It's finally here! I have corrected what I saw as the major flaw in the zombie-infestation model I described earlier, and proceeded to draw a number of interesting conclusions about the effects of zombies on a human population based on my enhanced model.
I even went to the length of learning LaTeX so as to write it up properly.
Here it is! [pdf] (link updated Nov 2010)
There's actually a lot more I could do, but I don't have enough time at the moment. I have run a lot of simulations, implemented in Ruby, which guided me to the approach I took. It would be interesting to parameterise the difference between my model and the Ottawa one - they had zombies becoming corpses when killed by humans, whereas I have made them destroyed altogether - I could add another parameter γ which is the proportion of killed zombies which are destroyed. It would be zero in the original model and one in mine, and I could calculate how values less than one affect my conclusions.
There is a serious side to this. In accepting the approach taken by Munz, Hudea, Imad and Smith? I constrained myself, while improving the model, to using the same basic technique. If I could include some phenomenon in the model as a rate of change of a population variable, I did. If I couldn't model it in that way, I didn't include it. So including a natural decay rate of zombies would be easy, but introducing the effects of age on humans or zombies would be very difficult.
Most strikingly, I didn't make any correction to an obvious error in the model, that humans and zombies do not achieve better "combat" results by outnumbering the enemy. I didn't do it because the line I did take was more interesting. But Mencius Moldbug's law of sewage applies - if you base a conclusion on N assumptions, and one of them is crap...
Of course, nobody would really rely on such crude mathematical treatments when planning for unlikely events, would they?