Welcome Guest [Log In] [Register]
Speculative biology is simultaneously a science and form of art in which one speculates on the possibilities of life and evolution. What could the world look like if dinosaurs had never gone extinct? What could alien lifeforms look like? What kinds of plants and animals might exist in the far future? These questions and more are tackled by speculative biologists, and the Speculative Evolution welcomes all relevant ideas, inquiries, and world-building projects alike. With a member base comprising users from across the world, our community is the largest and longest-running place of gathering for speculative biologists on the web.

While unregistered users are able to browse the forum on a basic level, registering an account provides additional forum access not visible to guests as well as the ability to join in discussions and contribute yourself! Registration is free and instantaneous.

Join our community today!

Username:   Password:
Add Reply
Evolution simulator Gedankenexperiment
Topic Started: Aug 23 2016, 09:30 AM (322 Views)
Sceynyos-yos
Member Avatar
dheubewes wedor
 *  *  *  *  *  *
A recent discussion made me embark on this train of thought: what would it take to make a fully-fledged, realistic, or at least somewhat so, evolution simulator? The very thought of it, the very concept is overwhelming and if one doesn't know how to go about it, he isn't likely to do any better than a blindman in a maze, before giving up shortly afterwards. There's need for structured approach.

The scope of the goal

'Fully-fledged, realistic evolution simulator' sounds great and all, but it's not well defined. What functionality exactly should it have? Simulating the whole biosphere is obviously out of the question. There are billions upon billions of species, most of them with millions or billions of living specimen. In order for the simulator to run in reasonable and realistic times, we'll have to make a number of assumptions and simplifications. If we want the simulation to run for anything less than a year (or rather anything less than great many years), our first two simplifications are quick to reach:

Simplification 1: The ES (evolution simulator) will simulate only a single species.

Simplification 2: Only a small subset of the simulated species' population will be simulated, which will be assumed to be representative and yet large enough to be significant.

And just to make our life easier some more we can have

Simplification 3:
Only a fixed area of terrain will be simulated, which will be static over time.

Now that we have these three simplifications in place, we can actually do something. (I know many of you will already have an objection - about S1; we'll get to that later.) But even so the scope is still as broad as 'build me a city'. What should an ES actually do? One would say: have it receive as input species X plus simulation variables and output (evolved) species Y. Or rather, most people would start long convoluted descriptions intending to convey that meaning, not necessarily managing to; you get my point.
Or would they? Some would expect it to show actual evolution in action, as this is what the name implies. And they would be right.
So which of the two approaches are we going for? Neither one seems more correct, so while we're at it, we may as well go for both. Finally, we can more or less define the functionality of the simulator:

Functionality 1: The ES would have two modes of operation - fast-forward mode, simulating many generations over at once, and realtime mode, showing the intermediary or final organisms in action. Those two modes would be switchable between at any time.

Functionality 2: At the start of the simulation the ES would accept a starting species and simulation variables such as terrain shape, environmental factors and evolution time.

Functionality 3: The fast-forward mode would just be a genetic algorithm, as these algorithms resemble evolution quite closely.

Functionality 4: The realtime mode would be some physics simulation with species and environment interacting. Due to its very nature, no actual evolution would occur in it.

So this is it. Now that we have the basic functionality and premises outlined, we can built upon them until the whole Gedankenexperiment is over. Let me put a spoiler and say I've mostly carried it out and there are only a few major technical hurdles to get the SE done. Won't post everything at once though.

'Hold on! What about S1? No actual evolution would occur with only a single species being simulated.'
That's right, the single species thing was only for the point. If there is only a single species present, it'd necessarily be an autotroph and would quickly degenerate to some algea-like form if it isn't already. Fortunately we still don't have to simulate billions a species to have an actual evolution. Only a few would do for a reasonable amount of adaptaions and evolution. So let's restate our first simplification to

Simplification 1: The ES (evolution simulator) will simulate only an autotroph, a primary consumer and a secondary consumer.

This configuration of just three species is still enough to have a basic arms race that would fuel the evolution.

'But just a single autotroph won't degenerate to algea-like forms - interspecies competition would lead to specialisation and-'
Um, no. This is for a much later part of the Gedankenexperiment, but let me state it from now: speciation is much trickier than it looks. Let's make our lives even easier (at least for now) and have our first assumption:

Assumption 1: All populations are always connected and no speciation event ever occurs.

This concludes my first post out of several. The next one may take quite some time.
Offline Profile Quote Post Goto Top
 
1 user reading this topic (1 Guest and 0 Anonymous)
« Previous Topic · General Discussion · Next Topic »
Add Reply