Sorry your browser is not supported!

This entry is will be quite different from the others and consist mainly of theoretical discussion. Proposing something as ludicrously ambitious as creating an entire tree of life is beyond the scope of a month's coding -- even if I could code for 31 days straight -- but I hope to at least make a start and tackle some of the problems involved in creating such a system. This project requires a lot of research and I'm certain I will learn a lot about evolutionary biology, so it wont be a waste of time regardless of the results I achieve. I have been very vague about the details of this proposed system because I honestly have no idea how detailed and authentic I will be able to make it, which isn't a great starting point for a project, but consider this a research project.


Species Variation

To get a general idea of the task that lies ahead, I have decided to look at one group of organisms in isolation in order to experiment with ways of reproducing the kind of inter-species variations observed in nature. I have chosen mushrooms as my first subject, they are fairly simple organisms but have a great deal of variation, which should serve my experimentation well.

mushroom drawings
I have recently taken up the hobby of mushroom picking, so I know a little about them already, but I couldn't name all of the varieties in this drawing. There is quite an astonishing variety but most of these differences are subtle mutations of size, shape, texture and colour. I could imagine how these variations could be produced by the same algorithm, but the first detail that struck me as an obstacle was on the undersides of the caps. You see, some mushrooms have gills while others have pores. How on Earth am I going to create that kind of a distinction?

This type of "this or that" distinction is of course far more pronounced when looking at a wider range of species, compare a mushroom to a bat and there are a lot of "this or that" distinctions. I cannot specify all of these differences manually, there are far too many, so I must find a way for them to emerge from the same algorithm. Is this possible? How can I produce an algorithm that can switch early on between such distinct paths? Surely it would need to be immensely complicated to the point where I may as well be specifying each difference manually. I think what I require is a set of rules that in turn, when provided with input, produce one of a new set of secondary rulesets that can produce wildly different outcomes from each other. Emergent rulesets? Sounds complicated and messy...

It might be more straightforward to go for another aspect altogether. I can understand trying to adapt the details of the phenotype (shape) is pretty complicated because every change has to have a function within an established set of laws. This is especially difficult when it comes to gills vs pores because that must have something to do with air flow, which is pretty difficult to simulate.

But if you used a physics library you could make actual animals. At first they wouldn't do much but they could adapt to move and compete for food, you could even specialise different limbs, for example to respond to touch. This would be a lot easier than the first option because you have simple mechanics to deal with, not fluid dynamics.

Another approach would be to skip adapting the body altogether. Create a set body (a cube, say) with built in functions, like movement, sight, the ability to change colour, eat, etc, and from there focus on evolving only their behaviour. Watch as different "species" arise and how they react to each other, you might get a predator-vs-prey dynamic emerging.

Good luck though, sounds like a fascinating project!

Libervurto your screenshot doesn't work it gives a 403 error.

If I get this right, your trying to create a procedure for creating the features of a mushroom? I think this is a very interesting idea.

Maybe something to keep in mind is the way mushrooms (or anything else) evolved. You talk about gills and pores but maybe there was something before either gills or pores that was common to all mushrooms. Let's call it a gilre (pronounced "geeler"). Now, at some point in the past all mushrooms had gilres and some mushroom went east and other mushrooms went west. In the east, there was something in the air that made the mushroom's gilres mutate into what we think of as gills and in the west there was something in the soil that mutated the gilres in pores.

Obviously I have no real idea of actual mushroom evolution, or much of an idea of what I'm talking about, but what maybe there is or was or can be something in between a gill and pore, or some connection between the two (gills being elongate pores, say), that makes the alogorithim easier to work out.

Don't forget, what we don't see today, are all the mushroom species that are extinct and all the mutations that didn't work but all of it would still feature in the tree.

I suppose it depends on what you trying to achieve but good luck anyway.

@29 games - thanks, that helps. I see why people don't usually try to do this sort of thing because it is mind-blowing even trying to work out where to start!

I am planning to start experimenting with very basic shapes and create different evolutionary paths from them. I will be looking at simple single- and multi-cell organisms and try to make something that produces similar organisms.