Slashdot Mirror

Even "genetic algorithms" are not evolution in the strict Darwinian sense, because people intentionally design algorithms and then let them compete.

Not always. For example, see psoup, Tierra, and Avida.

For this to be truly genetic, the algorithms themselves would have to evolve.

For weather simulations to be truly predictive, the algorithms would need to condense out of the atmosphere.

For traffic simulations to really work, they would need to pave little digital roads and build little digital cars.

For nuclear physics simulations to really be nuclear, they would have to require lead shielding around the computer.

There's a squishy robot called Muu, which is actually social with other robots as well. Muu has had promising results with autistic children.

Previous work. But as you said, not for $250. I don't think the Wii is useful for this though, it's just a crude accelerometer. Good enough to distinguish between distinct gestures but not to record trajectories.

The reason viruses don't evolve is that current software systems are too brittle. A single bit error can cause a program to fail. It is possible to implement a virtual machine for a more flexible language. That's essentially what Tom Ray did in Tierra.

"Bizarre things are going on that we are just beginning to get a handle on," says Marcus Pembrey, a clinical geneticist at the Institute of Child Health in London. Consider the pregnant Dutch women who starved during the famine of the Second World War. Not unexpectedly, they had small babies. Far more surprisingly, those babies went on to have small babies, even though the postwar generation was well fed and no genes had been tinkered with.

[interesting article with more examples of non-genetic inheritance and gene imprinting. Google on "heritable methylation" for many more.]

Parasites and evolution:
Evolution, Ecology and Optimization of Digital Organisms by Thomas S. Ray
http://www.his.atr.jp/~ray/pubs/tierra/tierrahtml. html

see also Lynn Margolis' work

non-constancy of evolution - see "punctuated equilibrium" - related to parasite co-evolution in simulations see Ray above

Sexual selection often unrelated to fitness - a commonplace. See peacocks, stags, Pamela Anderson, homosexuals etc. The signs of desirability become divorced from the reality and lead to more waste in display than ultimate reproductive or survival benefit.

Mitochondria as indispensable commensal organisms with an independent genetic lineage, e. coli and other gut organisms as affecting fitness either way (improved digestion/ peritonitis) - (look it up yourself)
Further, the mitochondria are inherited through the ova's cytoplasm.

Immune systems in mammalian infants are initialized from the mothers to a large degree through the colostrum (first milk) and additionally through regular milk. Breastfeeding: Unraveling the Mysteries of Mother's Milk Medscape Women's Health eJournal 1(5), 1996.
by Margit Hamosh, PhD, Georgetown University Medical Center
http://www.asklenore.info/breastfeeding/additional _reading/mysteries.htm

http://www.calfnotes.com/pdffiles/CN050.pdf
the number of leukocytes in colostrum can easily exceed 1,000,000 cells/ml. Colostral leukocytes are primarily composed of lymphocytes (23%), neutrophils (38%) and macrophages (40%). ...Colostral lymphocytes can survive in the intestinal tract due to the lack of proteases found in the intestine during the first 24 hours after birth and the presence of protease inhibitors such as trypsin inhibitor. Further, leukocytes have been shown to be absorbed into
bloodstream of the newborn.

http://www.sscnet.ucla.edu/anthro/faculty/fessler/ pubs/F&A2004InfantMouthing.pdf
IgA from breast milk provides protection against all microbes the mother has or has had in her intestinal tract, as the mother's intestinal Peyer's patches send SIgA against current antigens to the mammary glands, and memory lymphocytes able to target past antigens congregate there as well during lactation [39]. These passively transmitted antib

I first stumbled on this over ten years ago.

http://www.his.atr.jp/~ray/tierra/

The Tierra project has been around for many years, but seems to be pretty slow moving. It works in a somewhat similar fashion, but has its issues, such as only really optimising for reproduction speed (which is correlated with small size), and so you miss some potentially interesting results as the system tends away from complexity.

A friend and I have been talking about writing something that will use some of the ideas from this system, and a bunch of our own, but haven't really gotten very far yet, aside from writing some notes and some prototype code.

The original idea of a evolving computer programs was proposed by Thomas Ray in the early 90's. Check his program http://www.his.atr.jp/~ray/tierra/ Tierra . Many ideas in AVIDA could be traced back to it.

Tom Ray's Tierra

Thanks for telling me about this.

Look up something like Tierra and actually play with it. (I wrote a simpler variant that displays the same behavior). Open-ended evolution that develops new behavior. The 'organisms' I evolved quickly developed ways of using the opcodes I provided that never occurred to me.

How did a simple but robust single-cell organism spontaneously "evolve" into a more complex multi-cell organism?

You should check out Tierra then. Given the simple framework, you can see just what mutation can do to an "organism" over a few generations.
Tierra simulates programs, which just reproduce by coping and "forking". However, the simulated cpu is faulty causing slight mutations. Over time, the programs thus evolve. Some become parasites, others simply more efficient. In some cases, cooperative programs evolve, that rely on each other to "fork". Very cool stuff imho.

It seems like a natural progress of artificial life and as such reminds me more about Tierra than Blade Runner's replicants. If you don't know Tierra, there is an interesting description on Wikipedia:

Tierra is a computer simulation developed by ecologist Thomas S. Ray in the early 1990s in which computer programs compete for central processor unit (CPU) time and access to main memory. The computer programs in Tierra are evolvable and can mutate, self-replicate and recombine. Tierra is a frequently cited example of an artificial life model; in the metaphor of the Tierra, the evolvable computer programs can be considered as digital organisms which compete for energy (CPU time) and resources (main memory).

The basic Tierra model has been used to experimentally explore in silico, the basic processes of evolutionary and ecological dynamics. Processes such as the dynamics of punctuated equilibrium, host-parasite co-evolution and density dependent natural selection are amenable to investigation within the Tierra framework. A notable difference to more conventional models of evolutionary computation, such as genetic algorithms is that there is no explicit, or exogenous fitness function built into the model. Often in such models there is the notion of a function being "optimized"; in the case of Tierra, the fitness function is endogenous: there is simply survival and death. According to Ray and others this may allow for more "open-ended" evolution, in which the dynamics of the feedback between evolutionary and ecological processes can itself change over time (see evolvability).

While the dynamics of Tierra are highly suggestive, the significance of the dynamics for real ecological and evolutionary behavior are still a subject of debate within the scientific community. Tierra is an abstract model, but any quantitative model is still subject to the same validation and verification techniques applied to more traditional mathematical models, and as such, has no special status. More detailed models in which more realistic dynamics of biological systems and organisms are incorporated is now an active research field (see systems biology).

It is very important to remember that given sufficient space and complexity, the difference between carbon-based form of life as we know it and any "artificial" form thereof is only that of a medium. Very interesting read. I hope it will go much further during the next few years and we will see some unimaginable implications of this new idea.