[Geowanking] building social modelling tools

[Anselm Hook]

What would it take to build a social software model of our communities
and watersheds?

Recently I've been captivated by Jared Diamond's book 'Collapse'. 
This is a study of several independent communities that each of which
either responded to or failed to respond to their local  geography.
Particularily compelling are the studies that control the number of
factors being examined - such either the kind of population or the
kind of geography.  For example:

	1) According to Jared the Norse died out in Greenland in part because
they applied traditional sheep and cow farming techniques to a land
that was  too fragile to sustain such activity.  At the same time he
notes that the Inuit were more successful and more able to continue
surviving there because they didn't have the same cultural
prohibitions (such as against eating fish) and were in some ways more
technologically advanced.  He also notes that the Norse had a
judeo-christian mythology that may have contributed to their inability
to respect the Inuit people or adapt to the techniques that the Inuit
must have  been demonstrating at the same time.

	2) He points out how the Dominican Republic and the Haiti share the
same island yet that Haiti is denuded of all trees.  And that the only
reason the Dominican Republic is still so well protected was because
of draconian top down land management laws by one of their tin pot
dictators named Joaquin Balageur.  From Jared's book is the comment
"To stop logging, the [Dominican] armed forces initiated a program of
survey flights and military operations, which climaxed in 1967 in one
of the landmark events of Dominican environmental history, a night
raid by the military on a clandestine large logging camp. In the
ensuing gunfight a dozen loggers were killed".  This google maps
example is not perfectly clear but it does show the lack of
tree-cover.  The left side (without trees) is Haiti:

http://maps.google.com/maps?ll=18.050072,71.746902&spn=0.034809,0.038813&t=h&hl=en

	3) He also walks through first world countries - showing that many of
the same trends that destroyed smaller communities are at work in
these societies as well:

	"Australia has an exceptionally fragile environment, damaged in a
multitude of ways incurring enormous economic costs.  Some of these
costs stem from past damage that is now irreversible, such as some
forms of land degradation in the extinctions of native species.  Most
of the type of damage are still ongoing today, or even increasing and
acelerating as in the case of old-growth forest logging in Tasmania. 
To those of us inclined to pessimism or even just to realistic sober
thinking, all of those facts give us reason to wonder whether
Australians are doomed to a declining standard of living in a steadily
deteriorating environment.  This is an entirely realistic scenario for
Australia's future - much more likely than either a plunge into an
Easter Island like population crash and political collapse, or a
continuation of current consumption rates and population growth."

	4) He includes assessments of what is happening here in the west in
small scale as well:  "the sardine fishery of Northern California
collapsed early in the 20th century, the abalone industry of Southern
California collapsed a few decades, and the rockfish fishery of
Southern California is now collapsing and...[so on]"

The one thing that can be said for most human activity is that almost
always the side-effects of any decision always drown out the intended
effects.  We seem to pursue bold agenda's such as 'War on Terror' or
the like only to find that our best intentions are completely overrun
by the things we did not plan for.  Part of the problem is that our
communities exist at a scale that makes it difficult to plan for or
anticipate the effects of our actions.  The time frame and the time
scales are not the same as human consciousness.  (IMO) Forces have
tremendous inertia, moving on time frames below our awareness, and
when they finally start to climb their exponential growth curve then
they approach us at a velocity that is too high for us to think
through fast enough to deal with.

How could we build tools to help us model these kinds of concerns
before they become crisis?

When I was a teen I was influenced by Will Wright's 'Sim City'.  I
remember in fact a friend of mine (S0ren Gr0nbech) had a similar idea
- to simply let people play with a model of their own cities; to lay
down roads and pipe and electricity and homes and to see if they could
build places that functioned.  At the time my watercooler friends and
I thought 'What a crazy idea - who would want to play that"! But
clearly that series of games has a tremendous following.  It provides
a way for people to explore something that is important to them.  Like
many video games (and a few movies such as 'Groundhog Day') it
explores an idea in 'depth' rather than in 'length'.  One gets a kind
of experiential fluidity with all of the variations of a theme rather
than simply having a single linear storyline with very little depth at
each point.  Joseph Campbell has always spoken about how humans are
mythology machines; how we like stories and morality plays... but as
much as we may like that we also I think have a very strong playful
desire to explore complex problem spaces looking for islands of
stability.  We seem to be wired to be playful and exploratory and
perhaps that instinct can be put to service.

One can imagine a social tool that would let us collectively describe
key factors affecting our local communities or watersheds.  We should
be able to describe things such as the water supplies, the arable
land, what kinds of subsidies and funding exist, populations in as
great a detail as possible even down to individual houses, vehicle
usage and street layouts.  Tree cover, wildlife and the relationships
between these and human population.  With sufficient detail such a
tool could be used to do short term predictive modelling of near term
futures.  We could take the rhetoric and posturing of an electoral
candidates position and turn that into a series of relationships that
model their stance to see if the outcomes match their expectations.

Ultimately any kind of social modelling tool is actually a game. 
Except in this case the stakes are higher.  And the players 'win' by
contributing real facts and real solutions into the simulation in
order to create unanticipated outcomes.

I was browing through Katie Salen's and Eric Zimmerman's 'Rules of
Play' that my friend Tom Longson lent to me.  It talks about patterns
for building games.  For example:

	"Conflict over knowledge is a different model for understanding the
way games simulate conflict.  In Trivial Pursuit for example it is
true that the pieces move about on a spatial territory of a board.  It
is also true that the players acquire a set of colored pieces within
an economy of parts in order to win the game.  But these ways of
framing Trivial Pursuit seems to leave out a key component of the game
conflict: the process of asking and answering trivia questions.  In
Trivial pursuit, as with many other games in which information itself
forms the arena of conflict, the contested "terrain" of the game is
knowledge."

And I was struck by the similarity to the book 'A Pattern Language' by
Christopher Alexander; which is simply a laundry list of design
patterns to use for building houses.  One example constraint from his
book (there are hundreds) is:

	1) Intimacy Gradient; homes should be designed with a series of
privacy levels radiating away from public entrances.

Of note is that many of these constraints conform to a fundamental
human irrationality.  We are not rational creatures; we are embodied
in a world that is so complex with so many with many subtle and varied
concerns that they cannot always be reasoned apriori but must simply
be observed.

Effectively what has happened in both the games domain and the home
building domain is that a body of expertise has arisen.  This
expertise has become codified as constraints that help provide a
conceptual scaffolding for limiting the kinds of choices that people
make.  Limiting choices is a way to reach a conclusion quickly. 
Constraints may be good or bad but at least by defining them clearly
and visibly they become explicit factors in design that can be
examined for how they contribute or detract from an overall solution.

Our environment also has similar constraints.  We know that we need
certain kinds of tree cover, or certain kinds of soil.  That sediments
from farming will shallow rivers and make it difficult for boats or
wild fish to navigate therein.  There are centuries of studies and
ground truth awareness about these issues.  Even recently we as a
society have become much more environmentally aware.  Recycling
(whether it does any good or not) is now common for example whereas
we'd just throw empty pop cans over our shoulders with a satisfied
belch when I was a kid.

In building any kind of social model of our community we probably
would be relying on some reasonably rigorous mathematical model.  My
experience at MathEngine and on the http://ode.org project is that
constraints can translate directly to a mathematical model and that a
computer can directly simulate them forward over time to do predictive
analysis.

MathEngine (and more recently CMLabs, Havoc and ODE) are tools for
modelling the physics of say 'people walking' or 'billiard tables' or
'cars driving'.  You could define a set of 3d constraints on 3d
geometrical objects - such as 'has wheels' or 'has a knee joint' and
then the system can do prediction on how that system would behave in
the real world.

If you look at ODE you see a series of well defined kinds of
constraints that you can pour into a simulation such as:

	1) Simple joint based constraints such as:  http://ode.org/pix/joints.jpg
	2) Complex joint constraints such as: http://ode.org/pix/hinge2.jpg
	3) Contact constraints such as: http://ode.org/pix/contact.jpg
	4) Forces such as http://ode.org/pix/amotor.jpg

What ODE does internally is it takes your definition of constraints;
turns each one into a linear equation and then does some simple
calculus to find the intersection of those points in the future. 
Doing this efficiently and believably can be a hassle; techniques such
as Runge-Katta are used.  Doing it quickly and cheaply (although less
accurately) can be possible using say spring based solutions.  We have
to recognize that computer models will not predict the same reality
but will just predict likely realities; they just develop for us a
sense or intuition about how things work.  There are various choices
of algorithms to try and be more accurate but the main value of
prediction at all, even inaccurately, is that it is better than what
we have now which is simply "rhetoric".

Mike Liebhold mentioned to me several modelling projects that various
research and academic institutions have done - such as the "Artificial
Anasazi" project ( http://www.u.arizona.edu/~mlittler/artanasazi.htm )
and David Gelernter (
http://www.cs.yale.edu/people/faculty/gelernter.html ) and somebody
else pointed me to this now disappeared link off the net:
 
http://64.233.179.104/search?q=cache:YrfLyZtOlMQJ:forum.rsuh.ru/archive/o_t__t_990__%D0%B3%D0%BE%D1%80%D0%BE%D0%B4-%D1%81%D0%B0%D0%B4-mvrdv-%D0%B1%D1%83%D0%B4%D0%B5%D1%82-%D0%B7%D0%B4%D0%B5%D1%81%D1%8C-%D0%B8%D0%BB%D0%B8-%D0%BD%D0%B5%D1%82.html+mvrdv+software+planning&hl=en&client=firefox-a

And of course Jim Fournier and Planetwork are pretty centered on these
kinds of ideas.

So there are nascent steps going in the direction of providing tools
to help us model our communities.

But I'm hoping for a broader perspective and to get a broader sense of
any community interest for these kinds of ideas.  In particular how
this could really and effectively be done as a community tool - as a
kind of game; not just as a research project used by a few but a tool
that is used by everybody.  I'd like to see the same stuff that is
happening with social cartography start to happen with social
modelling....  is that possible?

 - a