I have to admit that I was thinking about posting something about coffee (it is still early in the morning here) and an article in The Economist mentions how coffee is being used to power cars. Apparently the same energy that can power people can power cars too, no idea whether that would be a sign that cars are becoming more human-like. Still, reassuringly for me, what the cars would use is not directly double espressos but the leftover grounds which would mean I could potentially feed my espresso cravings and fuel my car right with the same effort and cost.
Still, the topic of the day (or week to be a bit more realistic) is, as it was the previous time, something I read in Jerry Coyne’s book Why Evolution is true. The book describes the use of mathematical biology by Dan-Eric Nilsson and Susanne Pelger in a paper entitled A pessimistic estimate of the time required for an eye to evolve. In their model, a patch of cells capable (at least initially) of sensing light is allowed to evolve in a way in which only those mutations that increase the survival advantage were allowed to spread. Their conclusion is that, even in the worst case scenario, vision would nature evolve in only a few hundreds of thousands of years. As most evolutionary processes (unlike cancer) take lengths of time that are difficult for the human mind to fully grasp, a mathematical model can be a very useful tool to explain the evolutionary origin of one of the most sophisticated and, deceptively, engineered-like biological features.
It’s a good question, and useful links – I just read a review of Coyne’s book in PLoS Biology this morning, and I’m kinda tempted to check it out.
The modelling aspect interests me, in particular. Richard Levins wrote an article in 19661 titled “The strategy of model building in population biology”, where he highlights that there are (at least) 3 desirable features that we can aim for in any model:
Slide taken from my course Introduction to Ecological Modelling
but as soon as we move towards any one of these goals, we trade off by moving away from at least one of the others. So, a model can never be maximally real, precise and general. This applies to all models, in ecology, evolutionary biology, Milan catwalks and beyond.
But with a little ingenuity, we can also make a relatively simple model do whatever the hell we want it to! A tweak here, a sneaky implicit assumption there, and hey presto! God can be shown to be an imaginary number…
1 Levins R. (1966) American Scientist 54 421-431
Hi Mike, thanks for your post. I agree that model building, a bit like engineering, requires compromises. I quite didn’t picture it as a triangle though, having in my mind the more conventional line separating generality from realism. I am not sure whether the best route is to start from a general/abstract one and then build realism and parametrise it better to make it more precise or maybe the other way round until we find a happy compromise. That’d leave out the alternative of starting with God as an imaginary number and then trying to explain how that makes him not an irrational one.