe and model complex physical systems
We design and build complex computational systems
Question
Which features do all complex systems share independently of their
nature?

Complexity & Interaction
. . . by a complex system I mean one made up of a large number
of parts that interact in a non simple way [Simon, 1962]
Laws of complexity
Some “laws of complexity” exists that characterise any complex
system, independently of its speciﬁc nature [Kauﬀman, 2003]
The precise source of what all complex systems share is still unknown
in essence
Interaction
We argue that interaction – its nature, structure, dynamics – is the
key to understand some fundamental properties of complex systems of
any kind

Interaction & Complex Systems Computational Systems
Outline
Complexity
Physical Systems
Social Systems
2 Perspectives
Coordination Models
3 Final Remarks

Interaction in Complex (Computational) Systems I
Interaction as a computational dimension
Interaction as a fundamental dimension for modelling and engineering
complex computational systems
For instance, a well-founded theory of interaction is essential to model
sociality [Castelfranchi et al., 1993] and situatedness
[Mariani and Omicini, 2013] in multi-agent systems (MAS)

Interaction in Complex (Computational) Systems II
Interaction as a source of complexity
Interaction is the most relevant source of complexity for
computational systems nowadays [Wegner, 1997]
roughly speaking, when interaction within a system is (not) relevant,
system properties cannot (can) be straightforwardly derived by
component properties
compositional vs. non-compositional systems
computer scientists vs. computer engineers
system formalisability vs. system expressiveness
For instance, interaction is the main source of emergent social
phenomena in MAS [Castelfranchi, 1998]

Interaction in Complex (Computational) Systems III
Interaction as a first-class issue
The inter-disciplinary study of interaction in many diverse scientific areas dealing
with complex systems basically draws the foremost lines of evolution of
contemporary computational systems [Omicini et al., 2006]
Interaction — an essential and independent dimension of computational systems,
orthogonal to mere computation
[Gelernter and Carriero, 1992, Wegner, 1997]
Environment — a first-class abstraction in the modelling and engineering of complex
computational systems, such as pervasive, adaptive, and multi-agent
systems [Weyns et al., 2007]
Mediation — environment-based mediation [Ricci and Viroli, 2005] is the key to
designing and shaping the interaction space within complex software
systems, in particular socio-technical ones [Omicini, 2012]
Middleware — provides complex socio-technical systems with the mediating
abstractions required to rule and govern social and environment
interaction [Viroli et al., 2007]

Interaction & Complex Systems Physical Systems
Outline
Complexity
Physical Systems
Social Systems
2 Perspectives
Coordination Models
3 Final Remarks

Interaction in Statistical Mechanics I
Independence from interaction
Some physical systems are described under the assumption of mutual
independence among particles—that is, the behaviour of the particles
is unaﬀected by their mutual interaction
e.g., ideal gas [Boltzmann, 1964]
There, the probability distribution of the whole system is the product
of those of each of its particles
In computer science terms, the properties of the system can be
compositionally derived by the properties of the individual
components [Wegner, 1997]
→ Neither macroscopic sudden shift nor abrupt change for the system as
a whole: technically, those systems have no phase transitions—of
course, while the “independence from interaction” hypothesis holds

Interaction in Statistical Mechanics II
Interacting systems
Introducing interaction among particles structurally changes the
macroscopic properties, along with the mathematical ones
Interacting systems are systems where particles do not behave
independently of each other
The probability distribution of an interacting system does not
factorise anymore
In computer science terms, an interacting system is
non-compositional [Wegner, 1997]

Interaction in Statistical Mechanics III
Interacting vs. non-interacting systems
Only interacting systems can describe real cases beyond the idealised
ones
e.g., they can explain phase transitions – like liquid-gas transition – and
much more, such as collective emerging eﬀects
While a system made of independent parts can be represented by
isolated single nodes, an interacting system is better described by
nodes connected by lines or higher-dimensional objects
From the point of view of information and communication theories,
an ideal non-interacting gas is a system of non-communicating nodes,
whereas an interacting system is made of nodes connected by
channels

Complexity in Statistical Mechanics I
The case of magnetic particles
The simplest standard prototype of an interacting system is the one made of
magnetic particles
There, individual particles can behave according to a magnetic field which
leaves their probabilistic independence undisturbed
At the same time, two magnetic particles interact with each other, and the
strength of their interaction is a crucial tuning parameter to observe a phase
transition
If interaction is weak, the effect of a magnetic field is smooth on the system
Instead, if the interaction is strong – in particular, higher than a threshold –
even a negligible magnetic field can cause a powerful cooperative effect on
the system
The system can be in one of two equilibrium states: the up and the down
phase

Complexity in Statistical Mechanics II
Interaction is not enough
Interaction is a necessary ingredient for complexity in statistical mechanics
but definitely not a sufficient one
Complexity arises when the possible equilibrium states of a system grow very
quickly with the number of particles, regardless of the simplicity of the laws
governing each particle and their mutual interaction
Roughly speaking, complexity is much more related to size in number, rather
than to complexity of the laws ruling interaction
→ we do not need complex interaction to make interaction lead to complexity
In the so-called mean field theory of spin glasses [Mézard et al., 1986],
particles do not just interact, but are alternatively either imitative or
anti-imitative with the same probability [Contucci and Giardinà, 2012]
Both prototypical cooperation and competition effects can be observed, and
the resulting emerging collective effect is totally new

Interaction & Complex Systems Social Systems
Outline
Complexity
Physical Systems
Social Systems
2 Perspectives
Coordination Models
3 Final Remarks

From Statistical Mechanics to Social Systems I
Large numbers
The key point in statistical mechanics is to relate the macroscopic
observables quantities – like pressure, temperature, etc. – to suitable
averages of microscopic observables—like particle speed, kinetic
energy, etc.
Based on the laws of large numbers, the method works for those
systems made of a large number of particles / basic components

From Statistical Mechanics to Social Systems II
Beyond the boundaries
Methods for complex systems from statistical mechanics have
expanded from physics to fields as diverse as biology
[Kauffman, 1993], economics
[Bouchaud and Potters, 2003, Mantegna and Stanley, 1999], and
computer science itself
[Mézard and Montanari, 2009, Nishimori, 2001]
Recently, they have been applied to social sciences as well: there is
evidence that the complex behaviour of many observed
socio-economic systems can be approached with the quantitative
tools from statistical mechanics
e.g., Econophysics for crisis events [Stanley, 2008]

From Statistical Mechanics to Social Systems III
Social systems as statistical mechanical systems
A group of isolated individuals neither knowing nor communicating
with each other is the typical example of a compositional social
system
No sudden shifts are expected in this case at the collective level,
unless it is caused by strong external exogenous causes
To obtain a collective behaviour displaying endogenous phenomena,
the individual agents should meaningfully interact with each other
The foremost issue here is that the nature of the interaction
determines the nature of the collective behaviour at the aggregate
level
e.g., a simple imitative interaction is capable to cause strong
polarisation eﬀects even in presence of extremely small external inputs

Perspectives Coordination Models
Outline
Complexity
Physical Systems
Social Systems
2 Perspectives
Coordination Models
3 Final Remarks

Modelling vs. Engineering
Physical vs. computational systems
Physical systems are to be observed, understood, and possibly
modelled
→ For physical systems, the laws of interaction, and their role for
complexity, are to be taken as given, to be possibly formalised
mathematically by physicists
Computational systems are to be designed and built
→ For computational systems, the laws of interaction have ﬁrst to be
deﬁned through amenable abstractions and computational models by
computer scientists, then exploited by computer engineers in order to
build systems

A Meta-model for Coordinated Systems I
The coordination meta-model [Ciancarini, 1996]
Coordination entities — the entities whose mutual interaction is ruled by
the model, also called the coordinables (or, the agents)
Coordination media — the abstractions enabling and ruling interaction
among coordinables
Coordination laws — the rules governing the observable behaviour of
coordination media and coordinables, and their interaction as
well

A Meta-model for Coordinated Systems II
interaction space
coordinable
coordination
medium
coordinable
coordinable
coordination
medium
coordination
medium

A Meta-model for Coordinated Systems III
The medium of coordination. . .
“ﬁlls” the interaction space
enables / promotes / governs the admissible / desirable / required
interactions among the interacting entities
according to some coordination laws
enacted by the behaviour of the medium
deﬁning the semantics of coordination

Coordinated Systems as Interacting Systems I
Coordination media for ruling interaction
Deﬁning the abstractions for ruling the interaction space in
computational systems basically means to deﬁne their coordination
model [Gelernter and Carriero, 1992, Ciancarini, 1996,
Ciancarini et al., 1999]
Global properties of complex coordinated systems depending on
interaction can be enforced through the coordination model,
essentially based on its expressiveness
[Zavattaro, 1998, Denti et al., 1998]
For instance, tuple-based coordination models have been shown to be
expressive enough to support self-organising coordination patterns for
nature-inspired distributed systems [Omicini, 2013]

Coordinated Systems as Interacting Systems II
The role of coordination models
Coordination models could be exploited
to rule the interaction space
so as to deﬁne new sorts of global, macroscopic properties for
computational systems, possibly inspired by physical ones

Coordinated Systems as Interacting Systems III
Research perspectives
We need to understand
how to relate methods from statistical mechanics with coordination
models
whether notions such as phase, phase transition, or any other
macroscopic system property, could be transferred from statistical
mechanics to computer science
what such notions would imply for computational systems
whether new, original notions could apply to computational systems
which sort of coordination model could support such notions

Perspectives Socio-technical Systems
Outline
Complexity
Physical Systems
Social Systems
2 Perspectives
Coordination Models
3 Final Remarks

Socio-Technical Systems
Humans vs. software
Nowadays, a particularly-relevant class of social systems is
represented by socio-technical systems
In socio-technical systems
active components are mainly represented by humans
whereas interaction is almost-totally regulated by the software
infrastructure
where software agents often play a key role
This is the case, for instance, of social platforms like FaceBook
[FaceBook, 2013] and LiquidFeedback [LiquidFeedback, 2013]

Physical & Computational Social Systems I
A twofold view of socio-technical systems
The nature of socio-technical systems is twofold: they are both social
systems and computational systems
[Verhagen et al., 2013, Omicini, 2012]
As complex social systems, their complex behaviour is in principle
amenable of mathematical modelling and prediction through notions
and tools from statistical mechanics
As complex computational systems, they are designed and built
around some (either implicit or explicit) notion of coordination, ruling
the interaction within components of any sort—be them either
software or human ones

Physical & Computational Social Systems II
Computational systems meet physical systems
In socio-technical systems, macroscopic properties could be
described by exploiting the conceptual tools from physics
enforced by the coordination abstractions
Socio-technical systems could exploit both
the notion of complexity by statistical mechanics, along with the
mathematical tools for behaviour modelling and prediction, and
coordination models and languages to suitably shape the interaction
space

Physical & Computational Social Systems III
Vision
We envision complex socio-technical systems
whose implementation is based on suitable coordination models
whose macroscopic properties can be modelled and predicted by
means of mathematical tools from statistical physics
thus reconciling the scientist and the engineer views over systems

Final Remarks
Conclusion I
Interaction in complex systems
Interaction is key issue for complex systems
Interacting systems in physics
Coordinated systems in computer science
Socio-technical systems such as social platforms
e.g., FaceBook, LiquidFeedback

Final Remarks
Conclusion II
The role of coordination models
Coordinated systems as interacting systems
Coordination models as the sources of abstractions and technology
for enforcing global properties in complex computational systems,
which could then be
modelled as physical systems, and
engineered as computational systems
Case study
Socio-technical systems such as large social platforms could represent a
perfect case study for the convergence of the ideas and tools from
statistical mechanics and computer science, being both social and
computational systems at the same time

Final Remarks
Conclusion III
Next steps
We plan to experiment with social platforms like FaceBook and
LiquidFeedback, by exploiting
coordination technologies for setting macroscopic system properties
statistical mechanics tools for predicting global system behaviour

Final Remarks
Further References
Paper
Reference [Omicini and Contucci, 2013]
APICe http://apice.unibo.it/xwiki/bin/view/
Publications/InteractioncomplexityIccci2013
Presentation
APICe http://apice.unibo.it/xwiki/bin/view/Talks/
InteractioncomplexityIccci2013
Slideshare http://www.slideshare.net/andreaomicini/complexity-interaction-
blurring-borders-between-physical-computational-and-social-
systems-preliminary-notes

Final Remarks
Acknowledgements
Thanks to. . .
everybody here at ICCCI 2013 for listening
Costin B˘adic˘a for inviting me for the Keynote Speech

References
References I
Boltzmann, L. (1964).
Lectures on Gas Theory.
University of California Press.
Bouchaud, J.-P. and Potters, M. (2003).
Theory of Financial Risk and Derivative Pricing: From Statistical Physics to Risk
Management.
Cambridge University Press, Cambridge, UK, 2nd edition.
Castelfranchi, C. (1998).
Modelling social action for AI agents.
Artiﬁcial Intelligence, 103(1-2):157–182.
Castelfranchi, C., Cesta, A., Conte, R., and Miceli, M. (1993).
Foundations for interaction: The dependence theory.
In Torasso, P., editor, Advances in Artiﬁcial Intelligence, volume 728 of Lecture
Notes in Computer Science, pages 59–64. Springer Berlin Heidelberg.

References
References II
Ciancarini, P. (1996).
Coordination models and languages as software integrators.
ACM Computing Surveys, 28(2):300–302.
Ciancarini, P., Omicini, A., and Zambonelli, F. (1999).
Coordination technologies for Internet agents.
Nordic Journal of Computing, 6(3):215–240.
Contucci, P. and Giardin`a, C. (2012).
Perspectives on Spin Glasses.
Cambridge University Press, Cambridge, UK.
Denti, E., Natali, A., and Omicini, A. (1998).
On the expressive power of a language for programming coordination media.
In 1998 ACM Symposium on Applied Computing (SAC’98), pages 169–177,
Atlanta, GA, USA. ACM.
Special Track on Coordination Models, Languages and Applications.

References
References III
FaceBook (2013).
Home page.
http://www.facebook.com.
Gelernter, D. and Carriero, N. (1992).
Coordination languages and their significance.
Communications of the ACM, 35(2):97–107.
Kauffman, S. A. (1993).
The Origins of Order: Self-organization and Selection in Evolution.
Oxford University Press.
Kauffman, S. A. (2003).
Investigations.
Oxford University Press.

References
References IV
LiquidFeedback (2013).
Home page.
http://liquidfeedback.org.
Mantegna, R. N. and Stanley, H. E. (1999).
Introduction to Econophysics: Correlations and Complexity in Finance.
Cambridge University Press, Cambridge, UK.
Mariani, S. and Omicini, A. (2013).
Event-driven programming for situated MAS with ReSpecT tuple centres.
In Klusch, M., Thimm, M., and Paprzycki, M., editors, Multiagent System
Technologies, volume 8076 of LNAI, pages 306–319. Springer.
11th German Conference (MATES 2013), Koblenz, Germany,
16-20 September 2013. Proceedings.
M´ezard, M. and Montanari, A. (2009).
Information, Physics, and Computation.
Oxford University Press, Oxford, UK.

References
References V
Mézard, M., Parisi, G., and Virasoro, M. A. (1986).
Spin Glass Theory and Beyond. An Introduction to the Replica Method and Its
Applications, volume 9 of World Scientific Lecture Notes in Physics.
World Scientific Singapore.
Nishimori, H. (2001).
Statistical Physics of Spin Glasses and Information Processing: An Introduction,
volume 111 of International Series of Monographs on Physics.
Clarendon Press, Oxford, UK.
Omicini, A. (2012).
Agents writing on walls: Cognitive stigmergy and beyond.
In Paglieri, F., Tummolini, L., Falcone, R., and Miceli, M., editors, The Goals of
Cognition. Essays in Honor of Cristiano Castelfranchi, volume 20 of Tributes,
chapter 29, pages 543–556. College Publications, London.

References
References VI
Omicini, A. (2013).
Nature-inspired coordination for complex distributed systems.
In Fortino, G., B˘adic˘a, C., Malgeri, M., and Unland, R., editors, Intelligent
Distributed Computing VI, volume 446 of Studies in Computational Intelligence,
pages 1–6. Springer.
6th International Symposium on Intelligent Distributed Computing (IDC 2012),
Calabria, Italy, 24-26 September 2012. Proceedings. Invited paper.
Omicini, A. and Contucci, P. (2013).
Complexity & interaction: Blurring borders between physical, computational, and
social systems. Preliminary notes.
In B˘adic˘a, C., Nguyen, N. T., and Brezovan, M., editors, Computational Collective
Intelligence. Technologies and Applications, volume 8083 of LNCS, pages 1–10.
Springer Berlin Heidelberg.
5th International Conference (ICCCI 2013). Craiova, Romania,
11–13 September 2013, Proceedings. Invited Paper.

References
References VII
Omicini, A., Ricci, A., and Viroli, M. (2006).
The multidisciplinary patterns of interaction from sciences to Computer Science.
In Goldin, D. Q., Smolka, S. A., and Wegner, P., editors, Interactive Computation:
The New Paradigm, pages 395–414. Springer.
Ricci, A. and Viroli, M. (2005).
Coordination artifacts: A unifying abstraction for engineering
environment-mediated coordination in MAS.
Informatica, 29(4):433–443.
Simon, H. A. (1962).
The architecture of complexity.
Proceedings of the American Philosophical Society, 106(6):467–482.
Stanley, H. E. (2008).
Econophysics and the current economic turmoil.
American Physical Society News, 17(11):8.
The Back Page.

References
References VIII
Verhagen, H., Noriega, P., Balke, T., and de Vos, M., editors (2013).
Social Coordination: Principles, Artefacts and Theories (SOCIAL.PATH), AISB
Convention 2013, University of Exeter, UK. The Society for the Study of Artiﬁcial
Intelligence and the Simulation of Behaviour.
Viroli, M., Holvoet, T., Ricci, A., Schelfthout, K., and Zambonelli, F. (2007).
Infrastructures for the environment of multiagent systems.
Autonomous Agents and Multi-Agent Systems, 14(1):49–60.
Special Issue: Environment for Multi-Agent Systems.
Wegner, P. (1997).
Why interaction is more powerful than algorithms.
Communications of the ACM, 40(5):80–91.
Weyns, D., Omicini, A., and Odell, J. J. (2007).
Environment as a ﬁrst-class abstraction in multi-agent systems.
Autonomous Agents and Multi-Agent Systems, 14(1):5–30.
Special Issue on Environments for Multi-agent Systems.

References
References IX
Zavattaro, G. (1998).
On the incomparability of Gamma and Linda.
Technical Report SEN-R9827, CWI, Amsterdam, The Netherlands.

Complexity & Interaction: Blurring Borders
between Physical, Computational, and Social Systems
Preliminary Notes
Andrea Omicini
andrea.omicini@unibo.it
with Pierluigi Contucci
pierluigi.contucci@unibo.it
DISI / DM
Alma Mater Studiorum—Universit`a di Bologna
ICCCI 2013
Craiova, Romania, 11 September 2013

Complexity & Interaction: Blurring Borders between Physical, Computational, and Social Systems. Preliminary Notes

More Related Content

What's hot (6)

Viewers also liked (7)

Similar to Complexity & Interaction: Blurring Borders between Physical, Computational, and Social Systems. Preliminary Notes (20)

More from Andrea Omicini (16)

Recently uploaded (20)

Complexity & Interaction: Blurring Borders between Physical, Computational, and Social Systems. Preliminary Notes