The demarcattion is very easy to find: the demarcation is Protein Science

(see Frauenfelder, Hans, and Peter G. Wolynes. "Biomolecules: where the physics of complexity and simplicity meet." Physics Today 47.2 (1994): 58-66.)

this is why I think studying the relation between sequence, structure and function of proteins is the best arena for studying complexity: you have reliable data instead of hand waving (think of crystal structures of proteins or their thermodynamic properties) and in the same time the complete impossibility to link perfectly known microscopic laws to the entire molecule behaviour.

Basically in my view complexity IS normal science, simply complexity does not allow scientists to be arrogant, given the point is that you cannot perform exact predictions from 'first principles' and you must have the humility to take Nature seriously and derive your regularities from observation knowing your knowledge must take care of the peculiarity of the studied objects...The tools are already there: multidimensional statistics, graph theory..going in this direction you will discover 'top-down' and 'bottom-up' causation naturally merge into 'middle-out' approaches:

http://journal.frontiersin.org/Journal/10.3389/fgene.2014.00083/abstract

and a lot of other interesting things..by teh way this was the work of chemists since many decades but chemistry is not considered an interesting topics for philosophers of science.

I recently found a nice graphical overview about the different research fields including respective well-known reserachers: http://www.art-sciencefactory.com/complexity-map_feb09.html. Maybe it can serve as a starting point.

It is obvious that the mentioned disciplines are not "the same" but they share similar ideas. To my understanding complexity science tries to understand how complex systems work in order explain phenomena where science has not found an explanation yet (descriptive knowledge). On the other hand, cybernetics tries to steer a system directly to influence its behavior (prescriptive knowledge). But I would never use this as a general demarcation criterion.

Here is the abstract of my talk at the 7th Complex Systems French Summer School (2013) :

Not all disciplines that use models and simulations study complex systems. Hence it is necessary to clarify and to classify the specific roles of models and simulations in the study of complex systems. This will be the main objective of this talk. It is not possible to do this without at the same time trying to characterize complex systems themselves or, what will be more modest and operative, our diverse complexity oriented visions on systems. I propose introducing the concept of "character of complexity" for a model. Enlarging the notion of "visions on complexity" (Deffuant et al. to appear), we may attribute four different characters of complexity to a model: 1. its character of notation (the convenience of its formalism), 2. its character of heterogeneity integration (related to its ability to integrate different aspects and/or scales of the target system), 3. its character of condensation (related to the Kolmogorov complexity measure), 4. its character of execution (related to the Bennett complexity measure). Thanks to these conceptual distinctions, some classical questions will be addressed and clarified if not incontestably answered, such as: To what extent can we say that a system under study is complex and not complicated? What is the difference between the model of a complicated system and the model of a complex one? What happens when models themselves become complex? Why the manipulation of complex models of complex systems still can be fruitful? To what extent can we say that a simple model of a complex system can lead to complex simulations? What kind of knowledge can we expect from complex simulations of complex systems? In these last two cases, another question arises: how are we sure that the two characters of complexity (the one of the target systems, the other of the model) are the same? Which means: how can we be sure that the model captured the type of complexity supposed to really occur in the target system?

Location: Conference held at the 7th Complex Systems French Summer School - ISCN (University of Le Havre - France)

Event Date: Jul 9, 2013

Here are my slides : https://www.academia.edu/6214043/_Complex_Systems_and_Models_

Dear Alessandro, thank you very much. I have already read your paper. Networks and biology is an appealing approach, indeed. Currently, I am about to finish a paper on quantum biology and complexity; I found a nice idea from your paper that I shall dare to cite.

Dear Franck, I appreciate your sharing your presentation and paper at the 7th CSFSS. I know the steady work that is going around the French Réseau on complexity: an outstanding (and in my opinion, robust) endeavor. Great!

Dear Carlos,

you are very kind, by the way 'complexity' is not an exoteric field, it is the way we look at nature if we are honest with ourselves....

I appreciate the work being done attempting general conceptual characterisations of complexity, whether in prose, maps or cognate; undoubtedly it will help to orient researchers to the task. But I do not hold out much hope for such endeavours to settle the question or provide deep enough insight. My reason is straightforward: ultimately complexity that is scientifically studiable must be a matter of the dynamical processes involved, but their character can change enormously with very small changes in detail, and these details seem to be endlessly various. I expect, e.g., that any maps drawn to have themselves 'porous', infinitely complex boundaries of similar kinds to those that occur in specific dynamical domains, complexity that is smoothed over/neglected by the mapping concepts. At present we do not seem to have sufficiently high-powered, dynamically based maps that could pick out the relevant domains. Indeed, there are many kinds of complexity that we have only barely discovered as yet.

To make this point clear I have written about the neglect of relevant constraint types in mathematical dynamics that form one crucial dimension for the characterisation of complex processes. Dynamical transformations of constraints in protein interactions would provide a nice study in point. Such dynamics takes us, e.g., well beyond the Santa Fe conception of intricate behaviours from simple many-bodied interactions. In any event, in a recent paper - “On the import of constraints in complex dynamical systems”, Foundations of Science, 18(4), 757-780, 2013. DOI 10.1007/s10699-012-9304-9 - I tried to provide a first reckoning on this topic. Of course, this subject is itself semi-infinite and so, correspondingly is my ignorance; I would be delighted to receive proof that my views were now obsolescent. Meantime, I have a suggested criterion of complex dynamics that will please non-one because of its implicative nature (a nice complex feature): complex dynamics are all those whose constraint structures are of the kinds characterised in the above article and their system behaviours are those of the solutions to dynamical evolution under those conditions.

I appreciate that this does not make life easy for those of us who think about complexity, but it does keep it very interesting.

Cliff Hooker

University of Newcastle, Australia

Dear Carlos, Dear Alessandro,

Kind of you, Carlos. I agree with Alessandro too : the essential is in the vision, the "way we look at" nature. Still, "objective" things can be said that help to conceptually discriminate between the different kinds of the different “ways we look at nature” (hence the diverse “visions on complexity” and probably the absence of a unique demarcation criterium).

Yes, there are many approaches to the idea of complexity, which sometimes seem to refer to completely different things. I believe the main divide is an ontological one, between models and hypotheses, with, on one hand theories referring to ideas, or rather models, and on the other, theories referring to reality. The former (cybernetics, agent models, continuous games, etc.) are mainly aimed at exploring the possible, or, rather, the logically possible. The promoters of the latter theories are more interested in concrete systems and the way they are supposed to work, in links and mechanisms, which they approach via objective indicators. Most anthropologists, many historians and sociologists favor such approach -- although they don't always recognize themselves as systemists --, and I take it you have the same divide in natural sciences.

Yes Pierre, I agree with your divide. But the problem is that it is often overlooked. And sometimes for theoretical reasons : e.g., in most systemist views either in natural or human sciences ("systemist" : i.e. you see parts of reality as true systems in the strong sense just as Bertalanffy defined it), you make no more ontological difference between ideas and realities (see my slides on this undue homogenization).

Dear Cliff, we all sincerity, that's already quite a lot, namely working and sharing for the sake of interest and enthusiasm. After all, the engine behind research and thinking does not know of utilitarian goals beforehand, but the freedom and joy that knowledge produces. Somehow in the tenure of a certain Aristotle. I appreciate your comment.

Bingo Pierre! That's the core of the question raised. I believe there is some need to screen out, so to say, a rather floppy approach form many regarding complexity. Even, as Franck correctly says, a variety or multiplicity of demarcation criteria.

Provided that the critical mass of and on complexity has been steadily growing, it would not be inconvenient to sketch out several criteria that allow us to solidly step beyond the easily confused fields of cybernetics, etc.

Franck et al., don't you think that it could be convenient to recall the fact that, at least from the point of view of quantum physics, we can speak of a system of one particle? In other words, contra Bertallanfy and many others, not only two or more interacting systems are to be called as such. There are, in many levels of reality a system that interacts with itself. Would you agree on this?

Pierre's answer offers a very deep distinction. I'd only add that I hope there is not too deep a divide. Even as someone who generally has little truck with cybernetic, logic-based sorts of "complexity," I have come to value the rhetorical value of this logical systems for articulating specific features of the "real," "in the wild" systems. So long as the relation does not invite equating the logical with the flesh-and-blood, I see there being a nice, creative tension there. The cybernetic models are usually always cartoons, but they are good for articulating hypotheses to support reckless confidence necessary for storming out into the messier empirical data outside. However, I share Cliff's reluctance to envision a divide. We're somewhat knee deep in our own complexity: we study complex systems and project hypotheses that we (ourselves, complex systems) have fabricated. Hard to see how we can navigate and use the different types of complexity all in the same scientific process if they're so well demarcatable.

"The only laws of matter are those which our minds must fabricate, and the only laws of mind are fabricated for it by matter." -James Clerk Maxwell

Hello.

Demarcation criteria has been done for every science ever done since now.

Maybe its early to apply the same procedures of demarcation for sciences of complexity. Instead we could be letting it to a longer development and after that apply some kind of demarcation criteria.

regards.

I agree with Damian : this is a kind of hope, and even – I would say – of gamble. The interesting point in some recent works of the pope of DEVS - Bernard P. Zeigler - on the UAMA “Universal Automated Modeler Agent” (i.e. the project to build a computer framework that – through a generalized agent-based approach - would be able to coin and test all by itself many different models of complex systems on these complex systems themselves) is that it shed a crude light on the implicit internalization of the “reality” it is based on, so that the implemented systematic comparison between (models of) data and models of ideas can work in the computational framework. I tried some comments in a paper entitled : “Framework for M&S with Agents in regard to Agent Simulations in Social Sciences: Emulation and Simulation” available at https://www.academia.edu/2208574/Framework_for_M_and_S_with_Agents_in_regard_to_Agent_Simulations_in_Social_Sciences_Emulation_and_Simulation

Dear Damian and Franck, I find very interesting your take on the problem suggested. And your paper, Franck, a delight. Thank very much to you, both.

As we know, the methods and approaches on complexity at large are widely applied and extended on various fields and phenomena. It is a young field, indeed, and yet a big progress is being made, currently. The pace seems to me to be always rather faster. Great vitality, hence!

I am not sure if this applies: Luhmanns systems theory is all about demarcation. As his social system is a communication system, every single action (communication) defines an "inside" - so, it works as demarcation of this system, while reproducing the system as a whole.

Dear Philipp, à la lettre, it does not correspond. Luhmann is - to say it straightforwardly - a sociology without subjects. He does know about structuralism and system science; however he knows nothing about complexity science. Thank you any way!

 May be insightful to consider Murray Gell-Mann's attempt, focusing more on complex adaptive systems. 

https://www.google.com/url?sa=t&source=web&rct=j&ei=SJpdVZSMDYSMuASB04KABg&url=http://www.santafe.edu/~mgm/Site/Publications_files/MGM%2520113.pdf&ved=0CCIQFjAA&usg=AFQjCNEJ0t4sP76G-3yRVpr0LSyh5NSVAg&sig2=cEDV8NGvd2vmcMKzO-DYvQ

Dear all,

A recent publication on this topic in "Natures Sciences Sociétés" : it is in french but soon will be translated. https://www.researchgate.net/publication/275645416_Visions_de_la_complexit._Le_dmon_de_Laplace_dans_tous_ses_tats

Article Visions de la complexité. Le démon de Laplace dans tous ses états

I think the reality is that these things are all related in some sense.  What would be the reason for demarcation, what insight would be gained from splitting this?  It is a genuine question that one should answer.  I am keen to understand this, so I am hoping that the cited paper (thanks Franck) will soon be available in English.

Dear Franck, I did read the paper you mentioned: Vision de la complexité. I must tell you that I liked it very much. Laplace's demon is an always tempting (and dangerous) peril, out there. Thank you very much, indeed.

Dear J. H., The reason of my question lies in that, as you know, the increasing literature on "complex systems", does include as many fields as cybernetics, system theory, system science, complexity thought (i.e. Morin and others), complexity science - to mention but some f the most important. And yet, it seems to me they all are not the same. We can and should trace distinctions, particularities, differences, among some of these fields. To the best of my knowledge such an endeavor has not yet done, as yet.

I have been trying to work, to some extent, on some demarcation criteria. In any case, it is important to take into account that a demarcation criteria does not entail tracing hierarchies among them; absolutely not.

I think that complexity can exist in every science. In other words, complexity approach just changes the philosophy and methodology but not the object of study.

For example, what is complexity in economics and social sciences? It is a deviation from the standard economic assumptions, like a collection of fully rational identical agents interacting only via markets and living in the world without space.

Complexity in economics emerged in Santa Fe Institute (USA, New Mexico). Here are some features as given by its creators.

Brian Arthur, Steven N. Durlauf, and David A. Lane describe several features of complex systems that deserve greater attention in economics (see https://en.wikipedia.org/wiki/Complexity_economics ):

Dispersed interaction—The economy has interaction between many dispersed, heterogeneous, agents. The action of any given agent depends upon the anticipated actions of other agents and on the aggregate state of the economy.

No global controller—Controls are provided by mechanisms of competition and coordination between agents. Economic actions are mediated by legal institutions, assigned roles, and shifting associations. No global entity controls interactions.

Cross-cutting hierarchical organization—The economy has many levels of organization and interaction. Units at any given level behaviors, actions, strategies, products typically serve as "building blocks" for constructing units at the next higher level. The overall organization is more than hierarchical, with many sorts of tangling interactions (associations, channels of communication) across levels.

Ongoing adaptation—Behaviors, actions, strategies, and products are revised frequently as the individual agents accumulate experience.

Novelty niches—Such niches are associated with new markets, new technologies, new behaviors, and new institutions. The very act of filling a niche may provide new niches. The result is ongoing novelty.

Out-of-equilibrium dynamics—Because new niches, new potentials, new possibilities, are continually created, the economy functions without attaining any optimum or global equilibrium. Improvements occur regularly.

Dear Yuri, I do appreciate your answer. Thank you so much, indeed. In fact, as you mention, one of the ways in which complexity has been conceived is: it answers (or pretend to answer) to the philosophical issues that normal mainstreamed science is not able to solve. This matches, I believe, very well with the beginning of your contribution.

Thanks, Carlos. Apart from philosophy, complexity typically requires different mathematical approach. I would not say that "complexity" and "nonlinear science" are synonyms, but to great extent they are close concepts.

Complexity does not necessarily mean huge volume of equations, or agents and the necessity to use a lot of computation time to solve the problem. Very often one needs to derive core differential equation (or PDE, or system of nonlinear equations) to get non-trivial results. The simplest example of strange attractor is Lorenz system of only 3 nonlinear differential equations (see https://en.wikipedia.org/wiki/Lorenz_system ). It can describe  phenomena in chemistry, optics (laser) and several more applications.