Slashdot Mirror
Digital Life and Evolution
mrivorey writes "Discover Magazine has a story about The Digital Evolution Lab at Michigan State University. Scientists there have created virus-like computer programs that replicate, mutate randomly, and compete with each other... in other words, they evolve. Among such feats as learning to add and compare numbers, these digital life forms also once avoided scientists attempts at "killing" them, by playing dead.
You can download the project yourself from SourceForge." We first mentioned this in early 2003, but it appears to have developed a good deal since then.
The simple reason, from a biological perspective, is for gene mixing.
Nuff said.
The offspring of two sexual creatures is a blend of their genetic material, creating a more diverse species able to endure changing conditions better since there are variations which can adapt. Asexual species exchange genetic material far less and are more similar overall, meaning that come next climate change, they could be screwed, whereas the sexual species might have enough diversity to not only adapt, but thrive under the new conditions.
"If we let things terrify us, life will not be worth living."
- Seneca
Sounds like Tierra from the early 1990s, written by Thomas S. Ray. Artificial life, artificial intelligence, evolution, this is trully fascinating stuff. I hate it when so called "creation scientists" jump into threads like this only to force their superstitious mambo jumbo upon our throats saying that digital life couldn't have possibly evolved, it is complex therefore it must have been designed by an intelligent designer. *cough*ockham's*razor*cough*
Sincerely,
Pan Tarhei Hosé, PhD.
"Homo sum et cogito ergo odi profanum vulgus et libido."
.. one of the best games ever. Digital Life in Creatures. This simulates biochemistry, neural activity, genetics among other this and is great fun.
d ex.php
p hp
http://www.gamewaredevelopment.co.uk/creatures_in
Go get yourself a free copy of Docking Station (the online version of this game) for Linux or Windows:
http://www.gamewaredevelopment.co.uk/ds/ds_index.
Roman Kennke
Both asexual and sexual reproduction offer the benefit of mutation, which is the key to adaptation and evolution. Asexual reproduction offers the addditional benefit of efficiency, but restriccts you to the (benefitial) mutations within your single parent and their ancestors. Sexual reproduction has a penalty for efficiency, but allows your offspring to benefit from the mutations from two separate gene pools. In many cases, with larger life forms, it also offers the additional benefit of more than one parent to care for the offspring and teach them. (the ability to teach is basically a non-genetic form of evolution, and is much more rapid than genetic evolution) The faster you can evolve, the more successful your species is likely to be.
Asexual is "preferred" by microscopic life because even a poorly evolved microbe can still do well if it can reproduce rapidly and efficiently. In the larger kingdoms though, sexual reproduction encourages more rapid evolution, which is key when competing for the more limited resources of the macro world.
I work for the Department of Redundancy Department.
These "digital life forms" only exist within the confines of the host applicastion. That is their "universe", so I don't think we have to worry about Skynet with this particular program. I do worry about viruses using this methodology, but I don't think they could replicate fast enough to evolve before Symantic and McAfee shut 'em down.
I see your point on one level...but on another... AI is so far from anything you talk about happening, it's not even funny. I remember how dissapointed I was when I took my first AI class in undergrad--everything seemed just like hacks to me. don't worry--that world class chess AI is no closer to figuring out, well anything that doesn't involve a chess game than..I can't even come up with an analogy to illustrate my point :-p
Suffice it to say that AI as it stands today is not intelligent. A chess program can play chess, but that's all it can do. A robot designed to get from point a to point b can do that, maybe well, but it can't play chess--it's not like AI has an IQ that can be transferred to having a conversation or thinking about taking over the world.
Likewise, learning systems have a long ways to go too. My Prof. was not a fan of neural networks, so I could be biased, but even HOLLYWOOD neural networks have a rather limited use.
I would worry about any one of about a trillion things before I would worry about AI taking over the world.
Actually I kind of object to the term AI in general, for reasons above..
Viruses replicate by taking over the mechanisms of a host cell. They have no ability to replicate on their own.
What these researches have created are "digital organisms" which are intended to emluate cells. They don't need to invade other systems to replicate, but do it on their own within the runtime enviroment the researches set up.
I seem to be unable to find any source material for this study. I searched for documents coming out of the University of Kalisz from 1997 to date using various keyword approaches and haven't found anything that looks related. Perhaps I'm not choosing my keywords judiciously.
I'm especially interested in tracking down source material on the experiment you describe because of some of the language you're using. In what sense could they "tell" each other information? How did they "try" to figure out the binary format of other processors? And given the results you're describing, why wasn't there any publicity about this event? It seems something likely to make headlines, especially in the kinds of journals I tend to read...
Could you direct me to a link or a reference containing more information about the experiment you are describing, please? It would be greatly appreciated.
What he wants is more important that what I want. What he wants is also more important that what you want.
Well, it is a bit noteworthy that you need an intelligent being to create the program to kick off the evolving software.
Your reality is lies and balderdash and I'm delighted to say that I have no grasp of it whatsoever. - Baron Munchausen
These "digital life forms" only exist within the confines of the host applicastion. That is their "universe", so I don't think we have to worry about Skynet with this particular program. I do worry about viruses using this methodology, but I don't think they could replicate fast enough to evolve before Symantic and McAfee shut 'em down.
sex one of the other major driving influences besides survival, and no doubt will be part of the driving force for AI. i've heard many sexual disorders stem from a desire to have a completely submissive and totally nonjudgemental partner.
much of the utilization of UNintelligent machines has been in undertaking jobs hazardous to humans.. from working around superheated metal to the little police rovers with cameras. there's much to be said for their disposability.
i don't know that replicating the survival instinct would be high on the priority list.
The program is needed to simulate physical laws. In The Real World, no-one would be needed to start things up, assuming the physical laws were already in place.
Now, people can argue from now until the end of time about how the physical laws came to be, but that doesn't have anything to do with evolution.
As an AI researcher, I can tell you that replicating human intelligence (or something better) actually is the ultimate goal of most if not all AI researchers. The point is that nobody expects to achieve anything like that during their own lifetime. But it is all about making small steps.
Maybe it's because millions of CPU-hours is peanuts compared to the Earth over its lifespan? An unparalleled massively parallel biochemical lab would be comparable to untold trillions of CPU hours.
I suffer from attention surplus disorder.
Well their mechanism for transforming genotype to phenotype isn't exactly complex. That results in a limited search space. Their means for simulating competition is pretty weak too (they simply race the organisms, there's no competition for resources).
How we know is more important than what we know.
But it's only a cheat if you disallow the Waterloo & City line because it wasn't always run by LT. Or if you allow Moorgate to Finsbury Park because it used to be.
Unless you stipulated that 1975 or earlier maps should be used.
"She's furniture with a pulse"
There is one big difference between CRobots, Core Wars and the programs mentioned in this article. CRobots code never evolved, it was static code that reacted to certain circumstances, such as the distance to another robot, level of shields left, etc.
The programs mentioned here are evolving "on the fly". This would be equivelent to CRobots code learning from the last battle against an oppenent and applying the knowledge learned against a new opponent, something I personally have never seen.
Something closer to the mark would be Tierra developed in the early 90s.
In those AI distopia movies I always root for the machines. Human emotion is stupid. Love does not conqueror all. Hate just makes people drive airplanes into buildings and build up nuclear stock piles. A rational, higher then human intelligence could actaully save us from ourselves.
------ Take away the right to say fuck and you take away the right to say fuck the government.
More importantly, Sexual reproduction offers something that's fairly lacking in asexual reproduction: Significant genetic exchange
Actually it offers something else: Increased selection speed.
With asexual reproduction, you basically have to wait until nature kills it. A minroly disabiling problem may allow 50 generations of the organism to survive, just barely, before eventually going kaput. Huge waste of resources, no? Sexual reproduction allows the mate to "screen" the organism. With any degree of intelligence at all, the mate can decide that it's not worth mating after all, in advance, because he/she can see the writing on the wall.
This is a form of genetic algorithm (for selection) running on these virtual organisms. It's controlled for a number of reasons:
(1) The only language in which they can operate is what you might call a virtual machine language running on those systems (the Avida system) [it IS a custom language, even if you don't like the VM monikor]
(2) They can evolve, but their success is being tested against a known "fitness equation" (e.g. can it add two numbers). The "playing dead" was a result of him attempting to throw a pre-determined test at the organisms and constraining those that did too well. If they are out-right bad (trying to break out rather than add), they will not survive.
So in order to 'escape', they would have to:
(A) Develop both the known fitness function (addition) as well as the ability to hack out of the Avida (which requires a bug in Avida, something I wouldn't discount outright)
AND (B) hack Avida in such a way that either Avida continues to run in the background, giving the organism CPU cycles or completely evolve all of the code required to copy itself outside of the Avida system (and I would condider the chance of the developed bytes being both legal Avida instructions and legal processor/OS instructions very small)
Given that all we know about evolution shows that complex features evolved from simple features (something the Avida system itself helps to show) the chance of jumping from nothing to a full-fledged stand-alone program is very, very unlikely.
You will notice that there is no place in there for Atheism, since Atheism defaults to selfishness, which in turn implies no troublesome descendents.
Not necessarily. "Selfishness" may lead to altruistic behavior if altruism is rewarding (i.e. activates brain reward systems). Because there are selective benefits to altruism in many circumstances (reciprocal altruism, nepotism) there are likely genes that cause individuals to enjoy being altruistic, quite independently of their religious beliefs.
In those AI distopia movies I always root for the machines :)
When the war between humans and machines begins, you'll be one of the first to go.
Human emotion is stupid
Emotions are nothing more than instructions in the wiring of our brains. They aren't anything magical, a set of inputs gives a set of outputs. We don't even understand how emotions formed, and perhaps their importance to development. Some of what drives us to learn and advance is in part due to emotions. Perhaps robots without emotions never become a threat to humans, because they just don't care about anything.
Why would robots destroy humans? Because they would perceive us as a threat, but perception of threat and reaction is what drives the emotion of fear. Love is driven by our need to reproduce and care for the young. Robots may in fact function as if they had the emotion of love and fear, and a host of other emotions.
Love does not conqueror all
Nope, but it makes life a helluva lot more fun.
Hate just makes people drive airplanes into buildings and build up nuclear stock piles
As opposed to the reasoning which would have the robots wipe out humanity. If robots have a survival instinct, and they perceive that all humans are a threat to their survival they would in fact function as if they hate us.
A rational, higher then human intelligence could actaully save us from ourselves.
Or we can try to take care of that ourselves.
D6 63 0D 70 89 81 BB 8E 7B 7C 5F 5D 54 EA AB 73
I have some time, let me count the dumb things in your comment...
1. Altruism is correlated with reproduction? WTF? By Darwinian standards, reproduction is the ultimate selfish act - one aimed at getting your genes access to more resources. On a social level, you will find the countries with the highest birth rate are the ones where having more children increases your chance for survival and wealth. In countries with a proper retirement system and health care, the selfish reasons for having children are minimized. Guess what: That's why the Europeans and the Japanese are having so few children.
2. Why do you think that people who expect the world to end will "eschew luxury"? Wouldn't they instead be maxing out their credit cards, screwing in bathhouses and living it up? Anyway, why would people who expect the the world to end be having children? Wait, is it because they're altruistic and like to see their children die? I see.
3. ... Oh, forget it, I'm bored with your stupid post. Just one more thing about the atheism comment: I don't think atheists are more selfish than anyone else. They do tend to have fewer children than the average, but not when you adjust for income and education. You see, atheists are on average far more educated and wealthy than others, and all such people, atheists or not, have fewer children. (Again, this is because such people lack the selfish reason to reproduce, since their long-term comfort is assured even without children.)
Perhaps the giardia that used sexual reproduction evolved much more quickly and are now different species altogether?
A slashdotter who didn't build his own computer is like a Jedi who didn't build his own lightsaber.
> Darwinism doesn't explain everything as tidily as some may think.
ID doesn't explain anything at all.
> Behe goes on to say some systems can't be produced by natural selection because "any precursor to an irreducibly complex system that is missing a part is by definition nonfunctional."
His IC argument ignores the possibility of changing the function of a system, which is probably the most common way evolution acts.
> Heavy stuff
I would have said "deep".
ID is nothing but creationist apologetics, bowlderized to try to sneak it past the US court system.
Sheesh, evil *and* a jerk. -- Jade
So basically, according to you, sexual reproduction is preferred in organisms with brains because those brains can select better genes (mates).
It sounds plausible, but how do you explain the vast amount of sexual reproduction in plants? Last time I checked, they don't have brains.
You will notice that there is no place in there for Atheism, since Atheism defaults to selfishness, which in turn implies no troublesome descendents.
Could you please explain how? I really can't understand where you got this from. I would have thought atheism leads to the most selfless acts. See, if you are religious you (in most religions) get rewarded (or not) after you die. Every good thing gets rewarded. You have a limited time on Earth but get rewarded for *eternity* for whatever you do.
An atheist doesn't think there is a god (and therefore an afterlife), no matter what an atheist does they will (in their mind) never get rewarded after death (you just rot in the ground). The *only* time a atheist has is right now during their life. Any help you receive limits the amount of time they have to enjoy their (comparatively) very limited life, while a religious person has eternity to enjoy "heaven" (or wherever they believe they go).
If you want a analogy: If I knew I was going to win a billion (eternal life), giving you $100 is nothing. If I only had $1000 (limited atheist life) giving you $100 is a bloody lot.
And by the way, Darwin himself, at the end of his life, denied evolution as the explanation for how we got here.
No, he didn't.
And even if he had, what difference would it make? Evolution is a fact, not Darwin's opinion.
If Einstein had renounced his theories on his deathbed would relativity be any less true?
Curiosity was framed. Ignorance killed the cat.
Computer simulations are used to enlighten us as to the nature of Reality. The computer you used to create your post, and the computer network that brought that post to my eyes, were all made possible via computer simulation. The idea that "the program gave you exactly what you predetermined in the code" is only true in that our computers are technically deterministic, but your implication that computer simulation can only, at best, "validate what you expect as you design it", is utter nonsense.
Whether you're doing finite element analysis, circuit design, or evolutionary computation, computer simulations can tell you new things, things you didn't expect. Verification is only one of the purposes of computer simulation... discovery is another. And, in the case of evolutionary computation, it can be argued that computer simulations aren't a simulation of evolution... they are in instance of it. The changes that occur in a population of imperfect replicators is evolution, whether those imperfect replicators are made of bits or atoms.
wants to be the first monkey to touch the monolith
Also note that the point of this simulation is not to "prove" evolution, but to try and better understand how it works. We have plenty of examples of evolution in the real world, but since much of the information about previous states has been lost (we only have the sparse fossil record and even sparser antique DNA) it's hard to trace out exactly how the process occurred. With a simulation, you can look at a mutation by mutation record.
Oh, and the theory of evolution isn't trying to explain how we got here - it's trying to explain how evolution works. The fact that organisms do evolve over time is so patently obvious that it's a given. General relativity isn't a proof of gravity, its an attempt to describe gravity. Gravity itself is so patently obvious that it's a given. Think about it. Gravity is a measurable physical phenomenon - so is the change of biological organisms over time.
[javac] 100 errors