Re: Any non-biological complex systems?

Posted by Stephen Guerin-5 on May 28, 2017; 6:11pm
URL: http://friam.383.s1.nabble.com/Any-non-biological-complex-systems-tp7589731p7589799.html

On Thu, May 25, 2017 at 12:00 AM, Russ Abbott <[hidden email]> wrote:

What about my revised question. Can we think of anything that is non-biological, non-human, and not a biological or human artifact that would qualify as an agent based system?

Let me take a stab at what this could mean - but I first have to understand what you mean by agent-based system.

To me, "agent-based system" is an interesting mashup between two research disciplines - "Multi-Agent Systems (MAS)" from the Distirbuted AI community and "Agent-Based Modeling (ABM)" from the Complexity research community.

In MAS, systems and models are one and the same. It is a deployed system comprised of agents. It is not trying to model anything.

In our use, an agent-based model is one possible modeling formalization of a system. A coupled differential equation would be an alternative. Or a discrete-event queuing model a potential third way to model a system.

To make the distinction at the office, when we deploy systems with agents we are not modeling anything. Here, I try to consistently use the term "agent-oriented programming" or "agent-based system" instead of "agent-based model". And in these cases, the agents are semi-autonomous when compared to "object-oriented programming" or "object-oriented systems" where there is a greater reliance on a centralized "main()" thread that is controlling all the objects. I don't think this is what you mean by agent-based system.

Stu Kauffman introduced the term "Autonomous Agent" which I'll always use in capitals, as it really captures something greater than merely "semi-autonomous agents" which I think of as simple interacting finite state machines. Autonomous Agents by contrast, are approaching living systems with the use of energy gradients, sensing of gradients using informational kinematic flows vs mass-based force interactions and the constructions of constraints (equivalent with information btw) to extract work as well as applying work to maintain constraints to realize work-cycles.

So I'm guessing, Russ, that you're asking "can we think of any non-biological examples of Autonomous Agents?" - my answer would initially say "no, I can't".

And if I had beer with you I'd give a subtler murkier answer. I would say there's a more complete definition of living systems than Autonomous Agents that looks at the whole breakdown channels of mass and energy flows as Harold Morowitz and Eric Smith have been describing. Life is a property of the process in the full ecological interactions and not a property of a given entity. It ceases to make sense to ask if a virus or an Autonomous Agent is alive or dead. Life is not a property of an individual. (apologies for butchering this, Eric). I would add to Harold and Eric's description that all Autonomous Agents have a dual in their ecological interactions that is seeking to dissipate their gradients as much as the Autonomous Agent is seeking to extract work from theirs. I'm guessing the interactions of chloroplasts and mitochondria in an autotrophic plant cell are close to this.

As you get systems of coupled complex systems that are all dissipating gradients that are generated by the other and all complex systems are open to matter and energy flows, the boundary of a living system continues to expand out to the point that I start to sound like a pantheist and the world as a whole system is alive. I choose to take it on faith that it is. At that point, what does a non-biological system mean in our world :-)

-S

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove