Complexity research

A new york times article on the use and misuse of powerpoint has gotten some attention recently. The picture below was presented as a horrifying example of how bad powerpoint graphics could be.




As the NY Times put it: "A PowerPoint diagram meant to portray the complexity of American strategy in Afghanistan certainly succeeded in that aim."

While we can agree that something like this shouldn't be presented on a powerpoint slide, it's not a typical example of bad powerpoint. Nothing like the awful powerpoint that was partially to blame for the space shuttle columbia disaster. There's not even a bullet point here.
But if you click on the image and examine it in detail, notice what the arrows try to convey and especially the green boxes to the right.
You'll notice, this is actually a system dynamics diagram! This gives an interesting insight into the thinking of the US military.

Those of you familar with system dynamics will recognize feedback loops here, some of which seem straightforward. A high insurgent capacity and effectiveness increases their experience, training and leadership which in turn increases their capacity, but with a time delay. A longer loop imply that the more territory not under allied control, the bigger the power of the insurgents.

But as you examine each arrow "A -> B" you notice that every arrow implies "A increases B". So every feedback loop in this diagram is a positive feedback. In system dynamics, we know that most systems tend to have both positive feedbacks and negative, goal-seeking or self-limiting loops. This is strangely absent from this military analysis.

We know that if there are only positive feedback loops in a system, then it will tend to extremes. Either things grow into the sky or they end in catastrophy. Either the allies will score a decisive victory against the insurgents or the insurgents will overrun Afghanistan. Couldn't it be that the war will drag on? That there will be no victory or defeat but that the insurgents will continue to be a force to be reckoned with in Afghanistan for decades? No. In this diagram, a stalemate is impossible. That the war does not end, is literally unthinkable in this model.

So perhaps this diagram, while messy in display and badly lacking in system dynamics principles, still reveals something important about the thinking in the US military.

If you wish to study complexity in an academic fashion or if you'r just a practitioner using ideas from complex systems in one way or another, maybe you find it difficult to orient yourself. To get started, it helps to familiarize yourself with the complex systems community. Or "do networking" if you like. :-)

A particularly good place to start is the recent ASSYST newsletter.
Together with the Complex systems Society this organisation arranges conferences (one paper deadline coming up soon!) maintains a directory of people involved with complex systems and post job offers. And if you'r an academic, it's in places like these that strong grant proposals are chiseled out.

The newsletter also links to videos of sessions from ECCS09. Robert Mackay's talk on emergence looks interesting, Michael battys talk looks interesting too. I love the video-feed they have here, where you can see the presenter and the slides in separate windows and also move back and forth in the slides yourself, without interrupting the presentation.

A tip for newcomers to the conference scene: Talks in the plenary sessions tend to be the most accessible ones and to have the best presenters. 

Since September i've been working on a project. I have neglected posting about it here because i wanted to really dig in and understand it myself before i felt ready to share or debate.

Basically I want to redefine what we consider to be alive as our reasons for excluding what most consider "inanimate objects" just doesn't make sense to me and I would argue don't make sense even according to our current ideas of what is alive. It also solves the question of when did life "arise" on the planet as its always been here, just in a way that is different than what we think of as life. The earth itself is alive, so instead of there being this almost mystical change over from "inanimate matter" into "life" we have simply an evolution from one form of life to another, molecules into single cell organisms. It makes a lot more sense science usually prefers the simpler explanation.

Anway i've done a lot of writing and a lot more thinking on the subject and most of it is posted on my blog which I welcome you to read, comment on and question with me.

XPosted

I came across a paper by David H. Wolpert (a researcher I've mentioned before in relation to the no free lunch theorem) This paper is called An incompleteness theorem for calculating the future (NB: The pdf has the pages in reverse order for some reason) and the abstract reads:

This paper proves that one can not build a computer which can,
for any physical system, take the specification of that system’s state
 as input and then correctly predict its future state before that future state
actually occurs. Loosely speaking, this means that one can not build a physical computer
which can be assured of “processing information faster than the universe”.
This result holds even if one restricts one’s attention to predicting the states
of systems which are finite, purely classical, and obey dynamics which is not chaotic,
 and even if one uses an infinitely fast, infinitely dense computer.

It's from 1996 and has not been widely cited despite the interesting conclusion:  
there are finite, classical systems whose state we humans, even with the aid of electronic computers, can
not compute ahead of time.

There are similar papers elsewhere but it's main contribution seems to be to derive the result without making annoying extra assumptions about the physical system or the computer and that it actually takes "how long does it take to compute?" into account. Sortof a gödel theorem for physics, the author promises us. The author speculates on a physics analogue to the chomsky hierarchy and open problems both computational and philosophical without getting hand-waving and vague. There seems to be some gems in this paper that you can only arrive at through serious theoretical computational science. Take this one for instance:

"“remembering” an event from the past, formally speaking, means predicting that event accurately, using only information from the present"

So how is the claim in the abstract proven? I won't spoil it, but for fans of Douglas Hofstadter and self-referential stuff in general, this should be as obvious as "the butler did it!".  
Hint: The computer is also a physical system

There's a listing of PhD and postdoc positions related to complex systems over here:
http://main.csregistry.org/tiki-in…

"one up-to-3-year position on "quality collectives" and two 1-year positions on blog networks and opinion diffusion."

"Multi-scale and reduction techniques for biofilm modelling"

Judging from the postings, Europe from the UK to Hungary is quite a hotbed of complex systems research

Scholarships for Courses on Complex Systems, June MA, USA

NECSI have funding for a limited number of partial scholarships
for the NECSI Summer School courses on complex systems
concepts and methods to be held June 8-26, 2009 in Cambridge, MA.

Week 1: Complex Physical, Biological & Social Systems
Week 2: Complex Systems Modeling and Networks
Week 3: Methods for the Study of Complex Systems

Members of underrepresented groups are especially encouraged to apply.

Applications should be sent to programs@necsi.edu. NECSI will provide these scholarships on a first come first served and need basis.

For more information and registration:
http://www.necsi.edu/events/summer…

Summer school in France, July and August
Third Annual French Complex Systems Summer School
Lyon and Paris, July 20-August 14, 2009

This new series of international summer schools set in France is co-organized by the two French Complex Systems Institutes, in Lyon and Paris, in coordination with the overarching National Network of Complex Systems (RNSC). It is divided into two periods:

"Fundamentals and Practice of Complex Systems" will take place in Lyon, at the Complex Systems Institute Rhône-Alpes (IXXI), Ecole Normale Supérieure, from July 20 to 31

"Advanced Topics in Complex Systems" will take place in the heart of the Latin Quarter in Paris, at the Complex Systems Institute Paris Ile-de-France (ISC-PIF), from August 3 to 14

Early tuition rate for the french school is 400 Euro, the homepage is: http://www.iscpif.fr/CSSS2009

This article which describe how the herbicide Atrazine and phosphate runoff harm frogs in the US, paints a picture of a bewildering mesh of effects, which required detailed experiments to uncover.


Atrazine is known to kill off pickerel frogs directly, but for the declining northern leopard frog of Minnesota, the story turned out to be more complex. Researchers quantified more than 240 separate environmental factors, finding a strong correlation between atrazine and parasite infections in the frogs. But a correlation isn't the same as cause and effect, so experiments were needed.

What the researches found, were that atrazine weakened the immune system of frogs, as had also been shown earlier. But both atrazine and phosphate runoff caused more algae to grow, which became food for snails, boosting their numbers. The snails are  for a short time  a host for the trematode parasite, which later tries to infect frogs and their tadpoles.  Indirectly, atrazine therefore boosted the number of parasites in the water. So did phosphate, but it was only together with atrazine's immunity-weaking effect that it could deliver a double-whammy to the frog population.  With a weak immune system, the frogs died from infections or had deformed limbs. 

From a complexity science viewpoint, the finding is a reminder that datamining can't reveal everything, the dots were only connected after old-fashioned experiments were performed, with controlled variables. But note that the article describe cause and effect without invoking any of the causal loops familiar to many in complexity science, especially those familiar with system dynamics. Most notably the lifecycle of the trematode parasite. After leaving the snail and having infected the frog, the trematode must wait for a bird or other predator to eat the frog. The predator then excretes the trematode, which is ready to once again infect snails.

The loop is what unlocks the questions needed to be asked in managing this problem. Would reducing atrazine and phosphate every second year be enough to keep the trematode population in check? Is atrazine ok if there are other runoffs that kill snails? Or what if there's some additional runoff that  kill off whatever is competing with the snail for algae? Where are the birds and fish that prey on snails? 

The complex mesh of effects seen in nature means it's harder to spot if and how what we do harm the environment. But when these networks start getting mapped, it's my opinion that we get a much larger toolbox for fixing the problems. When "everything is connected to everything else" there's so many more knobs we could tweak to  improve the odds for these frogs.




Trematode larva being ejected from a snail

A recent article in New Scientist provides a heads up for lovers of complexity science. 

This intelligence report by a NS reporter has it that the next target for Intelligent Design and creationists are neuroscience. Attacking things such as mainstream science's inability to tackle the "hard problem of consciousness" and criticising reductionist-materialistic causation they could well be taken to be proponents of some kind of complexity science, and herein lies our problem.

The workings of the brain is still not fully understood and complexity science and philosophy could well have a role in revealing its workings. But our theories risk being misrepresented and tainted by association with creationism. A theory such as "punctuated equilibrium" was earlier used to argue in favour of ID, but the theory was well enough known not to be tainted by that. It could be different for the many theories that for instance claim to shed light on the hard problem of consciousness. (How something self-aware arises from matter)

It wouldn't be the first time that something from complexity theory has been misused. Fractals and chaos theory inspired everything from new age religion to sloppy analogies in the social sciences, but creationists tend to carry a bit more "punch". Those of us in the US may want to think twice about who we lend our support to. We who live outside the US will recognize creationism as mainly a US movement, but it reminds us to be even a bit more clear and thorough about how we represent our ideas. 

A call for papers is up for the "Approaching Complexity" Workshop as part of the IT revolutions conference in Venice December 17-19, 2008: http://www.iscpif.fr/ITR2008
Submission deadline is 27th of October and here are the flyer.

( Collapse )

A nice experiment involving a tank of glycerin with a convection current that suddenly changes direction, explains what might drive plate tectonics. I think this article and the following video explains it well:



The periodic changes in the direction of the flow in the experiment may be the same mechanism that drove plate tectonics to repeatedly gather and split up supercontinents.

( Collapse )