0
Darwin proven?
10 March 2005
"Scientists at Michigan State prove evolution works", says the cover of the latest (February 05) issue of Discover magazine. As an ardent believer in Darwin, that encouraged me to purchase the magazine as part of my in-flight reading for my trip south this year. Discover is one of my favourite magazines, by the way.
Not too surprisingly, the headline is just a trifle dense with hyperbole. What the scientists at Michigan State have come up with is an artificial life program called “Avida” – an “auto-adaptive genetic system.” Avida can be used to simulate simple organic behaviour - breeding, mutation, competition and eventually whether or not given the conditions set in the virtual environment, if the digital organisms evolve.
Avida is a simulation – more complex and more focused than Zoo Tycoon or SimCity, but nonetheless it is limited in what it can model. In Avida, primitive organisms are modeled as independent, virtual computers which interact with one another and their virtual environment in a complex manner that can be defined and adjusted by the user.
Each digital organism gets a “genome” – a set of instructions that simulates DNA. This instruction set is written to a file and processed in sequence by the “organisms.” If you ever programmed machine language in the heady days of the 6502 or Z80 chips, you will find Avida’s CPU structure with its stacks and buffers familiar.
Cells don’t process DNA in such a simplistic, linear fashion, but rather in a simultaneous-parallel manner. And Avida is limited to a two-dimensional environment, basically a simple grid. But with each organism running on its own, growing, mutating and splitting, you can end up with millions of creatures, making up in quantity what each lacks in complexity.
Simulating life on any scale that resembles even a bacteria’s basic functions would require a lot more processing power than most of us own, so Avida keeps the genetic algorithm simple, with only 26 basic instructions.
It’s not so simple that you can just download the program and start tinkering, mind you. You can get it (for free) but in order to make sense of Avida, you need both some coding experience and some basic understanding of genetics.
You can, of course, simply run the default instruction set that comes with the package, but the fun is in trying to create and modify the genomes and, like a modern Dr. Frankenstein, set your a-life creatures into action.
Like many other computer simulations, Avida collapses the time required by observation of real events into a shorter period in which results can be measured.
Outside viruses, the progress of evolution in the real world is so slow as to be almost impossible to measure. In Avida, assuming the rules are correct and the simulation is reasonably accurate, you can see evolution at work in the space of a single afternoon – or at least virtual evolution.
Avida is only the latest of several competing programs attempting to simulate life and the evolutionary process. From all reports, it is better than the rest, but it’s difficult for a novice to prove that assertion without considerable effort.
If you’ve been around computers long enough, you will likely have come across some of the more primitive a-life programs. John Conway, a British mathematician, began it all with a “cellular automation” program - the Game of Life- which was originally a paper game described in Scientific American in 1970. That was followed by Core War (also known as Darwin), a computer game in which simple programs competed and fought in a virtual arena. Thomas Ray created Tierra in the early 1990s, a simulation in which virtual organisms could replicate, mutate, and recombine. Avida is the latest, but there are others out there, easily found with a little Google searching (try Golem, for example…).
None of these programs are more than elementary efforts. Life is far richer and more complex than any single simulation can manage, but we need to start with baby steps. Unfortunately, these programs are generally too complex for the average user to experiment with and learn from. Avida has a simple front end interface which allows me to set the conditions and genome easily and quickly, but even if you don’t have to sweat over coding, it’s a complicated affair to understand and manipulate.
If you want to learn more, go to Avida home
dllab.caltech.edu/avida/ and get Avida (version 2.2.1 is the latest) to try for yourself. It may not conclusively prove Darwin’s theories for you, but you may be amused by its presumption…
Not too surprisingly, the headline is just a trifle dense with hyperbole. What the scientists at Michigan State have come up with is an artificial life program called “Avida” – an “auto-adaptive genetic system.” Avida can be used to simulate simple organic behaviour - breeding, mutation, competition and eventually whether or not given the conditions set in the virtual environment, if the digital organisms evolve.
Avida is a simulation – more complex and more focused than Zoo Tycoon or SimCity, but nonetheless it is limited in what it can model. In Avida, primitive organisms are modeled as independent, virtual computers which interact with one another and their virtual environment in a complex manner that can be defined and adjusted by the user.
Each digital organism gets a “genome” – a set of instructions that simulates DNA. This instruction set is written to a file and processed in sequence by the “organisms.” If you ever programmed machine language in the heady days of the 6502 or Z80 chips, you will find Avida’s CPU structure with its stacks and buffers familiar.
Cells don’t process DNA in such a simplistic, linear fashion, but rather in a simultaneous-parallel manner. And Avida is limited to a two-dimensional environment, basically a simple grid. But with each organism running on its own, growing, mutating and splitting, you can end up with millions of creatures, making up in quantity what each lacks in complexity.
Simulating life on any scale that resembles even a bacteria’s basic functions would require a lot more processing power than most of us own, so Avida keeps the genetic algorithm simple, with only 26 basic instructions.
It’s not so simple that you can just download the program and start tinkering, mind you. You can get it (for free) but in order to make sense of Avida, you need both some coding experience and some basic understanding of genetics.
You can, of course, simply run the default instruction set that comes with the package, but the fun is in trying to create and modify the genomes and, like a modern Dr. Frankenstein, set your a-life creatures into action.
Like many other computer simulations, Avida collapses the time required by observation of real events into a shorter period in which results can be measured.
Outside viruses, the progress of evolution in the real world is so slow as to be almost impossible to measure. In Avida, assuming the rules are correct and the simulation is reasonably accurate, you can see evolution at work in the space of a single afternoon – or at least virtual evolution.
Avida is only the latest of several competing programs attempting to simulate life and the evolutionary process. From all reports, it is better than the rest, but it’s difficult for a novice to prove that assertion without considerable effort.
If you’ve been around computers long enough, you will likely have come across some of the more primitive a-life programs. John Conway, a British mathematician, began it all with a “cellular automation” program - the Game of Life- which was originally a paper game described in Scientific American in 1970. That was followed by Core War (also known as Darwin), a computer game in which simple programs competed and fought in a virtual arena. Thomas Ray created Tierra in the early 1990s, a simulation in which virtual organisms could replicate, mutate, and recombine. Avida is the latest, but there are others out there, easily found with a little Google searching (try Golem, for example…).
None of these programs are more than elementary efforts. Life is far richer and more complex than any single simulation can manage, but we need to start with baby steps. Unfortunately, these programs are generally too complex for the average user to experiment with and learn from. Avida has a simple front end interface which allows me to set the conditions and genome easily and quickly, but even if you don’t have to sweat over coding, it’s a complicated affair to understand and manipulate.
If you want to learn more, go to Avida home
dllab.caltech.edu/avida/ and get Avida (version 2.2.1 is the latest) to try for yourself. It may not conclusively prove Darwin’s theories for you, but you may be amused by its presumption…
