[FRIAM] low population complexity : unclassified mail

[Stephen Guerin]

Hello John,

Welcome to Friam!

> I'm interested in seeing if there are ways to characterise or 
> classify complex systems - for example...[snip]

I like your list of features for classification.

One shorthand we use to classify CAS that you might add is to ask where the
adaptation/learning takes place in the model. Three generic areas are:
 1) internal to the agents (eg genetic algorithm)
 2) in the agent interactions (eg edge weights in neural networks,
customer/vendor selection in supply networks)
 3) in the environment (eg pheromone fields in ant foraging)

Of course, models can have adaptation happening in multiple locations but
it's a start for classification... 

So, from your brief description of your model, it sounds like most of the
learning is #1 - internal to the agents.

-Steve




> -----Original Message-----
> From: DCCCOEIA1, John Ardis [mailto:DCCCOEIA1 at dpa.mod.uk] 
> Sent: Wednesday, May 17, 2006 8:34 AM
> To: 'Friam at redfish.com'
> Subject: [FRIAM] low population complexity : unclassified mail
> 
> 
> 
> Dear Group,
> This is my first post, so hello all!
> I'm interested in complex systems that have low populations 
> but complicated participants. I understand these would be 
> known as "Fat Agents" in the contemporary complexity 
> vernacular. There will be some adaptation, but probably 
> little I the way of identifiable emergent behaviour. I think 
> two competing agents would themselves - collectively - 
> comprise a sort of CAS, even if the mutual trajectory seemed 
> to be dominated by chaos rather than systemic adaptation.
> The context is information warfare - intelligence, counter 
> intelligence, deception and counter deception (etc. etc.). My 
> problem is that people simplify things by throwing 95% of 
> available information away, then they promptly forget they 
> simplified matters and they go on to treat complex situations 
> as a series of elementary, independent events. I need a 
> language and model to allow people to express and recognise 
> complexity and valuable components.
> You'll see there are two sides to this; one agent striving to 
> recognise, express, understand or predict his own complexity 
> (e.g. strengths, assets, knowledge, liabilities, errors, 
> potential), and striving to compete with another agent, with 
> one or both of them executing information operations upon the 
> other (there's more symmetry, of course, in abundance!).
> 
> I'm interested in seeing if there are ways to characterise or 
> classify complex systems - for example,
> *	Population
> *	Complexity of individual
> *	Is there emergence?
> *	How much adaptation is there?
> *	Is the adaptation stable?
> *	How noisy is the system?
> *	Does the system interact with other complex systems? 
> (If so, how are
> these characterised? Population? Complexity of individual etc. ...) 
> 
> In my case, I have low population, high individual 
> complexity, low behavioural emergence, medium but pretty 
> unstable adaptation, high noise, system/system interaction 
> (with significant similarities between systems).
> 
> I'm new to complexity theory and am probably well behind the 
> curve on this matter, so bear with me :-). I'd appreciate 
> your thoughts.
> 
> Best regards,
> Jas 
> 
> UK MOD
> 
> 
> 
> 
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at cafe at St. John's College 
> lectures, archives, unsubscribe, maps at http://www.friam.org
> 
>