Complex adaptive systems

[James Martin]

Curt, 

I changed subject line because I want to hear more about these ideas regarding complex adaptive systems. 

Tell me more about "John Holland's fundamental elements in a complex adaptive system". You said 'tags' are one of these elements. What are the other elements?

And how can they help us better understand or do better systems science?

James

Sent from my iPad

On Mar 5, 2016, at 4:57 PM, Curt McNamara <cur...@gmail.com> wrote:

A tag is like short hand - a way to represent an idea or set of attributes. They often develop in a particular area of work / research, or in a human activity system. 

It is one of John Holland's fundamental elements in a complex adaptive system.

    Curt

On Sat, Mar 5, 2016, 2:48 PM James Martin <mart...@gmail.com> wrote:

Curt,

What do you mean by "tags"? And how are such tags useful in understanding or dealing with systems?

James

Sent from my iPad

On Mar 5, 2016, at 3:21 PM, Curt McNamara <cur...@gmail.com> wrote:

WRT Jack and Mike:

SoS, socio-technical, and leveraging could be seen as tags from a complex adaptive systems perspective.

    Curt

On Sat, Mar 5, 2016, 1:52 PM Mike Dee <mdee...@gmail.com> wrote:

Thinking in terms of Wymore, SoS is just a system wherein the implementation of certain functionalities (I apologize for using that word) is constrained by use of particular other implementations (Wymore:  Technology Constraints).    Therefore the only thing that differentiates a SoS from any other system is the degree to which the trade space is restricted only to the new elements of the system, and away from existing stuff that is mandated to perform such implementations.   No function design space is available, no implementation design space is available.

 

Most people just call that leveraging.

 

MD

 

From: syss...@googlegroups.com [mailto:syss...@googlegroups.com] On Behalf Of Curt McNamara
Sent: Saturday, March 05, 2016 12:29 PM
To: Sys Sci

Subject: Re: [SysSciWG] Digest for syss...@googlegroups.com - 13 updates in 2 topics

 

Speaking as a systems engineer ...

As I understand it, the term System of Systems is most useful when considering a situation which contains:

-- humans

-- human activity systems (organizations)

-- human designed systems
...

For example, a battle field with pilots, airplanes, air traffic control, airplane control systems, civilians, ...

The intent (as I understand it) is to remind system designers and engineers that their work will exist and be used in a complex environment. This has been a useful construct. 

There are other concepts of great use to a practicing engineer or designer (for example boundary), that have also been challenged by systems scientists. There is a lesson here ... perhaps more than one?

                   Curt

 

On Sat, Mar 5, 2016 at 6:31 AM, Mike Dee <michael....@gmail.com> wrote:

Paula:

I would add this to the (well worn) discussion of SoS...

As commonly used, the term SoS seems to be an engineering term related the CREATION of new functions by adding something new and incorporating functions from existing implementations or nature.   So, the new SoS provides the new functions, but DOES NOT include the other functions of the pre-existing "systems" that comprised the SoS.

So, our definition of a "system" (its boundary) is defined by the I/O that we (as observers) ascribe to it.

A SoS may utilize existing implementations that continue to provide functions unrelated to the desired functions of the "new" SoS.

I understand the desire to use the term SoS, but that term does not seem to add any new distinction.   The only caveat to this is that when we design a SoS, one requirement might be that we have NO deleterious effect on the other systems when our SoS is implemented.   In the SoS world we are making subsystems of things that pre-exist our design.  But how is that diffderent from re-use or leveraging?

Still I see no purpose in the designation SoS.

-----Original Message-----
From: syss...@googlegroups.com [mailto:syss...@googlegroups.com] On Behalf Of Paola Di Maio
Sent: Saturday, March 05, 2016 7:05 AM
To: syss...@googlegroups.com
Subject: Re: [SysSciWG] Digest for syss...@googlegroups.com - 13 updates in 2 topics

Glad to see some topics of interest on the list (an not only rants LOL)

Wanted to add a few comments

Do I read this right that people are still discussing the
boundary/definition for Systems, vs SOS?  Cant believe that people
have been discussing the same thing for decades.

 I have a very clearcut definition, that works (but happy to hear
objections if any)

A system is capable of delivering a system function ALONE (by itself)
that is what makes it a system. if it cannot, it is not a system, but
a subsystem. So, for me a liver is not a system, because it does not
work by itself, if you put a liver on a table, it does not do
anything.

when it comes to a wheel, well, one could argue that the function of
the wheel is to spin, therefore a wheel, properly mounted on a hub, is
a system whose function is to turn.

if we consider a wheel system, then a bicycle could be considered an
SOS, whereby the SOS functionality cannot be deliverd by one or even
two wheels alone, without the other components


RE. WORLDVIEWS, well, essentially a worldview is a statement that only
pertains to human sphere, that is what an individual or collective of
individuals perspective, what they can see. This is pertinent to
engineered systems because it is individuals shape their systems
solely based on their worldview, which is continually shifting

When it comes to natural systems (not engineered), the worldview
matters because it defines what people (general users and/or
engineers) can see. which impacts their ability to
use/manipulate/build upon natural systems.  any engineered system
interfaces and is deployed within a natural system (laws of physics
and all) therefore it is important that we understand , I think the
role of the worldview.  As we continue to learn, the worldview also
changes. if it doesnt. we cannot progress our understanding of the
world.

Just wanted to say these things

Greets to all,

PDM






On 3/5/16, syss...@googlegroups.com <syss...@googlegroups.com> wrote:
> =============================================================================
> Today's topic summary
> 

[Curt McNamara]

Complex adaptive systems are often described as having the following attributes. I am using the context of human activity systems (teaming):

-- aggregates (aka agents). These have the ability to interact with one another

-- tags (aka labels). These are used to pass messages between agents.

-- the environment

-- an ability to change / evolve

This is how I summarized it to my students for this week's assignment:

• Describe a complex adaptive system: chunks/aggregations; interactions; context; and evolution. 
• • Chunks or aggregates in a CAS would be at the level of an organism. This could be a person, a bee, a plant, a group of people in a department, or an organization. This will be your "unit" or "cell". 
  • At the boundary of the chunk there will be interactions. The interactions are how the chunks/aggregates message/respond/exchange/trigger one another. Your diagram should show the aggregates and their potential interactions. 
  • What is the "context" for the CAS and interactions? Is it home, office, traveling? i.e. where will the messages/interactions occur (be sent and received)?
  • Finally, comment on how change/learning/adaptation happens -- evolution is the process by which the overall behavior (made up of the interactions) changes over time. What drives the evolution or change? Could there be a tipping point or phase change into another kind of structure?

        Curt

Here is another view:
http://serendip.brynmawr.edu/complexity/course/emergence06/bookreviews/jferraioli.html

--
The SysSciWG wiki is at https://sites.google.com/site/syssciwg/.
 
Contributions to the discussion are licensed by authors under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.
---
You received this message because you are subscribed to the Google Groups "Sys Sci Discussion List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to syssciwg+u...@googlegroups.com.
Visit this group at https://groups.google.com/group/syssciwg.
For more options, visit https://groups.google.com/d/optout.

[Jack Ring]

Good. 

Is boundary decided by some designer or by the stimuli to which the system actually accommodates?

Does any system ever repeat, i.e., is never changed by its response to a stimulus?

[James Martin]

Curt,

How does your depiction of CASs deal with agents whose behavior is not triggered by information from another agent but by some relationship between agents?

For example, two siblings behave in a way with each other that is different from two strangers who interact with each other, even though the information exchanged is identical in both cases. This difference in behavior is because of their family relationship. 

I don't see anything in your CAS elements that accounts for this difference in observed behavior. In other words, you have interactions noted but relationships between agents. 

And it can be even more complicated. Two people may not know each other but one of them can know that this other person is a member of a gang who had beat up his friend at school. This would obviously affect his behavior towards the gang member even they don't have a direct relationship to each other. 

James

Sent from my iPad

[Lenard Troncale]

The focus should be maximally on HOW they accomplish change and adaptation. The mechanisms and if there are commonalities ACROSS those investigated. These are okay as first draft distinguishing characteristics or CAS but they don't get down to the real mechanisms and the essence. It is the essence that would distinguish them from other systems in general.

Len

[Lenard Troncale]

From the natural systems point of view, there are so many different types of CAS that it frustrates the intent of your good questions.

For example, engineered CAS have boundaries decided by a designer; but in many natural CAS the boundary is decided over long periods of time and selection by the environment or context of the system. Even the stimuli change over time. Any one lineage of CAS may add more and more stimuli they respond to over time to widen their sustainability (the cell certainly has). (and us too)

If a system repeats, which many certainly do, the endanger survivability or sustainability and so evolutionary selection often simply eliminates these non-responders (unless they are human; many of us stick around long after we are useful)

Sorry Curt, these were questions to you but I couldn't help responding. I am CAS.

Len

[Jack Ring]

I may not have asked the right question. 

After a system responds to a stimulus then encounters another instance of that kind of stimulus does it exactly repeat whatever functions and features that were engaged in generating the prior response?

I think not, both for engineered system and for naturally occurring systems.

Jack

[Curt McNamara]

Thanks! Yes, modeling may need to get complicated to capture a situation. Picking the right representation is crucial in system modeling.

For what you described, perhaps use a stock and flow model inside the agents? 

Then the messaging could have a range of effect depending on the amount in that stock.

What I described was very simple, yet has proven to be very useful in the study of emergent behavior.

       Curt

http://ccl.northwestern.edu/netlogo-ccl.shtml

http://www.cas-group.net/

http://www.complexityexplorer.org/

[Curt McNamara]

Len, I agree. However :-) we system folk can make it all seem complicated (which it is).

My goal is to pass on the basics, realizing there are many layers inside and outside those simple models.

    Curt

[James Martin]

Curt,

Stock and flow only addresses that which flows via interactions. But this does nothing to model the relationships that affect agent behaviors: social, familial, managerial, political, geographic, temporal, etc 

James

Sent from my iPad

[Jack Ring]

James,

In your hypothetical case regarding siblings your fallacy is the presumption that the information exchanged is identical when you should be considering whether the cumulative information exchange WAS identical. The ‘family relationship’ has conditioned their ladders of inference such that the next bit of incoming data has far more effect than if the ladder was not pre-conditioned.

In your second hypothesis, one person does, in fact,”know” the other if the person harbors a mental model about the other. 

Jack

[Jack Ring]

Len, 

Please consider that an engineered CAS may exhibit boundaries not intended by the designers due to ‘transcription' errors (coding errors or bugs) inserted by the development and maintenance folks or due to hardware malfunctions that let unwanted stimuli affect the system.  POSIWID.

Jack

On Mar 5, 2016, at 8:02 PM, Lenard Troncale <lrtro...@cpp.edu> wrote:

[Curt McNamara]

To me, stock and flow is a great choice if the model needs to represent behavior and learning/memory. On the other hand, agent based / CAS modeling could be a better choice if the focus is on external interactions.

Your example seems to require both types.

      Curt

[Mary Edson]

Tagging is one of three mechanisms, in addition to four basics, that Holland describes as common to all CAS. Tagging is a mechanism related to the property of aggregation (as in categories) Non-linearity, diversity, and flows are three other examples of properties. Internal models are examples of another mechanism. Holland's book (1995), Hidden Order: How Adaptation Builds Complexity is a distillation of his work and a great resource for learning more about CAS.

[Lenard Troncale]

Seems like "Tagging" to CAS is what "Identifying Features" and "Identifying Functions" is to Systems Processes Theory (SPT). Both describe properties associated with something. They are like the memes we cluster together to distinguish one thing (or concept) from another. Like red, round, tree fruit, taste, makes good juice for "apple." A useful primitive. But SPT has many other similar tags like "Meaurables," "discinyms," "types" to further elaborate its systems processes. Which systems processes of course enable or delivery manifest complex adaptive systems among others. Just sayin'

Len

[joseph simpson]

• TAGGING (Mechanism)
  Tagging is a form of identification and recognition for agents. Holland describes examples of tagging in real life as such things as banners, flags, logos, trademarks, etc. He sees the tagging mechanism as a pervasive feature of CAS as they enable selective interaction between agents and hence act as a mechanism for aggregation and boundary formation.  

Take care and have fun,

Joe

--

Joe Simpson

“Reasonable people adapt themselves to the world. 

Unreasonable people attempt to adapt the world to themselves. 

All progress, therefore, depends on unreasonable people.”

George Bernard Shaw

[James Martin]

Given what Joe provided, these Tags are more than what Len calls "Identifying Features" and "Identifying Functions" in his Systems Processes Theory (SPT). It would also include the "Recognition Features" and "Recognition Functions" perhaps, or something like that. 

I can see how a dog has certain color, shape and size that help me identify it as a certain breed. But then I can see how looking at how the dog reacts to my presence, how it behaves around other dogs, its general demeanor (eg, tail wagging, showing his teeth), etc will help me recognize whether that dog is friendly or not, whether it is hungry or not, whether it has an owner or not, etc.

Len, does your SPT theory account for the Recognition aspects in some way for each of the systems processes?

James

James

[Jack Ring]

Helps clarify classes vs. types.

[Lenard Troncale]

Interesting viewpoint and follow-on questions.

What I always meant for "Identifying" was the same as "recognizing" in your usage IMHO. What I most meant was that a novice would have memes to look for in finding a systems process or isomorphy or pattern in Duane's more favored usage in new or previously unknown material.

It seems you are saying identifying the process is one thing but knowing all the impacts or influences on the process is another. I think that is true and that is where the Linkage Propositions come in (and to some extent a true understanding of ID Functions because they point to specific ways the ISP satisfies the demands of the context and systemness). This would mean SPT fully covers both meanings in its 30 InfoCategories for each SP and in the LPs.

So beyond the dog example, could you take some of the proposed isomrophic systems processes and depict them in the "recognition" way you are pointing to for me to further understand what you mean if I have misunderstood.

Len

[Jack Ring]

You might find it useful to tag Len’s System Processes as simple, compound and complex (similar to what you learned about sentences and algebraic equations (had you been paying attention)). 

In doing so you may question the need for Linkage Propositions.

Or maybe not.

[joseph simpson]

In this case the tags serve a function.

"Tagging is a form of identification and recognition for agents."

"In the nineteenth century, Charles Sanders Peirce defined what he termed "semiotic" (which he sometimes spelled as "semeiotic") as the "quasi-necessary, or formal doctrine of signs", which abstracts "what must be the characters of all signs used by... an intelligence capable of learning by experience", and which is philosophical logic pursued in terms of signs and sign processes..."

The tag is identified by the agent...

The tag is interpreted by the agent.....

Take care, be good to yourself and have fun,

Joe