to represent agents, the nature of agent interactions, theagents’ joint environment, and the specific techniquesadopted to validate the model, that is, to ensure itsreliability as a source of insight about the target.The Iruba (
) project  is followingthis approach to construct and experiment with ageneral model of a guerrilla war sufficiently realistic tooffer new insights into dynamics or to further developexisting insights. In the model agents correspond toguerrilla bands, regime bases or outposts, and toheadquarters on each side. A particular objective is toestablish sets of conditions expressed in terms of themodel’s parameters and structures that guarantee thatan insurgency will succeed or, alternatively, fail.The Iruba model has been made broadly realistichaving regard to the relevant literature. In particular,the model is loosely based on (extensive descriptionsof) guerrilla wars that took place in the last century inIreland (1919-1921) and in Cuba (1956-1959), withsome further features drawn from the Arab Revoltagainst the Turks in the Hejaz (1917-1918) towards theend of the First World War. It is important to recognisethat in many published descriptions of insurgenciesthere is much that is inaccurate and biased to one sideor the other. This is certainly true of aspects of the Irishinsurgency as has been demonstrated by Hart . Nevertheless, reliable sources for the foregoingconflicts are  and .The main structural and behavioural concepts usedin building the Iruba model are drawn from the Irishinsurgency: near autonomous regions with only limitedcentral control; mobility; limited weaponry; theimportance of terrain; and the importance of ideologyand popular support. Correspondingly, the Iruba modelis structured as a network of 32 relatively autonomousregions on an island that vary in terrain and population.The population of a region provides a (finite)recruitment pool for both insurgents and regime forces.Initially the regime forces are relatively numerous,distributed in bases over the island, and relativelystatic, whilst the insurgents are small in number,localised, mobile and hard to find. As indicated,computational agents represent guerrilla cells/bandsand regime bases and insurgent and regimeheadquarters. Attacks take place within regionsfollowing simple rational strategies. For example, aguerrilla band may attack a poorly defended regime base, with the outcome dependent upon terrain, relativenumbers and weaponry, and random factors. Asuccessful attack may well lead to capture of weapons.Movement of insurgent or regime forces betweenneighbouring regions takes place under appropriateconditions. For example, neither the forces that aremoved nor those that remain behind are left atsignificant risk. Recruitment to insurgents and toregime forces (and defection from regime forces)reflects the numbers and attitudes of the so far uncommitted general population of the region inquestion. This population will partially support theinsurgents, and will partially be aware of theinsurgency, depending upon the conflict history in thatregion. These two “population attitude” variables andtheir use are intended to go some way towardscapturing the dynamics of population opinion and itsimpact upon the course of the insurgency.The core cycle of the model may be expressed inoutline pseudo-code as:
Repeat Attacks and their impact HQ decisions Recruitment Force movement Until termination
As indicated above, a degree of central control by“headquarters” agents is possible for both sides. If aninsurgency grows, regime force may be concentratedinto regions where the insurgency is at its strongest.Furthermore, faced with a dangerous insurgency theregime may take “all out” measures (comparable with,for example, the so called “Salvador option”).
On theother side, in appropriate circumstances the insurgentsmay be switched into “hyper-mobile” mode(comparable with the use of “flying columns” by theIRA in Ireland) and/or an “all out” attack across arange of regions or even the entire island, may betriggered (compare the Tet Offensive in the Vietnamwar).Victory in this model is a matter either of insurgent annihilation, or of the insurgents achievingmassive numerical superiority and hence, byassumption, political power. At several points themodel invokes chance factors using a pseudo-randomnumber generator. Consequently the success or failureof an insurgency may vary with the pseudo-randomnumber stream seed even if all other model setting arethe same. The Iruba model has been implemented inthe C programming language. Although some modelvariables (e.g. population support for insurgents) areupdated by simple mathematical relationships, manyaspects of the model structure are much more complex.For example, agents (guerrilla bands, regime bases andHQs) are essentially expressed as sets of conditionalrules. Thus, even if feasible, a more formalmathematical or logical specification of the modelindependent of the code would achieve nothing.
For the “Salvador option” see Michael Hirsh and John Barry, NEWSWEEK, Jan 10
Proceedings of the IEEE Workshop on Distributed Intelligent Systems: Collective Intelligence and Its Applications (DIS’06)0-7695-2589-X/06 $20.00 © 2006