Complete.Org: Mailing Lists: Archives: freeciv-dev: September 2003:
[Freeciv-Dev] (PR#6268) RFC: Clustering in Civil War
Home

[Freeciv-Dev] (PR#6268) RFC: Clustering in Civil War

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: undisclosed-recipients: ;
Subject: [Freeciv-Dev] (PR#6268) RFC: Clustering in Civil War
From: "Cameron Morland" <cameron@xxxxxxxxxx>
Date: Wed, 24 Sep 2003 08:06:53 -0700
Reply-to: rt@xxxxxxxxxxxxxx

Hi,

This is a request-for-comment on the attached patch.

Currently when a civil war occurs, each city chooses randomly whether or
not it will join the rebels. This causes a highly fractured
civilisation, and usually forces the human to give up. My patch causes
the nation to split into two clusters of cities (using the K-means
algorithm, incidentally.). What I'd like to discuss:

* What features should be used to determine membership in the cluster? I
currently use only Euclidean distance, but other things might be
important: continent number, courthouses, wonders, happiness, etc.

* Should the rebel nation be 1/2 of the whole nation? It seems to me
that cities will usually prefer the status quo, ie to stay loyal, so the
rebels should be smaller. (This is easy.)

* Would we prefer to have, eg, two small rebel nations instead of one
fairly large one? (This is easy.)

One example screenshot of this can be found at
<http://www.cns.bu.edu/~cjmorlan/freeciv/revoltmap1.png>. The Catalan
(red & yellow stripes) nation is my group of rebels, and the Australians
are loyalists. The matching savegame is 
<http://www.cns.bu.edu/~cjmorlan/freeciv/revoltsoon.sav.gz>; connect as
cjmorlan.


Attachment: kmeans.diff
Description: kmeans.diff


[Prev in Thread] Current Thread [Next in Thread]