Slashdot Mirror
Mathematical Model of Zombie Epidemics Reveals Two Types of Living-Dead Strains
KentuckyFC writes "Epidemiologists have long known how to model the way disease spreads through a population using a computer simulation. This generally involves three populations of individuals: those who are susceptible to disease, those who are infected and those who recover, return to the population and are no longer susceptible. Researchers then feed data about the number of infections and so on into the model which can then work out the disease characteristics such as infection rates. And with this information, they can predict the future evolution of the disease. Now researchers have used a similar model to simulate the spread of infection during a zombie epidemic. They've gathered infection data from real zombie movies, put this into the model and used it to predict the disease characteristics. The results show two clear types of zombie infection which differ in what happens to people after they die. In the first, epitomized by Night of the Living Dead, everybody who dies becomes a zombie. In the second, as in Shaun of the Dead, not everyone who dies becomes a zombie--contact with a zombie beforehand is required. This allows the interesting dynamic of escaping zombification by committing suicide. It also shows how close these zombies have come to winning. The research isn't entirely frivolous. The researchers say exactly the same process of model-building, data gathering and simulation works equally well on real diseases such as influenza. So their approach is a useful teaching tool for budding epidemiologists of the future."
"They've gathered infection data from real zombie movies"
Good! I was afraid they would just make things up.
>They then plug these figures into the model and iterate to find the set of parameters that best fit the data, a process known as Markov Chain Monte Carlo simulation. In total they run the simulations over up to 500,000 iterations.
The author makes Monte Carlo seem like a solver. It's not. You don't use Markov Chain Monte Carlo to model data. You use it to optimize finding solutions by reducing the number of samples required, which allows more complex models with less expensive hardware. You still need the rest of the picture to solve for the data.
That's like saying catalysts cause chemical reactions. No, they don't cause them, they help them go faster.
But this kind of pop media exposure is manna from heaven for researchers. The research itself is fatuous and risible, but the simple fact that a lot of eyes are now focused on these people means that the exposure of their "serious" work has been increased by several orders of magnitude. And often that's what really matters - not the underlying scientific value of your work - but that that work is attuned to tackle problems deemed more fashionable and relevant to society as a whole. Lacking a direct profit motive, fellowship committees have other priorities which are nevertheless rather worldly when determining the allocation of grant money.
If you assume an overnight conversion of 5% of America's population to Zombies (and not a gradual spread of infection) then you have 15.6 million survivors of 320 million people.
If only 1/3 of those people actively went hunting zombies, and managed to kill just one a day, All the zombies are dead in 2 months.
This is why I can't watch shows like The Walking Dead, where zombies are really easy to kill. The level of infection would never reach 95% in a real world scenario where you are required to be infected (by bite or scratch) and killed for it to spread.
According to Wikipedia, the US has an active manpower of almost 1.5 million people. When mobilised, It is safe to assume they with training and equipment they can kill at least 5 a day, meaning the epidemic is over in less than a fortnight.