Saturday, October 18, 2003

I was playing around a bit with epidemic models. Its not just a morbid fascination with zombies (a great word to repeat to yourself and have it lose its meaning) or the spread of disease. Models as to how diseases spread are now being re-purposed to model more abstract infections such as the spread of ideas.

The simplest model is the SEIR model which is a simplistic model of how a flu-like disease can spread among a sea of random interactions. I was playing with the Java model linked at the bottom of that page. You can see for yourself that the behavior never attains any sense of real complexity (scroll down to see the atoms of infection) I did come up with these two parameter sets:

Zombieville: Everybody goes through a phase of infection. The ratios all remain fairly constant as if a truce has been reached with the disease. (10,10,80,200,200,1000000,2,2000,0)

Exploding Zombies: The disease is infectious only at the end of the lifecycle. (20,20,1000,80,1000000,1000000,4,800,100)

More sophisticated models take into account both the complexities of how diseases really spread (types of interaction) and the way in which opportunities for spread are created.

(A fun applet which has made the rounds recently is this Zombie Infection Simulation)

No comments: