Posted by
Carlos Gershenson on
May 12, 2006; 2:18pm
URL: http://friam.383.s1.nabble.com/Advice-needed-on-CAS-tp521794p521797.html
Hi Jochen,
Also nice to meet you here, I've also read some of your papers from
the arXiv.
> Mamei et al. say in their article
> about "Case Studies for Self-Organization in Computer Science"
>
http://www.agentgroup.unimo.it/Zambonelli/PDF/JSA.pdf>
actually I am reading that paper now... just went out online in the
Journal of Systems Architecture...
> that "Selforganization appears to be the next 'big concept'
> in science in general". I doubt this is true. Would you agree?
Well, there is a renascent interest in it, e.g. there are workshops
being organized on the subject (e.g.
http://esoa.altarum.net/
esoa06/ ) (renascent because it was very popular in the 40's and 50's
within cybernetics)
Personally, I don't think it will be "a big hit" in science, because
the concept is very loose...
and mainstream science likes hard and crisp concepts.
As Ashby argued, you can call any dynamics system self-organizing. In
other words, dynamical systems have attractors, and if the observer
wants to call that attractor an "organized" state, then you have self-
organization, since the dynamics tend to the attractor. So there's a
bit of subjectivity always tied to self-organization, and many people
don't like that. (Well, I would argue that there's a bit of
subjectivity in all sciences, even mathematics, but that's another
story...)
In this paper
http://uk.arxiv.org/abs/nlin.AO/0303020 , we argue that
the question is not to see wether a system is self-organizing or not,
but when it is useful to speak about it or not...
> According to my experience, dealing with self-organization and
> "emergence" can be quite frustrating, because they are IMHO more
> the exception than the rule. They are interesting concepts, but
> one reason why they are interesting is that they are rare in
> daily life. Nothing organizes itself or appears out of nowhere.
It depends on how you see daily life... As I said before, you can see
any system as self-organizing, since it is just a description of your
observations. So you can describe any animal, any society, any cell,
any computer, any car... any system, as self-organizing. The only
question is wether this is a useful description or not... and for our
daily lives, it is not.
You can see emergence as a change of model, or level of description.
So it is not that things come out of nowhere, but the observer can
describe the observed as emergent or not... So for example anything
you touch has temperature, which its atoms don't have, so you can see
temperature as an emergent property. The same for life, cognition,
etc... Emergence all around us.
What I want to say is that describing something as self-organizing or
emergent does not change the thing itself, only our description of
it. As for now, when is this useful? It seems that only when the
usual ways of describing things run into trouble... The thing is that
nowadays things seem to run into trouble more and more often... So if
we'll change our way of describing daily things or not, that I don't
know...
> Moreover, there is a natural contradiction between self-organization
> and engineering: a self-organizing system with emergent properties
> is not only hard to find, it is obviously hard to design and to
> engineer. What is your own experience from your work towards a
> PhD thesis in this area ?
Well, it is hard because we still don't know how to do it properly...
I don't see a contradiction here, because self-organization becomes a
very useful concept when you have such a huge problem domain that you
cannot exhaust it, or when the problem domain is changing constantly,
so a solution needs to be actively sought for.
If the problem is closed and well defined, self-organization is a
redundant concept. But when the system you are engineering is beyond
your full comprehension, in the sense that you cannot tell the system
precisely what to do, the only choice is to build a system so that it
finds itself what to do... And this works for both cases: when you
don't know the solution, or when this changes constantly. If you use
a traditional approach, you either will get a simplified solution
that will not work, or will get a solution that will be outdated
before you can implement it...
With our current society, there is a demand for these types of
systems, but I don't think that the demand for the "traditional"
systems will decrease, so for a few decades at least, I don't see
self-organization as being "big"...
Hope this satisfied more or less your question...
Best regards,
Carlos Gershenson...
Centrum Leo Apostel, Vrije Universiteit Brussel
Krijgskundestraat 33. B-1160 Brussels, Belgium
http://homepages.vub.ac.be/~cgershen/ ?Tendencies tend to change...?