Wanted - Home Computers to Join in Research on Artificial Life(nytimes.com)
nytimes.com
Wanted - Home Computers to Join in Research on Artificial Life
http://www.nytimes.com/2009/09/29/science/29grid.html?_r=1
2 comments
Modeling a system theoretically capable of evolution is a great idea. However, I think EvoGrid would be better off modeling a system with some simple interactions rather than the real world biochemistry. The real world biochemistry is very redundant and the chances are higher that a simpler system would produce evolving life on the same computer power.
I agree wholeheartedly. However, there's a problem. To date, the only known physical system that has produced life is the one we are made of. To extend on that, we definitely have more knowledge now than before about the basic chemical interactions that make up many of the component molecules we're all built from. Despite this, we still haven't pieced together a bottom to top knowledge base of how you get from amino acids to dna/rna to functional, if simple, self-replicating systems.
In the last couple of years, advancements have been made in getting DNA and RNA to replicate independently in a test tube. Similarly, and more important, there have been at least a couple of experiments where completely synthetic RNA molecules (meaning manufactured and using amino acids not found in living creatures) have shown limited self-replication in test tube environments.
Other research efforts have been studying how certain collections of lipid (fat) molecules can form porous membranes, allowing some molecules to enter while blocking others. This is what our cell membranes do.
From the other direction, projects are in progress that are attempting to knock out all the un-necessary genes in a simple bacteria to create the simplest cell possible. Craig Venter is pretty far ahead on this. http://www.ted.com/talks/craig_venter_is_on_the_verge_of_cre...
As part of that project(I think) they have created a fully synthetic genome. The next part is to put it into an artificial cell and get it replicating.
So we have approaches from both the top and bottom that are seeking to meet in the middle.
I think the EvoGrid, if they are actually trying to simulate real biochemistry, is a very valid approach to help solve this problem. We're probably very close to understanding how to go from basic biochemistry all the way up. An artificial simulation that incorporates known physical properties might offer some additional insights on how this works.
My belief is that once we have a rough draft of a theory on how it all fits together that can be experimentally repeated, it will provide a firm starting point for figuring out the mathematics behind why it works. Think of it as being similar to the Standard Model of physics, but for life. Once we know one set of rules that works, then it opens up the way for experimenters to setup massive simulations with all different kinds of parameters to explore the full range of possibilities for life.
One very interesting extension of this idea is that once we know for sure how carbon-based life arises from the proverbial molecular soup, it lets us start experimenting with other elements. We can examine if its possible for life to arise in previously dismissed areas of the universe. We could determine how likely it is for silicon-based biochemistry to work. It also gives us the opportunity to engineer life using elements that really can never get together in the regular old universe due to natual laws.
In the last couple of years, advancements have been made in getting DNA and RNA to replicate independently in a test tube. Similarly, and more important, there have been at least a couple of experiments where completely synthetic RNA molecules (meaning manufactured and using amino acids not found in living creatures) have shown limited self-replication in test tube environments.
Other research efforts have been studying how certain collections of lipid (fat) molecules can form porous membranes, allowing some molecules to enter while blocking others. This is what our cell membranes do.
From the other direction, projects are in progress that are attempting to knock out all the un-necessary genes in a simple bacteria to create the simplest cell possible. Craig Venter is pretty far ahead on this. http://www.ted.com/talks/craig_venter_is_on_the_verge_of_cre...
As part of that project(I think) they have created a fully synthetic genome. The next part is to put it into an artificial cell and get it replicating.
So we have approaches from both the top and bottom that are seeking to meet in the middle.
I think the EvoGrid, if they are actually trying to simulate real biochemistry, is a very valid approach to help solve this problem. We're probably very close to understanding how to go from basic biochemistry all the way up. An artificial simulation that incorporates known physical properties might offer some additional insights on how this works.
My belief is that once we have a rough draft of a theory on how it all fits together that can be experimentally repeated, it will provide a firm starting point for figuring out the mathematics behind why it works. Think of it as being similar to the Standard Model of physics, but for life. Once we know one set of rules that works, then it opens up the way for experimenters to setup massive simulations with all different kinds of parameters to explore the full range of possibilities for life.
One very interesting extension of this idea is that once we know for sure how carbon-based life arises from the proverbial molecular soup, it lets us start experimenting with other elements. We can examine if its possible for life to arise in previously dismissed areas of the universe. We could determine how likely it is for silicon-based biochemistry to work. It also gives us the opportunity to engineer life using elements that really can never get together in the regular old universe due to natual laws.
I like your point that it's necessary to figure out the math of replication first and then apply the knowledge in design of alternative computer-based replicators.
In addition to the approaches from top and bottom there is one more (bottom-bottom) approach: turn an existing cellular automation like Conway's Game of Life into an evolving computer-based replicator. As far as I know, all existing examples of cellular automation can only produce replicating systems. They are not evolving because they don't have selection. The moment one of these systems is tweaked so that it would also have a possibility to mutate and endure natural (artificial would be a better word here) selection, there will be evolving life in a computer.
In addition to the approaches from top and bottom there is one more (bottom-bottom) approach: turn an existing cellular automation like Conway's Game of Life into an evolving computer-based replicator. As far as I know, all existing examples of cellular automation can only produce replicating systems. They are not evolving because they don't have selection. The moment one of these systems is tweaked so that it would also have a possibility to mutate and endure natural (artificial would be a better word here) selection, there will be evolving life in a computer.
Project site: http://www.evogrid.org/index.php/Main_Page They're looking for volunteer developers, don't seem to have anything running as yet.