Re: Modeling the Emergence of Political Parties

[Darren Schreiber]

Actually, the four things at the beginning are what I wanted to
demonstrate, not the predicates of the model.

If a coalition is winning, it won't move of the center of gravity because
it has achieved the objective.  However, there is only one winning
coalition (colored red).  If you look at the first 2d map you'll see that
they all stay at the center of gravity (because they are all purely
ideological).  In the next map,  all of the other coalitions (including the
winning coalition's subordinates) will try to move their issue positions to
gain members.  But in the 3rd 2d map the coalitions implement the very
simple strategy of recentering when there is a defection to or from their
coalition.  (Glen, thanks for the clarifying question, I will update the
site to be clearer on these points).

I'd be glad to answer the implementation questions...

I started with this spatial model since it is the classic one used in
political science.  But, the calculation of distances in this program is
based upon the weighted Euclidean distance.  Each voter and coalition has
its own weight matrix for the issues.  This is how I change the issue
salience in the final run where the parties realign.  The inital salience
of issue 2 is zero ((w2, w2)=0), then I change it to one.  Similarly, if
issue 1 and issue 2 are actually not independent, then we can change the
separability (e.g. (w1, w2)=0.5).

Also, as a side note, the model can be run in a multidimensional issue
space (up to 10 right now).  The display is set up so that you input which
two dimensions you want to visualize during that run in the 2d space.  I
thought this flexibility would be nice down the road.

Some of what you are suggesting Glen is akin to the "directional" model, a
rival to the "proximity" model I am using in this paper.  As I answer the
questions I have right now, I am going to look at a variety of expansions
of the framework.  As you can tell above, I spent a lot of time making the
code flexible so that I can permute it alot.

Thanks for your questions, I appreciate any others that come to mind.

I will likely be publishing the source code in the near future.

        Darren


>Cool model, Darren!
>
>I only have one modeling question:  When you say:
>
>        Then, we let the coalitions move towards the closest agent that
>        is not already a member.  And, we let agents (coalitions and
>        voters) defect to the closest coalition.
>
>Why is this in the model?  The 4 postulates you list at the
>top don't seem to justify the ability to move the platform
>off the "center of gravity" of the coalition.
>
>I have a bunch of implementation questions; but, they're not
>all that important.  The one that might be interesting is:
>Did you give any consideration to issues that might not
>be mappable onto a vector space?  In other words, in the model,
>issue 1 is, to some extent, independent of issue 2.  But,
>in most situations, the space isn't so well-behaved.  For
>example, I can imagine that moderates sit in a kind of
>gravity well near the origin in an "issue-space", if you
>will, whereas those who hold immoderate opinions on one
>or two issues are more sensitive to subtle changes in
>issues on which they do hold moderate opinions.  I.e.
>there might be a kind of annulus about the origin where
>the issues are highly volatile.  In any case, these are the
>thoughts that your model allows me to have. [grin]
>
>Any chance you'll publish the source?
>
>glen
>
>--
>glen e. p. ropella =><= Feeding the hamster wheel.  Hail Eris!
>Home: http://www.trail.com/~gepr                (505) 424-0448
>Work: http://www.swarm.com                      (505) 995-0818
>
>
>                  ==================================
>   Swarm-Modelling is for discussion of Simulation and Modelling techniques
>   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
>   please send a message to <address@hidden> with "help" in the
>   body of the message.
>                  ==================================


_____________________________________________

                 Darren Schreiber
                  Attorney at Law
                 Graduate Student
             Political Science, UCLA
                address@hidden


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================